PostHog vs Mixpanel

Agent access

1. ai-native userPoint an agent at llms.txt or agent-oriented docs

   weight 2 · round drawn
   PostHog✓fullprobed9/10

   PostHog serves a working llms.txt (probe confirms HTTP 200 with structured content describing the platform) and extensive agent-oriented docs like the MCP integration page (posthog-docs-12, 19, 24, 31) explicitly designed for AI agents/editors to consume. Individual docs pages also expose `.md` variants as seen in citations (e.g., product-analytics.md, session-replay.md), showing a broader agent-readable docs pattern. Missing for 10: a root docs.md aggregate returned 404, so machine-readable coverage isn't perfectly uniform across all docs.

   • [probe] “PROBE llms.txt: HTTP 200 at https://posthog.com/llms.txt # PostHog > PostHog is the platform for self-driving products. It pairs the full c…”
   • [claimed-docs] “"Show me the full stack trace for the most recent crash, then propose a fix." – Pulls the stack trace, error message, and metadata so the ag…”
   • [claimed-docs] “with just plain text questions your agents can ship a feature flag from a prompt, dig into a stack trace without leaving your editor, run a …”
   • [claimed-docs] “your agents can ship a feature flag from a prompt, dig into a stack trace without leaving your editor, run a HogQL query through Claude”
   • [probe] “official MCP server documented at https://posthog.com/docs/model-context-protocol”
   • [probe] “PROBE docs-md: HTTP 404 at https://posthog.com/docs.md”

   proof ↗⚑ flag

   Mixpanel✓fullprobed9/10

   A direct probe confirms Mixpanel serves a working llms.txt at docs.mixpanel.com/llms.txt (HTTP 200) with a curated index of docs, and Mixpanel also documents an MCP server for agent-native data access. Missing for 10: no independent/community corroboration of an agent actually consuming llms.txt successfully.

   • [probe] “PROBE llms.txt: HTTP 200 at https://docs.mixpanel.com/llms.txt # Mixpanel - [What is Mixpanel?](https://docs.mixpanel.com/docs/what-is-mixp…”
   • [claimed-docs] “Mixpanel provides a hosted Model Context Protocol (MCP) server that gives AI assistants direct access to your Mixpanel data.”
   • [probe] “official MCP server documented at https://docs.mixpanel.com/docs/mcp”

   proof ↗⚑ flag

2. ai-native userRun the product headlessly / in CI for automation

   weight 2 · round to PostHog
   PostHog✓fullprobed7/10

   PostHog's CLI is explicitly documented for use in 'local scripts, and CI/CD pipelines' alongside terminal and coding agent use, plus REST/HTTP APIs and OpenAPI spec enable headless automation of data capture, queries, and flag management. missing for 10: independent/hands-on confirmation of CI/CD usage, and no explicit CI example/config shown (e.g., GitHub Actions snippet).

   • [claimed-docs] “The PostHog CLI lets you use PostHog from your terminal, your coding agents, local scripts, and CI/CD pipelines.”
   • [probe] “official CLI documented at https://posthog.com/docs/cli”
   • [claimed-docs] “capture events from any language that can send HTTP requests”
   • [probe] “PROBE openapi: HTTP 200 at https://posthog.com/openapi.json — contains "openapi" key”
   • [claimed-docs] “Inject and upload source maps for PostHog error tracking”

   proof ↗⚑ flag

   Mixpanel~partialclaimed5/10

   Mixpanel exposes REST-based APIs (Raw Event Export, Query API, Engage API, JQL) that can be scripted and run headlessly/in CI for data export and querying, and warehouse connectors support automated syncs. However, there's no evidence of an official CLI, SDK for automation pipelines, or documented CI/CD integration pattern — the product is primarily a UI-driven analytics platform with API access as a secondary capability. missing for 10: dedicated CLI tool, CI/CD pipeline examples or GitHub Actions integration, headless automation SDK, and any hands-on/community evidence of running Mixpanel in automated pipelines.

   • [claimed-docs] “To export raw event data, use the Raw Event Export API. The events payload is returned as JSONL”
   • [claimed-docs] “The raw export API allows you to download your event data as it is received and stored within Mixpanel, complete with all event properties”
   • [claimed-docs] “The Mixpanel Query API endpoints can: Filter or break down Events * Return user profiles. * Return an activity/event stream for a user. * Re…”
   • [claimed-docs] “The JQL JavaScript code must define a main() function. This JavaScript will be compiled, and the main() function will be executed by the que…”
   • [claimed-docs] “The Query API allows you to obtain the same calculated and formatted results that you see from a report in our web app.”
   • [claimed-docs] “you can sync data from data warehouses like Snowflake, BigQuery, Databricks, Redshift, and Postgres to Mixpanel”
   • [claimed-docs] “Use the Feature Flags API to assign your users to variants for experiments, rollout, and releases.”

   proof ↗⚑ flag

3. ai-native userPlug MCP servers into this product so it can use their tools

   weight 3 · round drawn
   PostHog—none0/10

   All MCP-related evidence describes PostHog exposing its own MCP server so external AI clients/editors (Claude, Cursor, etc.) can query PostHog's data and tools — the reverse direction of this story. There is no evidence that PostHog itself acts as an MCP client that can plug in external MCP servers to use their tools.

   • [claimed-docs] “Run trends, funnels, retention, and SQL queries from any MCP client or AI editor.”
   • [claimed-docs] “"Show me the full stack trace for the most recent crash, then propose a fix." – Pulls the stack trace, error message, and metadata so the ag…”
   • [claimed-docs] “with just plain text questions your agents can ship a feature flag from a prompt, dig into a stack trace without leaving your editor, run a …”
   • [claimed-docs] “your agents can ship a feature flag from a prompt, dig into a stack trace without leaving your editor, run a HogQL query through Claude”
   • [claimed-docs] “with just plain text questions your agents can ship a feature flag from a prompt, dig into a stack trace without leaving your editor”
   • [probe] “official MCP server documented at https://posthog.com/docs/model-context-protocol”
   • [github] “You can steer it all from Slack, web, desktop (PostHog Desktop), or your own editor via the MCP.”

   proof ↗⚑ flag

   Mixpanel—none0/10

   Evidence only shows Mixpanel exposing its OWN hosted MCP server so external AI assistants can query Mixpanel data (mixpanel-docs-10, mixpanel-docs-19, mixpanel-probe-3) — this is the reverse direction of the story, which asks whether Mixpanel can consume/plug in external MCP servers to use their tools. No evidence exists of Mixpanel acting as an MCP client or supporting third-party MCP tool integration.

   • [claimed-docs] “Mixpanel provides a hosted Model Context Protocol (MCP) server that gives AI assistants direct access to your Mixpanel data.”
   • [claimed-docs] “you can query events, funnels, flows, retention, session replays, and more using natural language — no dashboard navigation required.”
   • [probe] “official MCP server documented at https://docs.mixpanel.com/docs/mcp”

   proof ↗⚑ flag

4. ai-native userConnect an agent via an official MCP server

   weight 3 · round drawn
   PostHog✓fullprobed8/10

   PostHog documents an official MCP server (docs/model-context-protocol) enabling agents to run trends, funnels, HogQL queries, ship feature flags, and pull stack traces directly from MCP clients like Claude, corroborated by the GitHub README describing steering via MCP and a dedicated probe confirming the docs page exists. Missing for 10: independent third-party hands-on validation of the MCP server (only vendor docs/GitHub and one unrelated community complaint about a different AI assistant feature).

   • [claimed-docs] “"Show me the full stack trace for the most recent crash, then propose a fix." – Pulls the stack trace, error message, and metadata so the ag…”
   • [claimed-docs] “with just plain text questions your agents can ship a feature flag from a prompt, dig into a stack trace without leaving your editor, run a …”
   • [claimed-docs] “your agents can ship a feature flag from a prompt, dig into a stack trace without leaving your editor, run a HogQL query through Claude”
   • [claimed-docs] “with just plain text questions your agents can ship a feature flag from a prompt, dig into a stack trace without leaving your editor”
   • [github] “You can steer it all from Slack, web, desktop (PostHog Desktop), or your own editor via the MCP.”
   • [probe] “official MCP server documented at https://posthog.com/docs/model-context-protocol”

   proof ↗⚑ flag

   Mixpanel✓fullprobed8/10

   Mixpanel documents a hosted, official MCP server that gives AI assistants direct access to Mixpanel data, exposing tools like natural-language querying of events/funnels/retention and dashboard creation, confirmed by both docs and a live probe of the docs page. missing for 10: independent/hands-on verification of the MCP server working in practice, and details on authentication/scopes beyond the docs.

   • [claimed-docs] “Mixpanel provides a hosted Model Context Protocol (MCP) server that gives AI assistants direct access to your Mixpanel data.”
   • [claimed-docs] “Create-Dashboard | Create a new dashboard with text cards and reports”
   • [claimed-docs] “you can query events, funnels, flows, retention, session replays, and more using natural language — no dashboard navigation required.”
   • [claimed-docs] “Once connected, you can query events, funnels, flows, retention, session replays, and more using natural language — no dashboard navigation …”
   • [claimed-docs] “you can query events, funnels, flows, retention, session replays, and more using natural language — no dashboard navigation required”
   • [probe] “official MCP server documented at https://docs.mixpanel.com/docs/mcp”

   proof ↗⚑ flag

5. ai-native userUse an official CLI

   weight 2 · round to PostHog
   PostHog✓fullprobed8/10

   PostHog documents an official CLI usable from terminal, coding agents, local scripts, and CI/CD pipelines, with concrete features like source-map injection/upload for error tracking, and this is corroborated by a dedicated docs page and a probe confirming the CLI documentation is live. missing for 10: independent hands-on community review of the CLI itself (community evidence only covers MCP/LLM features, not the CLI specifically).

   • [claimed-docs] “The PostHog CLI lets you use PostHog from your terminal, your coding agents, local scripts, and CI/CD pipelines.”
   • [claimed-docs] “Inject and upload source maps for PostHog error tracking”
   • [probe] “official CLI documented at https://posthog.com/docs/cli”

   proof ↗⚑ flag

   Mixpanel—none0/10

   The evidence pack documents Mixpanel's APIs, MCP server, JQL query language, and export methods, but contains no mention of an official command-line interface (CLI) tool for Mixpanel. A SaaS analytics platform could plausibly ship a CLI for automation/scripting, so the axis applies, but no evidence supports it.

   ⚑ flag

6. ai-native userDrive the product through a documented public API

   weight 3 · round to PostHog
   PostHog✓fullprobed9/10

   PostHog has a documented public API (posthog.com/docs/api.md, HTTP capture from any language), a live OpenAPI spec confirmed by probe (openapi.json), an official CLI for terminal/CI/CD use, and a documented MCP server enabling AI agents to run queries, ship flags, and inspect stack traces via plain text — all first-party documented and probe-verified. Missing for 10: independent hands-on corroboration of the API/OpenAPI usage beyond vendor docs and probes.

   • [claimed-docs] “capture events from any language that can send HTTP requests”
   • [probe] “PROBE openapi: HTTP 200 at https://posthog.com/openapi.json — contains "openapi" key”
   • [probe] “official MCP server documented at https://posthog.com/docs/model-context-protocol”
   • [probe] “official CLI documented at https://posthog.com/docs/cli”
   • [claimed-docs] “with just plain text questions your agents can ship a feature flag from a prompt, dig into a stack trace without leaving your editor, run a …”
   • [claimed-docs] “The PostHog CLI lets you use PostHog from your terminal, your coding agents, local scripts, and CI/CD pipelines.”
   • [claimed-docs] “"Show me the full stack trace for the most recent crash, then propose a fix." – Pulls the stack trace, error message, and metadata so the ag…”

   proof ↗⚑ flag

   Mixpanel✓fullprobed8/10

   Mixpanel documents a comprehensive public REST-style API surface (Query API, Raw Event Export API, Engage API, Feature Flags API, JQL) plus a hosted MCP server that lets AI assistants query events, funnels, retention, and replays via natural language, giving AI-native users both programmatic and agentic access. missing for 10: no discoverable OpenAPI/Swagger spec (probe found only 404s) and no independent/hands-on corroboration of the MCP server or APIs actually working reliably.

   • [claimed-docs] “To export raw event data, use the Raw Event Export API. The events payload is returned as JSONL”
   • [claimed-docs] “The raw export API allows you to download your event data as it is received and stored within Mixpanel, complete with all event properties”
   • [claimed-docs] “The Mixpanel Query API endpoints can: Filter or break down Events * Return user profiles. * Return an activity/event stream for a user. * Re…”
   • [claimed-docs] “The Query API allows you to obtain the same calculated and formatted results that you see from a report in our web app.”
   • [claimed-docs] “Use the Feature Flags API to assign your users to variants for experiments, rollout, and releases.”
   • [claimed-docs] “Create custom queries using Javascript code”
   • [claimed-docs] “Mixpanel provides a hosted Model Context Protocol (MCP) server that gives AI assistants direct access to your Mixpanel data.”
   • [claimed-docs] “you can query events, funnels, flows, retention, session replays, and more using natural language — no dashboard navigation required.”
   • [probe] “PROBE openapi: all candidate paths 404 (https://docs.mixpanel.com/openapi.json, https://docs.mixpanel.com/swagger.json, https://docs.mixpane…”
   • [probe] “official MCP server documented at https://docs.mixpanel.com/docs/mcp”

   proof ↗⚑ flag

7. ai-native userIssue scoped/least-privilege API credentials for an agent

   weight 2 · round drawn
   PostHog—none0/10

   No evidence describes scoped or least-privilege API key/credential issuance for agents — the API docs mention basic HTTP event capture and an OpenAPI spec exists, but nothing about permission scoping, token minting with restricted access, or credential management for AI agents specifically.

   ⚑ flag

   Mixpanel—none0/10

   Mixpanel offers an MCP server and various APIs, but the evidence pack contains no mention of scoped/least-privilege API keys, service accounts, or permission-limited tokens for agent use — only generic API/export/query docs and generic security incident commentary unrelated to credential scoping.

   ⚑ flag

8. ai-native userBuild against official SDKs

   weight 2 · round to PostHog
   PostHog✓fullprobed7/10

   PostHog documents official SDKs/APIs (HTTP capture from any language, OpenAPI spec, CLI, MCP server) that let AI-native builders integrate programmatically, including AI-specific tooling like LLM observability SDKs and MCP-based agent workflows (posthog-docs-11, posthog-probe-3, posthog-docs-20/24, posthog-gh-6/9). A hands-on community comment corroborates smooth SDK integration for a Swift iOS app (posthog-comm-3), though another notes friction with LLM/AI query features (posthog-comm-8), suggesting some rough edges. Missing for 10: an explicit enumerated list of officially maintained language SDKs and stronger independent third-party validation of SDK completeness/versioning.

   • [claimed-docs] “capture events from any language that can send HTTP requests”
   • [probe] “PROBE openapi: HTTP 200 at https://posthog.com/openapi.json — contains "openapi" key”
   • [claimed-docs] “The PostHog CLI lets you use PostHog from your terminal, your coding agents, local scripts, and CI/CD pipelines.”
   • [claimed-docs] “your agents can ship a feature flag from a prompt, dig into a stack trace without leaving your editor, run a HogQL query through Claude”
   • [github] “Capture traces, generations, latency, and cost for your LLM-powered app.”
   • [github] “AI observability: Capture traces, generations, latency, and cost for your LLM-powered app.”
   • [community] “I've been using PostHog with my app for about a week now, and so far the results have been good. Pretty straightforward to integrate with a …”
   • [community] “I really wish you guys would do something about the llm integration, I tried using it two days ago to create a cohort of users using a sql q…”

   proof ↗⚑ flag

   Mixpanel~partialclaimed4/10

   Evidence shows official APIs (Query API, Raw/Engage Export API, Feature Flags API) and a JQL scripting interface for building custom analytics, which support programmatic integration, but there is no explicit mention of official client SDKs (e.g., JS/Python/mobile libraries), language coverage, or code samples that an AI-native builder would need. Missing for 10: explicit SDK documentation/list of languages, code snippets, GitHub repos, and independent corroboration of SDK reliability.

   • [claimed-docs] “The Mixpanel Query API endpoints can: Filter or break down Events * Return user profiles. * Return an activity/event stream for a user. * Re…”
   • [claimed-docs] “The Query API allows you to obtain the same calculated and formatted results that you see from a report in our web app.”
   • [claimed-docs] “Use the Feature Flags API to assign your users to variants for experiments, rollout, and releases.”
   • [claimed-docs] “Create custom queries using Javascript code”
   • [claimed-docs] “The JQL JavaScript code must define a main() function. This JavaScript will be compiled, and the main() function will be executed by the que…”

   proof ↗⚑ flag

9. ai-native userSubscribe to events via webhooks

   weight 2 · round to PostHog
   PostHog~partialclaimed4/10

   PostHog's CDP pipelines can send event data to Slack, PagerDuty, and other SaaS destinations, and can filter/reshape events and route them 'wherever it needs to go' in realtime, which implies webhook-style outbound delivery — but no evidence pack item explicitly names 'webhook' as a subscription mechanism or ties this to AI-agent consumption of the event stream. missing for 10: explicit webhook endpoint/subscription documentation, confirmation that arbitrary event streams can be pushed to a caller-defined webhook URL, and any AI-native usage example of subscribing via webhook.

   • [claimed-docs] “It pulls data in from the tools you already use, lets you filter and reshape events as they're ingested, and sends the result wherever it ne…”
   • [claimed-docs] “It pulls data in from the tools you already use, lets you filter and reshape events as they're ingested, and sends the result wherever it ne…”
   • [claimed-docs] “Slack: Send events to Slack channels SaaS tools: Send events to SaaS tools like Braze, Customer.io, and more”
   • [claimed-docs] “Send alerts to Slack, PagerDuty, or monitoring tools”
   • [github] “Create workflows that automate actions or send messages to your users.”

   proof ↗⚑ flag

   Mixpanel—none0/10

   No evidence of Mixpanel offering webhooks to push/subscribe to events; the evidence only covers exports (raw export API, warehouse connectors, MCP server for querying data), which are pull-based, not event-driven webhook subscriptions.

   ⚑ flag

Agentic features

1. ai-native userGet AI-generated insights and suggestions from my data inside the product

   weight 2 · round to Mixpanel
   PostHog!disputedcontradicted5/10

   PostHog markets in-product AI ('PostHog AI', 'self-driving products') that can suggest code fixes from stack traces, auto-write Hog functions, and turn product signals into researched reports/PRs (posthog-docs-12, posthog-docs-28, posthog-gh-1, posthog-probe-1). However a hands-on community report describes the in-app LLM feature failing a basic task (creating a SQL-based cohort) and severely degrading device performance, directly contradicting the polished 'insights on demand' claim (posthog-comm-8). missing for 10: independent corroboration that AI-generated insights/suggestions work reliably at scale, no rebuttal or fix documented for the reported failure.

   • [claimed-docs] “"Show me the full stack trace for the most recent crash, then propose a fix." – Pulls the stack trace, error message, and metadata so the ag…”
   • [claimed-docs] “Pipelines are built from Hog functions, which you can configure from a template, write yourself, or have PostHog AI write for you.”
   • [github] “Turn signals in your product data (errors, rage clicks, failed queries, and more) into researched reports and pull requests you review and m…”
   • [probe] “PROBE llms.txt: HTTP 200 at https://posthog.com/llms.txt # PostHog > PostHog is the platform for self-driving products. It pairs the full c…”
   • [community] “I really wish you guys would do something about the llm integration, I tried using it two days ago to create a cohort of users using a sql q…”

   proof ↗⚑ flag

   Mixpanel✓fullclaimed7/10

   Mixpanel's own docs/marketing describe AI-driven insight surfacing ('Agents monitor your product continuously, surface insights before you think to ask' and example outputs like 'Signups dipped 19% last week'), AI trend-spotting across replays/heatmaps, AI summaries of replay playlists, and natural-language querying via its MCP server. These directly match the story of AI-generated insights/suggestions inside the product. Missing for 10: independent/hands-on validation of these AI insight features' accuracy or usefulness, and more concrete product detail beyond marketing copy.

   • [claimed-docs] “Agents monitor your product continuously, surface insights before you think to ask, and tell your team exactly what to do next.”
   • [claimed-docs] “Diagnose what changed, and why, in seconds, delivered as a dashboard you and your team can keep working from.”
   • [claimed-docs] “Signups dipped 19% last week — a brief pullback against otherwise steady growth.”
   • [claimed-docs] “Use AI to spot trends across hundreds of replays and heatmaps, then zoom in yourself to get to the “why” behind every change.”
   • [claimed-docs] “Save playlists of replays based on events or properties, and quickly review with AI summaries, making it easy to revisit groups of replays t…”
   • [claimed-docs] “you can query events, funnels, flows, retention, session replays, and more using natural language — no dashboard navigation required.”
   • [claimed-docs] “Once connected, you can query events, funnels, flows, retention, session replays, and more using natural language — no dashboard navigation …”

   proof ↗⚑ flag

2. ai-native userSet up automations that run autonomously in the background

   weight 2 · round to PostHog
   PostHog~partialcommunity6/10

   PostHog's CDP pipelines and workflows can run automatically in realtime or on a schedule (posthog-docs-18, posthog-gh-7), with Hog functions optionally written by PostHog AI (posthog-docs-28), and alerts/feature-flag automation exist (posthog-docs-3, posthog-docs-5). However, evidence centers on data-pipeline/workflow automation rather than an AI-native agent explicitly setting up and running autonomous background jobs end-to-end, and one community report notes friction with AI-driven query/cohort creation (posthog-comm-8). Missing for 10: first-party documentation of an AI agent configuring and monitoring a fully autonomous background automation, independent success reports of such automations running unattended.

   • [claimed-docs] “It pulls data in from the tools you already use, lets you filter and reshape events as they're ingested, and sends the result wherever it ne…”
   • [github] “Create workflows that automate actions or send messages to your users.”
   • [claimed-docs] “Pipelines are built from Hog functions, which you can configure from a template, write yourself, or have PostHog AI write for you.”
   • [claimed-docs] “Build insights, assemble dashboards, share them, and set alerts on the numbers that matter.”
   • [community] “I really wish you guys would do something about the llm integration, I tried using it two days ago to create a cohort of users using a sql q…”

   proof ↗⚑ flag

   Mixpanel~partialclaimed3/10

   Mixpanel markets background 'Agents' that continuously monitor data and surface insights/recommendations autonomously (mixpanel-docs-44), suggesting some agentic automation capability, but there is no documentation on how a user actually configures or triggers these automations, what actions they can take, or independent confirmation they run unattended. missing for 10: setup/configuration docs for these agents, description of trigger conditions and autonomous actions, independent/hands-on evidence of them running in the background.

   • [claimed-docs] “Agents monitor your product continuously, surface insights before you think to ask, and tell your team exactly what to do next.”

   proof ↗⚑ flag

3. ai-native userDelegate tasks to a built-in AI assistant inside the product

   weight 3 · round to Mixpanel
   PostHog!disputedcontradicted4/10

   PostHog documents an in-product AI assistant/agent ("PostHog AI") that can write Hog functions, turn product signals into researched reports and PRs, and be steered from Slack/web/desktop/editor (posthog-docs-28, posthog-gh-1, posthog-gh-2), which matches the story of delegating tasks to a built-in assistant. However, a hands-on community report describes the LLM integration failing to perform a requested task (creating a cohort via SQL) and causing severe performance degradation, directly contradicting smooth delegation (posthog-comm-8). Missing for 10: first-party detail on the assistant's task scope/reliability, more independent corroboration of successful delegation, and resolution of the reported failure.

   • [claimed-docs] “Pipelines are built from Hog functions, which you can configure from a template, write yourself, or have PostHog AI write for you.”
   • [github] “Turn signals in your product data (errors, rage clicks, failed queries, and more) into researched reports and pull requests you review and m…”
   • [github] “You can steer it all from Slack, web, desktop (PostHog Desktop), or your own editor via the MCP.”
   • [community] “I really wish you guys would do something about the llm integration, I tried using it two days ago to create a cohort of users using a sql q…”

   proof ↗⚑ flag

   Mixpanel~partialclaimed4/10

   Mixpanel's marketing copy references built-in AI features such as 'Agents' that 'monitor your product continuously... and tell your team exactly what to do next,' AI-generated replay summaries, and AI trend-spotting across replays/heatmaps, suggesting some in-product AI assistant capability. However, these are thin marketing snippets rather than detailed product docs describing how a user actually delegates multi-step tasks to an assistant, and the only well-documented AI integration (MCP server) is for external AI assistants to query Mixpanel data, not a built-in assistant for users. Missing for 10: detailed first-party documentation of the 'Agents' feature and how tasks are delegated/executed, concrete UI/workflow examples, and independent hands-on corroboration.

   • [claimed-docs] “Agents monitor your product continuously, surface insights before you think to ask, and tell your team exactly what to do next.”
   • [claimed-docs] “Use AI to spot trends across hundreds of replays and heatmaps, then zoom in yourself to get to the “why” behind every change.”
   • [claimed-docs] “Save playlists of replays based on events or properties, and quickly review with AI summaries, making it easy to revisit groups of replays t…”
   • [claimed-docs] “Mixpanel provides a hosted Model Context Protocol (MCP) server that gives AI assistants direct access to your Mixpanel data.”

   proof ↗⚑ flag

4. ai-native userOperate the product with natural-language commands

   weight 2 · round to Mixpanel
   PostHog!disputedcontradicted5/10

   PostHog documents natural-language operation via its MCP integration—shipping feature flags, running HogQL queries, and debugging stack traces 'from a prompt' or 'plain text questions' (posthog-docs-12, posthog-docs-19, posthog-docs-24, posthog-docs-31, posthog-probe-4). However, a hands-on community report directly contradicts this: a user tried the LLM integration to create a cohort via SQL and it failed with an incorrect response, plus caused severe performance degradation (posthog-comm-8). missing for 10: independent corroboration that the failure was fixed, broader hands-on validation of natural-language commands across other product areas (analytics, experiments, session replay) beyond MCP docs.

   • [claimed-docs] “"Show me the full stack trace for the most recent crash, then propose a fix." – Pulls the stack trace, error message, and metadata so the ag…”
   • [claimed-docs] “with just plain text questions your agents can ship a feature flag from a prompt, dig into a stack trace without leaving your editor, run a …”
   • [claimed-docs] “your agents can ship a feature flag from a prompt, dig into a stack trace without leaving your editor, run a HogQL query through Claude”
   • [claimed-docs] “with just plain text questions your agents can ship a feature flag from a prompt, dig into a stack trace without leaving your editor”
   • [probe] “official MCP server documented at https://posthog.com/docs/model-context-protocol”
   • [community] “I really wish you guys would do something about the llm integration, I tried using it two days ago to create a cohort of users using a sql q…”

   proof ↗⚑ flag

   Mixpanel✓fullprobed8/10

   Mixpanel documents a hosted MCP server that lets AI assistants query events, funnels, flows, retention, and session replays using natural language without dashboard navigation, directly matching the story of operating the product via NL commands. This is first-party documentation without independent hands-on corroboration of the NL experience itself. Missing for 10: independent/community validation of the MCP natural-language workflow in practice, and detail on the breadth/reliability of commands beyond the listed examples.

   • [claimed-docs] “Mixpanel provides a hosted Model Context Protocol (MCP) server that gives AI assistants direct access to your Mixpanel data.”
   • [claimed-docs] “you can query events, funnels, flows, retention, session replays, and more using natural language — no dashboard navigation required.”
   • [claimed-docs] “Once connected, you can query events, funnels, flows, retention, session replays, and more using natural language — no dashboard navigation …”
   • [claimed-docs] “you can query events, funnels, flows, retention, session replays, and more using natural language — no dashboard navigation required”
   • [claimed-docs] “Create-Dashboard | Create a new dashboard with text cards and reports”
   • [probe] “official MCP server documented at https://docs.mixpanel.com/docs/mcp”

   proof ↗⚑ flag

Api quality

1. ai-native userExplore an interactive API reference with runnable examples

   weight 2 · round to PostHog
   PostHog~partialprobed3/10

   The evidence shows PostHog publishes a machine-readable OpenAPI spec (openapi.json) and API docs referencing HTTP-based event capture, which implies some API reference exists, but there is no direct evidence of an interactive reference page with runnable/try-it examples. missing for 10: confirmation of an interactive API console, runnable code snippets, or live 'try it' functionality in the docs.

   • [probe] “PROBE openapi: HTTP 200 at https://posthog.com/openapi.json — contains "openapi" key”
   • [claimed-docs] “capture events from any language that can send HTTP requests”

   proof ↗⚑ flag

   Mixpanel—none0/10

   While Mixpanel documents various APIs (Query API, Raw Export API, JQL), there is no evidence of an interactive API reference with runnable/try-it examples; a probe for a standard OpenAPI/Swagger spec returned 404s across all candidate paths, suggesting no interactive API console exists.

   • [probe] “PROBE openapi: all candidate paths 404 (https://docs.mixpanel.com/openapi.json, https://docs.mixpanel.com/swagger.json, https://docs.mixpane…”
   • [claimed-docs] “The Mixpanel Query API endpoints can: Filter or break down Events * Return user profiles. * Return an activity/event stream for a user. * Re…”
   • [claimed-docs] “The Query API allows you to obtain the same calculated and formatted results that you see from a report in our web app.”

   proof ↗⚑ flag

2. ai-native userDownload a machine-readable API spec (OpenAPI or equivalent)

   weight 2 · round to PostHog
   PostHog✓fullprobed9/10

   A probe confirms PostHog serves a machine-readable OpenAPI spec at https://posthog.com/openapi.json (HTTP 200, contains 'openapi' key), and the API docs reference programmatic access via HTTP. Missing for 10: no explicit first-party docs page describing the OpenAPI spec's usage/versioning or independent community confirmation of its completeness.

   • [probe] “PROBE openapi: HTTP 200 at https://posthog.com/openapi.json — contains "openapi" key”
   • [claimed-docs] “capture events from any language that can send HTTP requests”

   proof ↗⚑ flag

   Mixpanel—none0/10

   Mixpanel has extensive REST/Query/Export APIs, but direct probes for a machine-readable OpenAPI/Swagger spec (openapi.json, swagger.json, etc.) all returned 404, and no documentation evidence points to a downloadable spec file—only human-readable reference docs are cited.

   • [probe] “PROBE openapi: all candidate paths 404 (https://docs.mixpanel.com/openapi.json, https://docs.mixpanel.com/swagger.json, https://docs.mixpane…”
   • [claimed-docs] “The Mixpanel Query API endpoints can: Filter or break down Events * Return user profiles. * Return an activity/event stream for a user. * Re…”
   • [claimed-docs] “The Query API allows you to obtain the same calculated and formatted results that you see from a report in our web app.”

   proof ↗⚑ flag

3. ai-native userTest against a sandbox environment without touching production data

   weight 1 · round drawn
   PostHog—none0/10

   The evidence pack covers self-hosting (Docker Compose/MIT license), MCP integration, feature flags, and experiments, but nothing describes a dedicated sandbox/staging environment or test-mode data isolation from production. Self-hosting is a deployment option, not a sandbox testing feature, so it does not satisfy this story.

   ⚑ flag

   Mixpanel—none0/10

   No evidence pack item mentions a sandbox, test/dev project, or any mechanism to let an AI assistant query or write data without touching the production Mixpanel dataset — the MCP server docs (mixpanel-docs-10, mixpanel-docs-19) only describe querying live production data. Missing for 10: any documented sandbox/test project mode, data isolation for AI testing, or non-production environment support.

   • [claimed-docs] “Mixpanel provides a hosted Model Context Protocol (MCP) server that gives AI assistants direct access to your Mixpanel data.”
   • [claimed-docs] “you can query events, funnels, flows, retention, session replays, and more using natural language — no dashboard navigation required.”
   • [probe] “official MCP server documented at https://docs.mixpanel.com/docs/mcp”

   proof ↗⚑ flag

4. ai-native userRely on versioned APIs with a documented deprecation policy

   weight 2 · round drawn
   PostHog—none0/10

   There is an OpenAPI spec probe (posthog-probe-3) confirming an API exists, but no evidence anywhere in the pack of API versioning scheme or a documented deprecation policy for that API.

   • [probe] “PROBE openapi: HTTP 200 at https://posthog.com/openapi.json — contains "openapi" key”

   proof ↗⚑ flag

   Mixpanel—none0/10

   No evidence pack item mentions API versioning, version numbers, or a documented deprecation/sunset policy for Mixpanel's APIs; the openapi.json probe returned 404s and no versioning docs were found. missing for 10: API version scheme documentation, deprecation/sunset policy, changelog or migration guides for breaking changes.

   • [probe] “PROBE openapi: all candidate paths 404 (https://docs.mixpanel.com/openapi.json, https://docs.mixpanel.com/swagger.json, https://docs.mixpane…”

   proof ↗⚑ flag

1. ai-native userPerform bulk operations across many items at once

   weight 2 · round to Mixpanel
   PostHog~partialcommunity4/10

   PostHog offers SQL for querying/modifying data at scale, a CDP pipeline that can reshape/filter events in bulk, and a CLI/API for scripting operations, which could underpin bulk actions, but there is no explicit documentation of a 'bulk update/delete/tag N items' feature, and a hands-on report shows the AI/SQL interface failing on a basic bulk task (creating a cohort via SQL). missing for 10: explicit bulk-operation API/UI (e.g., mass edit flags, batch delete events), documented bulk-scale limits, and independent confirmation that SQL/CLI bulk actions work reliably for AI-driven workflows.

   • [claimed-docs] “You can use SQL (Structured Query Language) throughout PostHog to manage, query, and modify data.”
   • [claimed-docs] “Sync Stripe, Postgres, Salesforce, HubSpot, and dozens of other sources, then query them alongside your events, persons, and sessions in one…”
   • [claimed-docs] “It pulls data in from the tools you already use, lets you filter and reshape events as they're ingested, and sends the result wherever it ne…”
   • [claimed-docs] “The PostHog CLI lets you use PostHog from your terminal, your coding agents, local scripts, and CI/CD pipelines.”
   • [community] “I really wish you guys would do something about the llm integration, I tried using it two days ago to create a cohort of users using a sql q…”

   proof ↗⚑ flag

   Mixpanel~partialclaimed5/10

   Mixpanel offers APIs (Raw Event Export, JQL, Query API) that allow programmatic, high-volume data operations (bulk export, custom scripted queries, bulk filtering/cohort retrieval), and its MCP server lets AI agents query events, funnels, retention, etc. via natural language. However, the MCP tool list only shows single-item actions like 'Create-Dashboard', with no documented bulk-operation tools (e.g., batch create/update/delete across many items) for an AI-native agent to invoke. missing for 10: explicit AI-agent-invocable bulk operation tools (batch create/update/delete), independent confirmation of bulk actions succeeding at scale via MCP or API in practice.

   • [claimed-docs] “To export raw event data, use the Raw Event Export API. The events payload is returned as JSONL”
   • [claimed-docs] “To export raw event data, use the Raw Event Export API... This method is recommended for high volume exports.”
   • [claimed-docs] “Create custom queries using Javascript code”
   • [claimed-docs] “The Mixpanel Query API endpoints can: Filter or break down Events * Return user profiles. * Return an activity/event stream for a user. * Re…”
   • [claimed-docs] “The JQL JavaScript code must define a main() function. This JavaScript will be compiled, and the main() function will be executed by the que…”
   • [claimed-docs] “Mixpanel provides a hosted Model Context Protocol (MCP) server that gives AI assistants direct access to your Mixpanel data.”
   • [claimed-docs] “Create-Dashboard | Create a new dashboard with text cards and reports”
   • [claimed-docs] “you can query events, funnels, flows, retention, session replays, and more using natural language — no dashboard navigation required.”

   proof ↗⚑ flag

2. ai-native userDefine rules that trigger actions automatically on events

   weight 3 · round to PostHog
   PostHog✓fullclaimed7/10

   PostHog's CDP/Pipelines ('destinations') and alerting system let users define trigger conditions on events (errors, rage clicks, property changes, etc.) that automatically fire actions like Slack alerts, CRM syncs, PagerDuty notifications, or custom Hog functions, and workflows can automate actions/messages to users. missing for 10: no independent/hands-on verification of complex multi-condition automation rules, and no detail on latency/reliability of the automation engine at scale.

   • [claimed-docs] “It pulls data in from the tools you already use, lets you filter and reshape events as they're ingested, and sends the result wherever it ne…”
   • [claimed-docs] “It pulls data in from the tools you already use, lets you filter and reshape events as they're ingested, and sends the result wherever it ne…”
   • [claimed-docs] “Pipelines are built from Hog functions, which you can configure from a template, write yourself, or have PostHog AI write for you.”
   • [claimed-docs] “Slack: Send events to Slack channels SaaS tools: Send events to SaaS tools like Braze, Customer.io, and more”
   • [claimed-docs] “Send alerts to Slack, PagerDuty, or monitoring tools”
   • [github] “Create workflows that automate actions or send messages to your users.”

   proof ↗⚑ flag

   Mixpanel—none0/10

   Mixpanel's evidence covers analytics reporting, exports, session replay, and an MCP server for querying data with natural language, but there is no documented feature for defining rule-based triggers that automatically fire actions on events (e.g., alerts, webhooks, workflow automation). The closest mention, 'Agents monitor your product continuously... tell your team what to do next' (mixpanel-docs-44), describes passive AI insight surfacing, not user-defined event-triggered automation.

   • [claimed-docs] “Agents monitor your product continuously, surface insights before you think to ask, and tell your team exactly what to do next.”
   • [claimed-docs] “Use the Feature Flags API to assign your users to variants for experiments, rollout, and releases.”
   • [claimed-docs] “you can query events, funnels, flows, retention, session replays, and more using natural language — no dashboard navigation required.”

   proof ↗⚑ flag

3. ai-native userSchedule recurring jobs or workflows

   weight 2 · round to PostHog
   PostHog~partialclaimed5/10

   PostHog's CDP/pipelines can run 'in realtime or on a schedule' (posthog-docs-18) and GitHub notes 'workflows that automate actions or send messages to your users' (posthog-gh-7), suggesting some recurring automation capability, but there's no documentation of a general-purpose recurring job/workflow scheduler accessible via API, CLI, or MCP for AI-native orchestration. Missing for 10: explicit docs on scheduling arbitrary jobs/workflows via API or agent tooling, cron-like configuration details, and independent confirmation that scheduled pipelines work reliably.

   • [claimed-docs] “It pulls data in from the tools you already use, lets you filter and reshape events as they're ingested, and sends the result wherever it ne…”
   • [github] “Create workflows that automate actions or send messages to your users.”
   • [claimed-docs] “Pipelines are built from Hog functions, which you can configure from a template, write yourself, or have PostHog AI write for you.”

   proof ↗⚑ flag

   Mixpanel—none0/10

   Mixpanel is an analytics platform with export APIs, MCP server for queries, and dashboards, but there is no evidence of any capability to schedule recurring jobs or workflows (e.g., scheduled reports, cron-like automation, recurring data syncs beyond warehouse mirroring). No scheduling/automation feature is documented anywhere in the evidence pack.

   ⚑ flag

Ai analytics ops

1. ai-native userHave an agent build a dashboard of my key metrics end-to-end via the API or MCP server

   weight 3 · round to Mixpanel
   PostHog~partialprobed6/10

   PostHog has a documented official MCP server and public API (openapi.json) that let an agent run trends, funnels, retention, and SQL queries, ship feature flags, and inspect stack traces from an editor/agent (posthog-docs-2, 12, 19, 24, 31, posthog-probe-4, posthog-probe-3). Dashboard assembly, sharing, and alerting are documented product capabilities (posthog-docs-3, 25), but the evidence never explicitly shows the MCP/API path being used to assemble and publish a full dashboard end-to-end, and a hands-on community report describes the AI/LLM integration failing on a related task (cohort creation) and causing severe UI slowdown (posthog-comm-8), raising doubts about reliability of agent-driven workflows. Missing for 10: explicit documentation/demo of an agent creating and saving a dashboard object (not just insights/queries) via MCP or API, and independent corroboration that this specific end-to-end flow works reliably.

   • [claimed-docs] “Run trends, funnels, retention, and SQL queries from any MCP client or AI editor.”
   • [claimed-docs] “Build insights, assemble dashboards, share them, and set alerts on the numbers that matter.”
   • [claimed-docs] “"Show me the full stack trace for the most recent crash, then propose a fix." – Pulls the stack trace, error message, and metadata so the ag…”
   • [claimed-docs] “with just plain text questions your agents can ship a feature flag from a prompt, dig into a stack trace without leaving your editor, run a …”
   • [claimed-docs] “your agents can ship a feature flag from a prompt, dig into a stack trace without leaving your editor, run a HogQL query through Claude”
   • [claimed-docs] “with just plain text questions your agents can ship a feature flag from a prompt, dig into a stack trace without leaving your editor”
   • [probe] “PROBE openapi: HTTP 200 at https://posthog.com/openapi.json — contains "openapi" key”
   • [probe] “official MCP server documented at https://posthog.com/docs/model-context-protocol”
   • [community] “I really wish you guys would do something about the llm integration, I tried using it two days ago to create a cohort of users using a sql q…”

   proof ↗⚑ flag

   Mixpanel✓fullprobed8/10

   Mixpanel provides a documented hosted MCP server explicitly with a Create-Dashboard action ('Create a new dashboard with text cards and reports') plus natural-language querying of events, funnels, flows, retention, session replays, complementing the Query/Export APIs an agent could also use end-to-end. This directly matches the story of an agent building a dashboard via MCP or API, though there's no independent hands-on report confirming a real agent successfully completed this end-to-end. missing for 10: independent/hands-on verification of an agent actually building a dashboard via MCP or API, and more detail on API-only dashboard creation outside MCP.

   • [claimed-docs] “Mixpanel provides a hosted Model Context Protocol (MCP) server that gives AI assistants direct access to your Mixpanel data.”
   • [claimed-docs] “Create-Dashboard | Create a new dashboard with text cards and reports”
   • [claimed-docs] “you can query events, funnels, flows, retention, session replays, and more using natural language — no dashboard navigation required.”
   • [claimed-docs] “Once connected, you can query events, funnels, flows, retention, session replays, and more using natural language — no dashboard navigation …”
   • [claimed-docs] “you can query events, funnels, flows, retention, session replays, and more using natural language — no dashboard navigation required”
   • [probe] “official MCP server documented at https://docs.mixpanel.com/docs/mcp”
   • [claimed-docs] “The Mixpanel Query API endpoints can: Filter or break down Events * Return user profiles. * Return an activity/event stream for a user. * Re…”
   • [claimed-docs] “The Query API allows you to obtain the same calculated and formatted results that you see from a report in our web app.”

   proof ↗⚑ flag

Dashboards

1. developerCreate and manage dashboards and saved insights programmatically via the API

   weight 2 · round to PostHog
   PostHog~partialprobed4/10

   The evidence confirms PostHog exposes a general REST API (openapi.json spec) and documents building/sharing dashboards and insights with alerts, implying programmatic access is plausible, but no citation explicitly documents dashboard/insight CRUD API endpoints or SDK examples for managing them programmatically. missing for 10: explicit API docs/endpoints for creating/updating dashboards and insights, code examples for programmatic dashboard management, independent confirmation of this specific API surface.

   • [probe] “PROBE openapi: HTTP 200 at https://posthog.com/openapi.json — contains "openapi" key”
   • [claimed-docs] “Build insights, assemble dashboards, share them, and set alerts on the numbers that matter.”
   • [claimed-docs] “Build trends, funnels, retention, paths, stickiness, and lifecycle insights on the events you already send to PostHog, then save them to das…”

   proof ↗⚑ flag

   Mixpanel~partialprobed3/10

   Mixpanel's hosted MCP server exposes a 'Create-Dashboard' tool that can create dashboards with text cards and reports, and lets you query events/funnels/retention/etc. programmatically, but this is MCP-tool access for AI assistants, not a documented general-purpose REST API for full CRUD (create/update/delete) of dashboards and saved insights. The Query API (docs-39, docs-43) only supports reading/filtering data, not creating or managing dashboard objects. Missing for 10: explicit REST API endpoints for creating, updating, and deleting dashboards/saved insights outside the MCP context, and any evidence of programmatic dashboard sharing/permissions management.

   • [claimed-docs] “Mixpanel provides a hosted Model Context Protocol (MCP) server that gives AI assistants direct access to your Mixpanel data.”
   • [claimed-docs] “Create-Dashboard | Create a new dashboard with text cards and reports”
   • [claimed-docs] “you can query events, funnels, flows, retention, session replays, and more using natural language — no dashboard navigation required.”
   • [claimed-docs] “The Mixpanel Query API endpoints can: Filter or break down Events * Return user profiles. * Return an activity/event stream for a user. * Re…”
   • [claimed-docs] “The Query API allows you to obtain the same calculated and formatted results that you see from a report in our web app.”
   • [probe] “official MCP server documented at https://docs.mixpanel.com/docs/mcp”

   proof ↗⚑ flag

2. product-managerCompose saved insights into shareable dashboards for my team

   weight 2 · round to PostHog
   PostHog✓fullclaimed8/10

   Docs explicitly state users can build insights, save them to dashboards, share them, and set alerts (posthog-docs-3, posthog-docs-25), directly matching the story of composing saved insights into shareable dashboards. Missing for 10: no independent/hands-on corroboration of the dashboard-sharing workflow specifically, and no detail on team-level permissions or collaboration features around shared dashboards.

   • [claimed-docs] “Build insights, assemble dashboards, share them, and set alerts on the numbers that matter.”
   • [claimed-docs] “Build trends, funnels, retention, paths, stickiness, and lifecycle insights on the events you already send to PostHog, then save them to das…”
   • [claimed-docs] “Build trends, funnels, retention, paths, stickiness, and lifecycle insights on the events you already send to PostHog”

   proof ↗⚑ flag

   Mixpanel~partialclaimed6/10

   Mixpanel's marketing/docs reference dashboards as an output (e.g. MCP tool 'Create-Dashboard' with text cards and reports, and marketing copy about 'a dashboard you and your team can keep working from'), implying saved insights can be composed into shareable team dashboards. However there is no first-party documentation detailing the dashboard-building UI, sharing/permissions workflow, or team collaboration mechanics beyond these brief mentions. Missing for 10: dedicated dashboard-creation docs, sharing/permission controls, and independent/hands-on evidence of team usage.

   • [claimed-docs] “Create-Dashboard | Create a new dashboard with text cards and reports”
   • [claimed-docs] “Diagnose what changed, and why, in seconds, delivered as a dashboard you and your team can keep working from.”
   • [claimed-docs] “Connect your product data to your company strategy in one interactive view, so teams can align faster, act with confidence, and stay account…”

   proof ↗⚑ flag

3. product-managerSubscribe to dashboards and get alerted when a key metric moves abnormally

   weight 1 · round to PostHog
   PostHog✓fullclaimed7/10

   PostHog docs explicitly state dashboards/insights can be saved, shared, and users can 'get alerted when they move' (posthog-docs-3, posthog-docs-25), and alerts can be routed to Slack/PagerDuty (posthog-intdir-2), directly matching the PM story of dashboard subscriptions with abnormal-metric alerts. Missing for 10: no independent/hands-on confirmation of the alerting UX, no detail on anomaly-detection thresholds, and no explicit mention of subscribing to an entire dashboard vs individual insights.

   • [claimed-docs] “Build insights, assemble dashboards, share them, and set alerts on the numbers that matter.”
   • [claimed-docs] “Build trends, funnels, retention, paths, stickiness, and lifecycle insights on the events you already send to PostHog, then save them to das…”
   • [claimed-docs] “Send alerts to Slack, PagerDuty, or monitoring tools”

   proof ↗⚑ flag

   Mixpanel~partialclaimed4/10

   Evidence shows Mixpanel has AI 'Agents' that monitor products continuously and surface anomalies (e.g., 'Signups dipped 19% last week') and can 'tell your team exactly what to do next,' suggesting some proactive alerting capability, but there is no explicit documentation of a dashboard subscription mechanism or configurable alert thresholds for PMs. missing for 10: explicit dashboard subscription feature, alert configuration/thresholds, delivery channels (email/Slack) for alerts.

   • [claimed-docs] “Agents monitor your product continuously, surface insights before you think to ask, and tell your team exactly what to do next.”
   • [claimed-docs] “Signups dipped 19% last week — a brief pullback against otherwise steady growth.”
   • [claimed-docs] “Diagnose what changed, and why, in seconds, delivered as a dashboard you and your team can keep working from.”

   proof ↗⚑ flag

Identity

1. developerBulk-import historical events from another analytics tool or a data export into the platform

   weight 1 · round to Mixpanel
   PostHog~partialcommunity3/10

   There's no first-party documentation of a dedicated bulk-import tool for historical events from another analytics tool, but the data warehouse/CDP features allow syncing external data sources into PostHog for unified querying, and a community comment notes Mixpanel-library compatibility making switching 'really easy.' missing for 10: explicit import API/tool for historical event data, documented data-export ingestion pipeline, and first-party confirmation of the Mixpanel-compatibility claim.

   • [claimed-docs] “Sync Stripe, Postgres, Salesforce, HubSpot, and dozens of other sources, then query them alongside your events, persons, and sessions in one…”
   • [claimed-docs] “It pulls data in from the tools you already use, lets you filter and reshape events as they're ingested, and sends the result wherever it ne…”
   • [community] “Great idea to be compatible with Mixpanel libraries! Makes switching over really easy”

   proof ↗⚑ flag

   Mixpanel~partialclaimed6/10

   Mixpanel supports ingesting historical/bulk data via Warehouse Connectors (syncing from Snowflake, BigQuery, Databricks, Redshift, Postgres) and has a general import/ingestion API implied by its event JSON model, but the evidence pack lacks explicit documentation of a dedicated bulk 'Import API' for historical event backfill or migration from another analytics tool. missing for 10: explicit historical/bulk import API docs, guidance on importing from other analytics platforms (e.g., Amplitude/Segment migration), and confirmation of backdated event timestamp support.

   • [claimed-docs] “you can sync data from data warehouses like Snowflake, BigQuery, Databricks, Redshift, and Postgres to Mixpanel”
   • [claimed-docs] “With Warehouse Connectors, you can sync data from data warehouses like Snowflake, BigQuery, Databricks, Redshift, and Postgres to Mixpanel.”
   • [claimed-docs] “you can sync data from data warehouses like Snowflake, BigQuery, Databricks, Redshift, and Postgres to Mixpanel.”
   • [claimed-docs] “SET CHANGE\TRACKING = TRUE; # or ALTER VIEW SET CHANGE\TRACKING = TRUE; \\\ Mixpanel will create and manage the necessary STREAM objects in …”
   • [claimed-docs] “SET TBLPROPERTIES (delta.enableChangeDataFeed = true); \\\ \\Considerations when using Mirror with Databricks:\\ \ Mirror is not supported o…”
   • [claimed-docs] “Every data point sent to Mixpanel is stored as JSON in our data store.”

   proof ↗⚑ flag

2. developerIdentify users across devices and merge anonymous pre-signup activity into their identified profile

   weight 2 · round drawn
   PostHog—none0/10

   The evidence pack covers event capture, analytics, session replay, feature flags, experiments, and CDP integrations, but nowhere mentions identify()/alias() calls, distinct_id merging, or cross-device/anonymous-to-identified user reconciliation. This is a standard axis for an event-ingestion/analytics product, but no evidence supports it here. Missing for 10: any mention of identify/alias APIs, anonymous-to-known user merging, or cross-device identity resolution documentation.

   ⚑ flag

   Mixpanel—none0/10

   The evidence pack contains no mention of Mixpanel's identity resolution features (e.g., identify(), $merge, alias, distinct_id merging) for tying anonymous pre-signup activity to an identified user profile. While this is a well-known Mixpanel capability generally, none of the provided docs, community posts, or probes reference it, so per the evidence-only rule this must be judged as unsupported.

   ⚑ flag

Instrumentation

1. product-managerCapture clicks, pageviews, and form interactions automatically without writing tracking code for each event

   weight 2 · round drawn
   PostHog—none0/10

   The evidence pack discusses capturing custom events via SDKs/HTTP requests and building insights, but nowhere states that PostHog automatically captures clicks, pageviews, or form interactions without instrumenting code (no mention of autocapture). Missing for 10: any documented autocapture feature, evidence of automatic click/pageview/form tracking, or an example of a PM enabling capture with zero code.

   • [claimed-docs] “Custom events you capture from any SDK, each with the properties you attach to them.”
   • [claimed-docs] “capture events from any language that can send HTTP requests”

   proof ↗⚑ flag

   Mixpanel—none0/10

   The evidence pack covers Mixpanel's event model, funnels, retention, exports, warehouse connectors, and session replay, but nowhere documents an autocapture/autotrack feature that automatically records clicks, pageviews, and form interactions without custom instrumentation. Session Replay (mixpanel-docs-4/23/28/33) records video sessions but does not equate to automatic event-level tracking for analytics events.

   ⚑ flag

2. developerSend events from web, mobile, and backend apps through official SDKs for the major languages and platforms

   weight 3 · round to PostHog
   PostHog✓fullcommunity7/10

   Docs confirm event capture from "any SDK" and from any HTTP-capable language, and community evidence corroborates real-world use of the iOS SDK integrating smoothly, supporting broad SDK coverage across platforms. Missing for 10: an explicit enumerated list of official SDKs per language/platform (web, Android, Node, Python, etc.) and independent confirmation for backend/server SDKs beyond the HTTP API and iOS mention.

   • [claimed-docs] “Custom events you capture from any SDK, each with the properties you attach to them.”
   • [claimed-docs] “capture events from any language that can send HTTP requests”
   • [community] “I've been using PostHog with my app for about a week now, and so far the results have been good. Pretty straightforward to integrate with a …”
   • [community] “Great idea to be compatible with Mixpanel libraries! Makes switching over really easy”

   proof ↗⚑ flag

   Mixpanel—none0/10

   The evidence pack never mentions any specific SDKs (JavaScript, iOS, Android, Python, Java, etc.) for sending events from web, mobile, or backend apps — it only discusses tracking methods generically (warehouse connectors, event ingestion server rewrite, export APIs) without documenting an SDK catalog. Missing for 10: explicit mention of official web SDK, mobile SDKs (iOS/Android), backend/server SDKs, and any platform coverage list.

   • [community] “Mixpanel rewrote their event-ingestion server from Erlang to Python after two years, citing that 'the code has become difficult to maintain.…”
   • [claimed-docs] “Every data point sent to Mixpanel is stored as JSON in our data store.”
   • [claimed-docs] “SET CHANGE\TRACKING = TRUE; # or ALTER VIEW SET CHANGE\TRACKING = TRUE; \\\ Mixpanel will create and manage the necessary STREAM objects in …”

   proof ↗⚑ flag

Ai analytics ops

1. ai-native userHave an agent create, target, and toggle feature flags through the API or MCP server

   weight 2 · round to PostHog
   PostHog~partialprobed6/10

   PostHog has an official MCP server (posthog-probe-4) and docs explicitly state that 'with just plain text questions your agents can ship a feature flag from a prompt' (posthog-docs-19, posthog-docs-24, posthog-docs-31), showing agent-driven flag creation via MCP. An OpenAPI spec exists (posthog-probe-3) implying REST API access to flags, but no evidence explicitly documents targeting rules or toggling flags via API/MCP endpoints specifically. missing for 10: explicit documentation of flag targeting (cohorts/properties) and toggle actions via MCP/API, and independent hands-on confirmation that agent-driven flag creation works reliably.

   • [claimed-docs] “with just plain text questions your agents can ship a feature flag from a prompt, dig into a stack trace without leaving your editor, run a …”
   • [claimed-docs] “your agents can ship a feature flag from a prompt, dig into a stack trace without leaving your editor, run a HogQL query through Claude”
   • [claimed-docs] “with just plain text questions your agents can ship a feature flag from a prompt, dig into a stack trace without leaving your editor”
   • [probe] “official MCP server documented at https://posthog.com/docs/model-context-protocol”
   • [probe] “PROBE openapi: HTTP 200 at https://posthog.com/openapi.json — contains "openapi" key”

   proof ↗⚑ flag

   Mixpanel~partialclaimed4/10

   Mixpanel documents a Feature Flags API for assigning users to variants for experiments, rollout, and releases (mixpanel-docs-36), and a separate hosted MCP server for natural-language querying and dashboard creation (mixpanel-docs-10/11/19). However, there is no evidence that flag creation or toggling is exposed via the MCP server, nor documentation confirming full CRUD (create/toggle) operations through the API beyond targeting/assignment. missing for 10: MCP tool list including feature-flag creation/toggle, explicit API endpoints for creating and toggling flags, and any hands-on confirmation of agent-driven flag management.

   • [claimed-docs] “Use the Feature Flags API to assign your users to variants for experiments, rollout, and releases.”
   • [claimed-docs] “Mixpanel provides a hosted Model Context Protocol (MCP) server that gives AI assistants direct access to your Mixpanel data.”
   • [claimed-docs] “Create-Dashboard | Create a new dashboard with text cards and reports”
   • [claimed-docs] “you can query events, funnels, flows, retention, session replays, and more using natural language — no dashboard navigation required.”

   proof ↗⚑ flag

Experimentation

1. product-managerRun A/B experiments with goal metrics and statistical significance reported on the results

   weight 3 · round to PostHog
   PostHog✓fullclaimed8/10

   PostHog's Experiments docs explicitly describe defining variants, picking goal metrics, randomizing users, tracking exposures, and running statistics (Bayesian or frequentist) to determine significance, which directly matches the story. Missing for 10: independent/hands-on validation of the statistical reporting accuracy and no community corroboration specific to experiments (only analytics/session-replay community feedback exists).

   • [claimed-docs] “PostHog randomizes your users, tracks their exposures, and runs the statistics – Bayesian or frequentist – so you get a real answer instead …”
   • [claimed-docs] “You define variants, pick the metrics you care about, and PostHog randomizes your users, tracks their exposures, and runs the statistics – B…”
   • [claimed-docs] “You define variants, pick the metrics you care about, and PostHog randomizes your users, tracks their exposures, and runs the statistics – B…”
   • [claimed-docs] “PostHog randomizes your users, tracks their exposures, and runs the statistics – Bayesian or frequentist – so you get a real answer instead …”

   proof ↗⚑ flag

   Mixpanel~partialclaimed5/10

   Mixpanel documents a Feature Flags API for assigning variants to experiments/rollouts and marketing copy claims users can 'design, run, and analyze tests and rollouts' alongside analytics, indicating some experimentation capability. However, there is no explicit documentation of goal-metric configuration or statistical-significance calculation/reporting for A/B tests. missing for 10: explicit stat-sig reporting docs, goal-metric setup workflow, independent corroboration of experiment results UI.

   • [claimed-docs] “Close the gap between shipping fast and shipping right by designing, running, and analyzing tests and rollouts in the same place you do all …”
   • [claimed-docs] “Use the Feature Flags API to assign your users to variants for experiments, rollout, and releases.”

   proof ↗⚑ flag

Feature flags

1. developerRoll out features behind flags targeted by user properties, cohorts, and percentage rollouts

   weight 3 · round to PostHog
   PostHog✓fullclaimed8/10

   PostHog docs explicitly describe flagging changes and rolling out to a percentage of users, and targeting by person property, cohort, or group, which directly matches the story. missing for 10: no independent/hands-on corroboration of targeting accuracy or percentage rollout behavior beyond vendor docs.

   • [claimed-docs] “Wrap a change in a flag, roll it out to 1% of users, watch what happens, and turn it off the moment something looks wrong – no redeploy, no …”
   • [claimed-docs] “Roll out by person property, cohort, or group, then see the session replays, events, and exceptions from the people who got the flag”

   proof ↗⚑ flag

   Mixpanel~partialclaimed5/10

   Mixpanel documents a Feature Flags API for assigning users to variants for experiments, rollouts, and releases, and separately supports cohort creation via user properties, plus a marketing page claiming tests/rollouts are done in the same place as analytics. However, there's no documentation showing explicit percentage-rollout controls or cohort/property-based targeting rules specifically applied to feature flags. missing for 10: detailed docs on percentage-based rollout configuration, explicit flag targeting rules by user property/cohort, and any independent/hands-on confirmation of the flagging workflow.

   • [claimed-docs] “Use the Feature Flags API to assign your users to variants for experiments, rollout, and releases.”
   • [claimed-docs] “Properties allow you to create groups of users (aka cohorts) and also enable you to filter for certain events or users.”
   • [claimed-docs] “Close the gap between shipping fast and shipping right by designing, running, and analyzing tests and rollouts in the same place you do all …”
   • [claimed-docs] “The Mixpanel Query API endpoints can: Filter or break down Events * Return user profiles. * Return an activity/event stream for a user. * Re…”

   proof ↗⚑ flag

Behavioral analysis

1. product-managerBuild multi-step funnels with filters and breakdowns to find where users drop off

   weight 3 · round to Mixpanel
   PostHog✓fullclaimed7/10

   PostHog docs explicitly state funnels are a first-class insight type built on captured events ('Build trends, funnels, retention, paths, stickiness...') alongside dashboards, alerts, and saved insights, which covers the core funnel-building use case for a PM. However, the evidence pack doesn't explicitly detail funnel-specific filter/breakdown UI mechanics beyond the general mention. Missing for 10: explicit documentation or screenshots of step-level filters and breakdown-by-property configuration within funnels, and independent user validation of funnel usability.

   • [claimed-docs] “Build trends, funnels, retention, paths, stickiness, and lifecycle insights on the events you already send to PostHog”
   • [claimed-docs] “Build trends, funnels, retention, paths, stickiness, and lifecycle insights on the events you already send to PostHog, then save them to das…”
   • [claimed-docs] “Custom events you capture from any SDK, each with the properties you attach to them.”
   • [claimed-docs] “Build insights, assemble dashboards, share them, and set alerts on the numbers that matter.”

   proof ↗⚑ flag

   Mixpanel✓fullcommunity8/10

   Mixpanel's funnels report explicitly measures conversion through a series of events, properties enable filtering/cohorting, and the Query API supports breakdowns and filtering of events; marketing copy also touts discovering 'critical drop-off points.' Missing for 10: a hands-on community example specifically confirming breakdown usage inside funnels (only adjacent Flow feature has direct user testimonials) and independent verification of the breakdown UI itself.

   • [claimed-docs] “Measure conversions through a series of events”
   • [claimed-docs] “Properties allow you to create groups of users (aka cohorts) and also enable you to filter for certain events or users.”
   • [claimed-docs] “The Mixpanel Query API endpoints can: Filter or break down Events * Return user profiles. * Return an activity/event stream for a user. * Re…”
   • [claimed-docs] “Discover critical drop-off points and calibrate user journeys to optimize every touchpoint for maximum engagement, retention, and growth.”
   • [community] “On Mixpanel's Flow feature: 'This is the coolest thing I have seen in Analytics for a long time. This really gets to the heart of most of th…”

   proof ↗⚑ flag

2. product-managerAnalyze behavior at the account or company level, not just per user, for B2B products

   weight 2 · round drawn
   PostHog~partialclaimed4/10

   PostHog documents a group concept for targeting (person property, cohort, or group) used in feature flag rollouts, implying group-level segmentation exists, but the evidence pack never explicitly ties 'groups' to trends/funnels/retention insights or shows account-level rollups in analytics dashboards. missing for 10: explicit docs on group-level funnels/retention insights, account/company entity definition and configuration, examples of B2B account analytics dashboards.

   • [claimed-docs] “Roll out by person property, cohort, or group, then see the session replays, events, and exceptions from the people who got the flag”
   • [claimed-docs] “Build trends, funnels, retention, paths, stickiness, and lifecycle insights on the events you already send to PostHog”
   • [claimed-docs] “Build trends, funnels, retention, paths, stickiness, and lifecycle insights on the events you already send to PostHog, then save them to das…”

   proof ↗⚑ flag

   Mixpanel~partialcommunity4/10

   Mixpanel does support account/company-level (group) analytics for B2B use cases, but the evidence pack lacks first-party docs describing this feature directly — the only confirmation comes from a community report noting that group-level/account-level analytics 'useful for B2B SaaS' exists but is 'tucked away on one of Mixpanel's higher price plans,' limiting access for many customers. Missing for 10: first-party documentation of the Group Analytics feature, details on setup/limits, and independent hands-on validation beyond the pricing-gate complaint.

   • [community] “Group-level/account-level analytics useful for B2B SaaS is 'tucked away on one of Mixpanel's higher price plans,' limiting access for smalle…”

   proof ↗⚑ flag

3. product-managerMeasure retention over time and slice it by behavioral cohorts

   weight 3 · round to Mixpanel
   PostHog✓fullclaimed7/10

   PostHog explicitly ships retention as a core insight type (posthog-docs-15, posthog-docs-25) alongside cohort support elsewhere in the product (posthog-docs-6 references rolling out and analyzing by cohort, posthog-docs-8/17 SQL for custom cohort queries). This directly matches the PM story of measuring retention and slicing by behavioral cohorts, though the evidence doesn't show a dedicated doc excerpt showing retention insight breakdown-by-cohort UI directly. missing for 10: an explicit doc/screenshot showing retention insight with cohort breakdown control, and independent hands-on confirmation of this specific workflow.

   • [claimed-docs] “Build trends, funnels, retention, paths, stickiness, and lifecycle insights on the events you already send to PostHog”
   • [claimed-docs] “Build trends, funnels, retention, paths, stickiness, and lifecycle insights on the events you already send to PostHog, then save them to das…”
   • [claimed-docs] “Roll out by person property, cohort, or group, then see the session replays, events, and exceptions from the people who got the flag”
   • [claimed-docs] “You can use SQL (Structured Query Language) throughout PostHog to manage, query, and modify data.”
   • [claimed-docs] “SQL insights enable you to directly query your data with SQL commands like SELECT, FROM, JOIN, WHERE, GROUP BY”

   proof ↗⚑ flag

   Mixpanel✓fullclaimed8/10

   Mixpanel's Retention report explicitly assesses user engagement over time, and properties/cohorts allow filtering and grouping of users for behavioral segmentation, directly matching the story. Documentation confirms retention analysis, cohort creation via properties, and query/export APIs that support slicing by cohort. Missing for 10: independent hands-on evidence of retention-by-cohort slicing in practice and no community corroboration specifically on retention feature usability.

   • [claimed-docs] “The Retention report in Mixpanel is designed to assess user engagement over a specified period of time.”
   • [claimed-docs] “Properties allow you to create groups of users (aka cohorts) and also enable you to filter for certain events or users.”
   • [claimed-docs] “The Mixpanel Query API endpoints can: Filter or break down Events * Return user profiles. * Return an activity/event stream for a user. * Re…”
   • [claimed-docs] “The Query API allows you to obtain the same calculated and formatted results that you see from a report in our web app.”

   proof ↗⚑ flag

4. product-managerExplore the paths users actually take before or after a key event

   weight 2 · round drawn
   PostHog✓fullclaimed7/10

   PostHog explicitly documents 'paths' insights alongside trends, funnels, retention, stickiness, and lifecycle built on captured events (posthog-docs-15, posthog-docs-25), directly matching the path-analysis story. Missing for 10: no dedicated docs excerpt detailing path-specific configuration (e.g., start/end event anchoring, wildcard grouping) or independent/hands-on validation of the paths feature specifically.

   • [claimed-docs] “Build trends, funnels, retention, paths, stickiness, and lifecycle insights on the events you already send to PostHog”
   • [claimed-docs] “Build trends, funnels, retention, paths, stickiness, and lifecycle insights on the events you already send to PostHog, then save them to das…”
   • [claimed-docs] “Custom events you capture from any SDK, each with the properties you attach to them.”

   proof ↗⚑ flag

   Mixpanel✓fullcommunity7/10

   Mixpanel's 'Flows' report is explicitly designed to visualize the paths users take before/after a given event, and this is corroborated by independent hands-on community praise ('gets to the heart of most of the information you need on user flows') as well as the docs' framing around discovering drop-off points and calibrating user journeys. However, community feedback also flags real friction (hover-based UI hard to use, requests to unify page-view vs custom events) and no probe confirms path exploration works flawlessly today. missing for 10: recent independent corroboration of UI usability, explicit documentation of 'before event' path exploration (most docs emphasize funnels/retention/after-event flows).

   • [claimed-docs] “Discover critical drop-off points and calibrate user journeys to optimize every touchpoint for maximum engagement, retention, and growth.”
   • [claimed-docs] “you can query events, funnels, flows, retention, session replays, and more using natural language — no dashboard navigation required.”
   • [community] “On Mixpanel's Flow feature: 'This is the coolest thing I have seen in Analytics for a long time. This really gets to the heart of most of th…”
   • [community] “Paying Mixpanel customer feedback on Flow: 'My biggest request is the same as the biggest request for the rest of Mixpanel: remove the wall …”
   • [community] “Usability complaints about Flow's hover-based UI: 'Requires too much effort (rolling over hover states) to dig out information' and is 'Very…”

   proof ↗⚑ flag

Self serve insights

1. founderAnswer "which activation step loses users?" myself with funnels and drop-off analysis, without an analyst

   weight 3 · round drawn
   PostHog✓fullclaimed8/10

   PostHog's product analytics explicitly builds funnels, retention, trends, paths, stickiness, and lifecycle insights on captured events, with dashboards and alerts, all self-serve without requiring an analyst (posthog-docs-15, posthog-docs-25, posthog-docs-1). This directly supports self-service drop-off/activation analysis via funnels. missing for 10: no independent/hands-on evidence of a founder successfully using funnels to diagnose activation drop-off, and no case study demonstrating non-analyst usability in practice.

   • [claimed-docs] “Build trends, funnels, retention, paths, stickiness, and lifecycle insights on the events you already send to PostHog”
   • [claimed-docs] “Build trends, funnels, retention, paths, stickiness, and lifecycle insights on the events you already send to PostHog, then save them to das…”
   • [claimed-docs] “Custom events you capture from any SDK, each with the properties you attach to them.”
   • [claimed-docs] “Build insights, assemble dashboards, share them, and set alerts on the numbers that matter.”

   proof ↗⚑ flag

   Mixpanel✓fullcommunity8/10

   Mixpanel's core product is self-serve funnels ('measure conversions through a series of events') and drop-off analysis, with docs explicitly framing this as a no-code, click-based, self-serve capability ('interactive reports that let you query and visualize the results with just a few clicks', 'Discover critical drop-off points'), plus Session Replay to explore 'why customers drop off' — all designed for non-analyst self-service. Community evidence corroborates the self-serve UI power ('PMs spend hours in mixpanel quickly generating custom report after custom report') though also notes UI friction in an older Flow feature. Missing for 10: no first-party founder/non-technical-user case study specifically validating a founder (not a PM/analyst) independently diagnosing activation drop-off, and one community note flags usability complaints on a related flow visualization.

   • [claimed-docs] “Measure conversions through a series of events”
   • [claimed-docs] “Discover critical drop-off points and calibrate user journeys to optimize every touchpoint for maximum engagement, retention, and growth.”
   • [claimed-docs] “It enables you to track how users engage with your product and analyze this data with interactive reports that let you query and visualize t…”
   • [claimed-docs] “Now, with Mixpanel Session Replay, you can quickly follow up by exploring **why** customers drop off.”
   • [community] “'Getting a front end anywhere near as slick and powerful as Mixpanel's is an even bigger challenge than the data pipeline... I've seen PMs s…”
   • [community] “On Mixpanel's Flow feature: 'This is the coolest thing I have seen in Analytics for a long time. This really gets to the heart of most of th…”
   • [community] “Usability complaints about Flow's hover-based UI: 'Requires too much effort (rolling over hover states) to dig out information' and is 'Very…”

   proof ↗⚑ flag

Destinations

1. developerStream events onward to third-party destinations (CRM, ad platforms, webhooks) from the analytics platform

   weight 2 · round to PostHog
   PostHog✓fullclaimed8/10

   PostHog's CDP explicitly supports streaming events to third-party destinations including CRMs (Salesforce, HubSpot), ad platforms, webhooks, and SaaS tools (Braze, Customer.io, Slack, PagerDuty), with real-time or scheduled delivery via configurable Hog functions. missing for 10: independent/hands-on confirmation of destination reliability or throughput, and a full list of supported destinations beyond the named examples.

   • [claimed-docs] “It pulls data in from the tools you already use, lets you filter and reshape events as they're ingested, and sends the result wherever it ne…”
   • [claimed-docs] “It pulls data in from the tools you already use, lets you filter and reshape events as they're ingested, and sends the result wherever it ne…”
   • [claimed-docs] “Pipelines are built from Hog functions, which you can configure from a template, write yourself, or have PostHog AI write for you.”
   • [claimed-docs] “Sync user properties to CRM systems like Salesforce or HubSpot”
   • [claimed-docs] “Slack: Send events to Slack channels SaaS tools: Send events to SaaS tools like Braze, Customer.io, and more”
   • [claimed-docs] “Send alerts to Slack, PagerDuty, or monitoring tools”

   proof ↗⚑ flag

   Mixpanel—none0/10

   Evidence shows only inbound warehouse syncing (Snowflake/BigQuery/etc. into Mixpanel) and pull-based export APIs (Raw Event Export, Query API, CSV), none of which describe pushing/streaming events out to CRM systems, ad platforms, or webhooks. The partner integrations directory lists third-party platforms (mParticle, Census, Vendo) but their descriptions frame them as bringing data INTO Mixpanel, not Mixpanel streaming events out to them.

   • [claimed-docs] “you can sync data from data warehouses like Snowflake, BigQuery, Databricks, Redshift, and Postgres to Mixpanel”
   • [claimed-docs] “To export raw event data, use the Raw Event Export API. The events payload is returned as JSONL”
   • [claimed-docs] “With Warehouse Connectors, you can sync data from data warehouses like Snowflake, BigQuery, Databricks, Redshift, and Postgres to Mixpanel.”
   • [claimed-docs] “You can export data from your project using a variety of methods, including lightweight CSVs generated from the UI, large volume raw data ex…”
   • [claimed-docs] “mParticle is an AI Customer Data Platform that powers the entire marketing stack with real-time customer data. Companies like NBCUniversal, …”
   • [claimed-docs] “Census is the #1 Data Activation platform that unites revenue teams around a single source of truth in the cloud data warehouse. Our no-code…”
   • [claimed-docs] “Vendo is a marketing data platform for growth teams. It connects the tools you already use — Shopify, Google Ads, Meta Ads, TikTok, Klaviyo,…”

   proof ↗⚑ flag

1. ai-native userDo everything through the API that I can do in the UI

   weight 2 · round to PostHog
   PostHog✓fullprobed7/10

   PostHog exposes a full OpenAPI-documented REST API (posthog-probe-3, posthog-docs-11), an MCP server enabling agents to run queries, ship flags, and inspect errors (posthog-docs-2, -12, -19, -24, probe-4), a CLI for terminal/CI use (posthog-docs-20, probe-5), and SQL access throughout the product (posthog-docs-8, -17), collectively covering analytics, flags, experiments, error tracking, and data warehouse operations programmatically. missing for 10: an explicit vendor statement guaranteeing 1:1 API/UI parity, and independent verification that every UI workflow (e.g. survey building, dashboard sharing) is fully replicable via API without gaps.

   • [claimed-docs] “Run trends, funnels, retention, and SQL queries from any MCP client or AI editor.”
   • [claimed-docs] “capture events from any language that can send HTTP requests”
   • [claimed-docs] “"Show me the full stack trace for the most recent crash, then propose a fix." – Pulls the stack trace, error message, and metadata so the ag…”
   • [claimed-docs] “with just plain text questions your agents can ship a feature flag from a prompt, dig into a stack trace without leaving your editor, run a …”
   • [claimed-docs] “The PostHog CLI lets you use PostHog from your terminal, your coding agents, local scripts, and CI/CD pipelines.”
   • [claimed-docs] “You can use SQL (Structured Query Language) throughout PostHog to manage, query, and modify data.”
   • [probe] “PROBE openapi: HTTP 200 at https://posthog.com/openapi.json — contains "openapi" key”
   • [probe] “official MCP server documented at https://posthog.com/docs/model-context-protocol”
   • [probe] “official CLI documented at https://posthog.com/docs/cli”

   proof ↗⚑ flag

   Mixpanel~partialclaimed6/10

   Mixpanel exposes a broad API surface—Query API (mirrors report results), Raw Export/Engage APIs, Feature Flags API, JQL for custom queries, and an MCP server that can create dashboards and query via natural language—giving strong programmatic coverage of most analytics workflows (mixpanel-docs-39, mixpanel-docs-43, mixpanel-docs-36, mixpanel-docs-37, mixpanel-docs-10, mixpanel-docs-11). However, some UI-only capabilities are explicitly documented, e.g. Session Replay sampling is managed 'directly from the Mixpanel UI — no code changes or redeployment required' (mixpanel-docs-5, mixpanel-docs-29), indicating a gap rather than full UI/API parity. Missing for 10: explicit confirmation that dashboard configuration, session-replay settings, and all admin/UI-only controls are fully API-accessible, and independent verification of complete parity beyond vendor docs.

   • [claimed-docs] “The Mixpanel Query API endpoints can: Filter or break down Events * Return user profiles. * Return an activity/event stream for a user. * Re…”
   • [claimed-docs] “The Query API allows you to obtain the same calculated and formatted results that you see from a report in our web app.”
   • [claimed-docs] “Use the Feature Flags API to assign your users to variants for experiments, rollout, and releases.”
   • [claimed-docs] “Create custom queries using Javascript code”
   • [claimed-docs] “Mixpanel provides a hosted Model Context Protocol (MCP) server that gives AI assistants direct access to your Mixpanel data.”
   • [claimed-docs] “Create-Dashboard | Create a new dashboard with text cards and reports”
   • [claimed-docs] “teams can manage Session Replay sampling directly from the Mixpanel UI — no code changes or redeployment required.”
   • [claimed-docs] “teams can manage Session Replay sampling directly from the Mixpanel UI — no code changes or redeployment required”

   proof ↗⚑ flag

2. ai-native userExport all of my data in open formats and leave

   weight 3 · round to Mixpanel
   PostHog~partialprobed6/10

   PostHog is open-source (MIT-licensed self-host Docker Compose) and exposes data via SQL, an HTTP API, and data-warehouse sync, which together give strong data portability and self-hosting options implying users can retain full control of their data. However, there is no explicit documented 'export all your data' feature or open-format bulk export tool, and no independent evidence of someone successfully exporting/migrating a full production dataset. Missing for 10: explicit bulk-export/download-all-data documentation, confirmation of open standard export formats (e.g., CSV/Parquet dumps), and independent hands-on evidence of a full data migration/exit.

   • [claimed-docs] “PostHog is open-source and freely available for anyone to host themselves. We offer a free Docker Compose deployment under an MIT license.”
   • [claimed-docs] “We offer a free Docker Compose deployment under an MIT license.”
   • [claimed-docs] “You can use SQL (Structured Query Language) throughout PostHog to manage, query, and modify data.”
   • [claimed-docs] “SQL insights enable you to directly query your data with SQL commands like SELECT, FROM, JOIN, WHERE, GROUP BY”
   • [claimed-docs] “Sync Stripe, Postgres, Salesforce, HubSpot, and dozens of other sources, then query them alongside your events, persons, and sessions in one…”
   • [claimed-docs] “capture events from any language that can send HTTP requests”
   • [probe] “PROBE openapi: HTTP 200 at https://posthog.com/openapi.json — contains "openapi" key”

   proof ↗⚑ flag

   Mixpanel✓fullcommunity8/10

   Mixpanel documents multiple open-format export paths — Raw Event Export API returning JSONL (mixpanel-docs-7,9,18,31), CSV export from the UI (mixpanel-docs-25,30), Engage API for user profiles (mixpanel-docs-8), and Query API for aggregated data (mixpanel-docs-39,43) — enabling a user to pull out event, profile, and query data in standard formats. Community commentary even praises Mixpanel's export policy as 'great' when discussing data ownership (mixpanel-comm-13), corroborating that full data export works in practice. Missing for 10: no explicit statement of exporting *all* config/dashboard/schema metadata, and no independent bulk-export walkthrough beyond forum anecdote.

   • [claimed-docs] “To export raw event data, use the Raw Event Export API. The events payload is returned as JSONL”
   • [claimed-docs] “The raw export API allows you to download your event data as it is received and stored within Mixpanel, complete with all event properties”
   • [claimed-docs] “Every data point sent to Mixpanel is stored as JSON in our data store.”
   • [claimed-docs] “You can export data from your project using a variety of methods, including lightweight CSVs generated from the UI, large volume raw data ex…”
   • [claimed-docs] “You may export small batches of events as a CSV for use in quick analysis by navigating to the Events tab”
   • [claimed-docs] “To export User Profile data, use the Engage API.”
   • [claimed-docs] “The Mixpanel Query API endpoints can: Filter or break down Events * Return user profiles. * Return an activity/event stream for a user. * Re…”
   • [community] “Complaint about vendor lock-in: 'The tool is lovely, their export policy is great, but there's something about actually owning your data' — …”

   proof ↗⚑ flag

3. ai-native userRead the product's source under an open license

   weight 2 · round to PostHog
   PostHog✓fullclaimed9/10

   PostHog explicitly documents its open-source nature and MIT-licensed self-host deployment (Docker Compose), with the actual source hosted publicly on GitHub, satisfying the ability to read the product's source under an open license. Missing for 10: no explicit confirmation of the license covering the entire codebase (vs just the self-host component) or independent audit of license terms.

   • [claimed-docs] “We offer a free Docker Compose deployment under an MIT license.”
   • [claimed-docs] “PostHog is open-source and freely available for anyone to host themselves. We offer a free Docker Compose deployment under an MIT license.”

   proof ↗⚑ flag

   Mixpanel—none0/10

   Mixpanel is closed-source SaaS; no evidence of any open-licensed source code repository or license grant is present in the evidence pack, only product docs, integrations, and community commentary unrelated to source licensing.

   ⚑ flag

4. ai-native userSelf-host the core product

   weight 3 · round to PostHog
   PostHog!disputedcontradicted5/10

   PostHog's docs explicitly state the core product is open-source and offers a free MIT-licensed Docker Compose self-host deployment (posthog-docs-14/21), which is a clear ai-native/openness claim. However, a hands-on community report describes the production Dockerfile/posthog:latest image as 'busted' and failing to create a database, requiring significant debugging effort — a concrete contradiction of the self-host claim working out of the box. Missing for 10: independent confirmation the current self-host path works reliably, resolution or acknowledgment of the reported Docker issue, and details on feature parity between self-hosted and cloud versions.

   • [claimed-docs] “We offer a free Docker Compose deployment under an MIT license.”
   • [claimed-docs] “PostHog is open-source and freely available for anyone to host themselves. We offer a free Docker Compose deployment under an MIT license.”
   • [community] “You should mention on the README that the production dockerfile (and posthog/posthog:latest) are busted, they do not create any database. Sp…”

   proof ↗⚑ flag

   Mixpanel—none0/10

   Mixpanel is offered exclusively as a hosted SaaS product; no evidence pack item mentions a self-hosted or on-premise deployment option, source code availability, or Docker/Kubernetes install path. Absence of evidence for this applicable capability yields a 'none' verdict.

   ⚑ flag

Privacy first analytics

1. founderTrack product usage without third-party cookies so I can minimize or avoid cookie consent banners

   weight 2 · round drawn
   PostHog—none0/10

   The evidence pack contains no mention of cookie-based vs cookieless tracking, consent banners, or privacy-compliant capture configuration—only general analytics, session replay, and SDK/API capture details. Without any documentation addressing cookieless tracking or consent-banner avoidance, this axis has no supporting evidence.

   ⚑ flag

   Mixpanel—none0/10

   The evidence pack contains no documentation on cookieless tracking, first-party identity resolution, or reducing/avoiding cookie consent banners — Mixpanel's docs focus on events, funnels, retention, session replay, exports, and MCP, none of which address cookie usage or consent-banner avoidance. Missing for 10: any statement on cookieless tracking methods, first-party data collection design, or consent-banner impact.

   ⚑ flag

1. ai-native userChoose where my data is stored (region/residency)

   weight 2 · round to Mixpanel
   PostHog—none0/10

   The evidence pack has self-hosting/open-source docs (posthog-docs-14, posthog-docs-21) which relate to deployment flexibility but no mention of data residency/region selection (e.g., EU vs US cloud region choice) for PostHog Cloud. Missing for 10: explicit region/residency selection options, data storage location controls, compliance documentation on where data resides.

   • [claimed-docs] “We offer a free Docker Compose deployment under an MIT license.”
   • [claimed-docs] “PostHog is open-source and freely available for anyone to host themselves. We offer a free Docker Compose deployment under an MIT license.”

   proof ↗⚑ flag

   Mixpanel✓fullclaimed7/10

   Mixpanel documents an EU Data Residency Program allowing customers to process and store personal data in Europe, directly addressing regional data residency choice [mixpanel-docs-38], backed by GDPR-oriented deletion/retrieval APIs [mixpanel-docs-12]. Missing for 10: no detail on which regions beyond EU are supported, no independent/hands-on confirmation of the residency mechanics or migration process, and no mention of per-project region selection UI.

   • [claimed-docs] “Mixpanel provides you with the option to process and store your customers' personal data in Europe via our EU Data Residency Program.”
   • [claimed-docs] “Mixpanel deletion and retrieval APIs are in place to help Mixpanel implementations meet the requirements outlined by the General Data Protec…”

   proof ↗⚑ flag

2. ai-native userPrevent my data from being used to train AI models

   weight 3 · round drawn
   PostHog—none0/10

   No evidence in the pack addresses AI-training data opt-out or any privacy control preventing customer data from being used to train AI models; nothing in the docs or community items mentions this capability.

   ⚑ flag

   Mixpanel—none0/10

   No evidence in the pack addresses AI-training opt-out or any policy preventing customer data from being used to train AI models; privacy docs only mention GDPR deletion/retrieval APIs and EU data residency, not AI-training exclusion.

   • [claimed-docs] “Mixpanel deletion and retrieval APIs are in place to help Mixpanel implementations meet the requirements outlined by the General Data Protec…”
   • [claimed-docs] “Mixpanel provides you with the option to process and store your customers' personal data in Europe via our EU Data Residency Program.”

   proof ↗⚑ flag

3. ai-native userControl data retention and deletion

   weight 2 · round to Mixpanel
   PostHog—none0/10

   The evidence pack contains no mention of data retention settings, deletion APIs, GDPR-style data deletion, or retention policy controls anywhere in the docs, CLI, API, or CDP descriptions. This is a fair and expected axis for an analytics/data platform handling user event data, so absence of evidence means the story is unmet rather than inapplicable.

   ⚑ flag

   Mixpanel✓fullclaimed7/10

   Mixpanel documents explicit deletion and retrieval APIs built to meet GDPR requirements, plus an EU Data Residency Program for regional data storage/processing, giving users concrete levers to control retention and deletion of their data ([mixpanel-docs-12], [mixpanel-docs-38]). Missing for 10: independent/hands-on verification of the deletion APIs actually working, and documentation of configurable retention windows or automated data-expiry settings beyond GDPR compliance language.

   • [claimed-docs] “Mixpanel deletion and retrieval APIs are in place to help Mixpanel implementations meet the requirements outlined by the General Data Protec…”
   • [claimed-docs] “Mixpanel provides you with the option to process and store your customers' personal data in Europe via our EU Data Residency Program.”

   proof ↗⚑ flag

4. ai-native userOpt out of telemetry and usage tracking

   weight 2 · round drawn
   PostHog—none0/10

   The evidence pack covers PostHog's product analytics, session replay, feature flags, MCP/CLI integrations, and self-hosting, but contains no mention of an opt-out setting for PostHog's own telemetry/usage tracking (e.g., CLI or SDK phone-home data collection). No documentation or community evidence addresses this capability.

   ⚑ flag

   Mixpanel—none0/10

   Evidence pack shows Mixpanel's GDPR deletion/retrieval APIs and EU data residency options for customer data stored in Mixpanel, but there is no mention of any mechanism for an AI-native user (e.g., of the MCP server or AI features) to opt out of telemetry/usage tracking collected about their own interactions with Mixpanel's tools.

   • [claimed-docs] “Mixpanel deletion and retrieval APIs are in place to help Mixpanel implementations meet the requirements outlined by the General Data Protec…”
   • [claimed-docs] “Mixpanel provides you with the option to process and store your customers' personal data in Europe via our EU Data Residency Program.”

   proof ↗⚑ flag

Replay

1. product-managerJump from a funnel drop-off or an error event straight to session replays of the affected users

   weight 2 · round to Mixpanel
   PostHog~partialclaimed6/10

   PostHog documents linking session replays directly to flag rollouts and to exceptions/errors ('see the session replays, events, and exceptions from the people who got the flag'), and session replay's synced DevTools panel shows errors at the moment they happened, supporting a PM-style jump from error/event to replay. However, the evidence never explicitly documents jumping from a funnel drop-off step to session replays, only from feature-flag cohorts/exceptions. missing for 10: explicit funnel-to-replay linking documentation, independent/hands-on confirmation of this specific workflow.

   • [claimed-docs] “Roll out by person property, cohort, or group, then see the session replays, events, and exceptions from the people who got the flag”
   • [claimed-docs] “Session replay records what real users do in your product and plays it back like a DVR, with a synced DevTools panel showing console logs, n…”
   • [claimed-docs] “with a synced DevTools panel showing console logs, network requests, and errors at the exact moment they happened”
   • [claimed-docs] “plays it back like a DVR, with a synced DevTools panel showing console logs, network requests, and errors at the exact moment they happened”

   proof ↗⚑ flag

   Mixpanel~partialclaimed7/10

   Mixpanel explicitly documents Session Replay as a way to 'follow up by exploring why customers drop off' from funnels, and playlists can be saved 'based on events or properties' for review, supporting drop-off-to-replay workflows. However, there is no explicit documentation of jumping from an error event specifically into replays, nor independent/hands-on confirmation of this workflow. Missing for 10: explicit error-event-to-replay linkage, independent user corroboration of the workflow.

   • [claimed-docs] “Now, with Mixpanel Session Replay, you can quickly follow up by exploring why customers drop off.”
   • [claimed-docs] “Now, with Mixpanel Session Replay, you can quickly follow up by exploring **why** customers drop off.”
   • [claimed-docs] “Save playlists of replays based on events or properties, and quickly review with AI summaries, making it easy to revisit groups of replays t…”
   • [claimed-docs] “Discover critical drop-off points and calibrate user journeys to optimize every touchpoint for maximum engagement, retention, and growth.”
   • [claimed-docs] “Use AI to spot trends across hundreds of replays and heatmaps, then zoom in yourself to get to the “why” behind every change.”

   proof ↗⚑ flag

2. product-managerWatch recordings of real user sessions with sensitive input masked by default

   weight 3 · round to Mixpanel
   PostHog—none0/10

   Evidence confirms PostHog's session replay lets PMs watch DVR-style recordings with synced DevTools panel (posthog-docs-26, posthog-docs-29, posthog-gh-8), but nothing in the pack mentions input masking, privacy controls, or sensitive-data redaction being applied by default. missing for 10: any documentation of input masking/privacy defaults, evidence of what is masked or how, and confirmation it's on by default rather than opt-in.

   • [claimed-docs] “Session replay records what real users do in your product and plays it back like a DVR, with a synced DevTools panel showing console logs, n…”
   • [claimed-docs] “plays it back like a DVR, with a synced DevTools panel showing console logs, network requests, and errors at the exact moment they happened”
   • [github] “Session replays: Watch real user sessions of interactions with your website or mobile app to diagnose issues and understand user behavior.”

   proof ↗⚑ flag

   Mixpanel~partialclaimed4/10

   Mixpanel's Session Replay lets PMs watch real user session recordings and even use AI to summarize/playlist replays (mixpanel-docs-4, mixpanel-docs-27, mixpanel-docs-42), but the evidence pack contains no documentation confirming that sensitive input fields (e.g., passwords, form data) are masked by default in these recordings. Privacy docs cover GDPR/EU data residency (mixpanel-docs-12, mixpanel-docs-38) but not replay-specific input masking. Missing for 10: explicit documentation of default input/text masking or redaction controls within Session Replay, and any independent confirmation that masking works as claimed.

   • [claimed-docs] “Now, with Mixpanel Session Replay, you can quickly follow up by exploring why customers drop off.”
   • [claimed-docs] “Save playlists of replays based on events or properties, and quickly review with AI summaries, making it easy to revisit groups of replays t…”
   • [claimed-docs] “Mixpanel Session Replay provides the fastest way to gain a comprehensive understanding of your customers and make informed product decisions”
   • [claimed-docs] “Mixpanel deletion and retrieval APIs are in place to help Mixpanel implementations meet the requirements outlined by the General Data Protec…”
   • [claimed-docs] “Mixpanel provides you with the option to process and store your customers' personal data in Europe via our EU Data Residency Program.”

   proof ↗⚑ flag

Ai analytics ops

1. ai-native userHave an agent answer growth questions from live product data by running queries through the API or MCP server

   weight 3 · round to Mixpanel
   PostHog!disputedcontradicted5/10

   PostHog documents a first-party MCP server and CLI that let agents run HogQL/SQL queries, ship flags, and answer product questions from live data (posthog-docs-2, posthog-docs-19, posthog-docs-24, posthog-probe-4, posthog-docs-8/17 for SQL, posthog-docs-9 for warehouse queries). However, a hands-on community report describes the LLM/query integration failing to create a cohort via a SQL query and causing severe UI slowdowns, directly contradicting the smooth 'agent answers growth questions via query' promise (posthog-comm-8). missing for 10: independent corroboration that MCP-driven SQL queries reliably work at scale, and resolution of the reported failure/performance bug.

   • [claimed-docs] “Run trends, funnels, retention, and SQL queries from any MCP client or AI editor.”
   • [claimed-docs] “with just plain text questions your agents can ship a feature flag from a prompt, dig into a stack trace without leaving your editor, run a …”
   • [claimed-docs] “your agents can ship a feature flag from a prompt, dig into a stack trace without leaving your editor, run a HogQL query through Claude”
   • [probe] “official MCP server documented at https://posthog.com/docs/model-context-protocol”
   • [claimed-docs] “Sync Stripe, Postgres, Salesforce, HubSpot, and dozens of other sources, then query them alongside your events, persons, and sessions in one…”
   • [claimed-docs] “SQL insights enable you to directly query your data with SQL commands like SELECT, FROM, JOIN, WHERE, GROUP BY”
   • [community] “I really wish you guys would do something about the llm integration, I tried using it two days ago to create a cohort of users using a sql q…”

   proof ↗⚑ flag

   Mixpanel✓fullprobed8/10

   Mixpanel documents a hosted MCP server that lets AI assistants query events, funnels, flows, retention, and session replays via natural language (no dashboard navigation), plus a Query API for filtering/breaking down events, cohorts, and profiles, and a raw export/JQL API for deeper queries—together covering agent-driven growth Q&A over live product data. Missing for 10: independent/hands-on verification of MCP server reliability or accuracy, and no evidence of complex ad-hoc SQL-like warehouse querying via MCP beyond the documented report types.

   • [claimed-docs] “Mixpanel provides a hosted Model Context Protocol (MCP) server that gives AI assistants direct access to your Mixpanel data.”
   • [claimed-docs] “you can query events, funnels, flows, retention, session replays, and more using natural language — no dashboard navigation required.”
   • [claimed-docs] “Once connected, you can query events, funnels, flows, retention, session replays, and more using natural language — no dashboard navigation …”
   • [claimed-docs] “you can query events, funnels, flows, retention, session replays, and more using natural language — no dashboard navigation required”
   • [claimed-docs] “The Mixpanel Query API endpoints can: Filter or break down Events * Return user profiles. * Return an activity/event stream for a user. * Re…”
   • [claimed-docs] “The Query API allows you to obtain the same calculated and formatted results that you see from a report in our web app.”
   • [probe] “official MCP server documented at https://docs.mixpanel.com/docs/mcp”

   proof ↗⚑ flag

Raw access

1. developerRun analytical queries and pull raw event data back through a documented query/export API

   weight 3 · round to Mixpanel
   PostHog✓fullprobed8/10

   PostHog documents a full HogQL SQL query interface, a data warehouse that lets you query events/persons alongside external sources, a public OpenAPI spec (confirmed live at /openapi.json), and API-based event capture/export, plus MCP support for running queries programmatically. This directly satisfies running analytical queries and pulling raw event data via a documented API. missing for 10: independent/hands-on confirmation of raw event export via REST API specifically (evidence is mostly first-party docs), and no third-party validation of API rate limits/completeness for large-scale export.

   • [claimed-docs] “You can use SQL (Structured Query Language) throughout PostHog to manage, query, and modify data.”
   • [claimed-docs] “SQL insights enable you to directly query your data with SQL commands like SELECT, FROM, JOIN, WHERE, GROUP BY”
   • [claimed-docs] “Sync Stripe, Postgres, Salesforce, HubSpot, and dozens of other sources, then query them alongside your events, persons, and sessions in one…”
   • [claimed-docs] “capture events from any language that can send HTTP requests”
   • [probe] “PROBE openapi: HTTP 200 at https://posthog.com/openapi.json — contains "openapi" key”
   • [claimed-docs] “your agents can ship a feature flag from a prompt, dig into a stack trace without leaving your editor, run a HogQL query through Claude”

   proof ↗⚑ flag

   Mixpanel✓fullclaimed9/10

   Mixpanel documents a Query API for analytical queries (filter/breakdown events, cohorts, activity streams) and a separate Raw Event Export API returning full JSONL event data, plus JQL for custom JS-based queries — directly matching the query/export API story. Missing for 10: no independent/hands-on developer corroboration of API reliability or rate limits beyond first-party docs.

   • [claimed-docs] “The Mixpanel Query API endpoints can: Filter or break down Events * Return user profiles. * Return an activity/event stream for a user. * Re…”
   • [claimed-docs] “The Query API allows you to obtain the same calculated and formatted results that you see from a report in our web app.”
   • [claimed-docs] “To export raw event data, use the Raw Event Export API. The events payload is returned as JSONL”
   • [claimed-docs] “The raw export API allows you to download your event data as it is received and stored within Mixpanel, complete with all event properties”
   • [claimed-docs] “To export raw event data, use the Raw Event Export API... This method is recommended for high volume exports.”
   • [claimed-docs] “Create custom queries using Javascript code”
   • [claimed-docs] “The JQL JavaScript code must define a main() function. This JavaScript will be compiled, and the main() function will be executed by the que…”

   proof ↗⚑ flag

2. developerQuery my raw events with SQL (or an SQL-equivalent query language) inside the platform

   weight 3 · round to PostHog
   PostHog✓fullclaimed9/10

   PostHog documents HogQL, a SQL-like query language, that lets developers directly query raw events with SELECT/FROM/JOIN/WHERE/GROUP BY, and also join across data warehouse sources alongside events/persons/sessions in one SQL query, plus MCP/agent access to run HogQL queries. This is well-documented first-party capability with clear syntax examples. missing for 10: independent hands-on verification of SQL query correctness/performance at scale (one community comment notes LLM confusion with cohorts, not the SQL feature itself).

   • [claimed-docs] “You can use SQL (Structured Query Language) throughout PostHog to manage, query, and modify data.”
   • [claimed-docs] “SQL insights enable you to directly query your data with SQL commands like SELECT, FROM, JOIN, WHERE, GROUP BY”
   • [claimed-docs] “Sync Stripe, Postgres, Salesforce, HubSpot, and dozens of other sources, then query them alongside your events, persons, and sessions in one…”
   • [claimed-docs] “with just plain text questions your agents can ship a feature flag from a prompt, dig into a stack trace without leaving your editor, run a …”
   • [claimed-docs] “your agents can ship a feature flag from a prompt, dig into a stack trace without leaving your editor, run a HogQL query through Claude”

   proof ↗⚑ flag

   Mixpanel~partialclaimed5/10

   Mixpanel doesn't offer SQL, but its JQL (JavaScript Query Language) lets developers write custom code queries against raw event data directly in the platform, and the Query API can filter/break down events and return activity streams — a functional but not syntactic SQL-equivalent. missing for 10: no actual SQL support, no evidence JQL/Query API is positioned as a full SQL-equivalent, and no independent/hands-on corroboration of its expressiveness for raw-event querying.

   • [claimed-docs] “Create custom queries using Javascript code”
   • [claimed-docs] “The JQL JavaScript code must define a main() function. This JavaScript will be compiled, and the main() function will be executed by the que…”
   • [claimed-docs] “The Mixpanel Query API endpoints can: Filter or break down Events * Return user profiles. * Return an activity/event stream for a user. * Re…”
   • [claimed-docs] “The Query API allows you to obtain the same calculated and formatted results that you see from a report in our web app.”

   proof ↗⚑ flag

Warehouse

1. developerSync events between the platform and my own data warehouse (query warehouse tables in-product or export events continuously to it)

   weight 2 · round to PostHog
   PostHog✓fullclaimed8/10

   PostHog's data warehouse feature lets developers sync external sources (Stripe, Postgres, Salesforce, HubSpot, etc.) and query them alongside native events/persons/sessions in one SQL query, and the CDP allows continuous export of events to external destinations in realtime or on a schedule, satisfying both directions of the sync story. Missing for 10: independent/hands-on verification of the warehouse sync reliability or performance at scale, and clearer documentation of specific export destinations (data warehouses) beyond CRM/Slack examples.

   • [claimed-docs] “Sync Stripe, Postgres, Salesforce, HubSpot, and dozens of other sources, then query them alongside your events, persons, and sessions in one…”
   • [claimed-docs] “You can use SQL (Structured Query Language) throughout PostHog to manage, query, and modify data.”
   • [claimed-docs] “SQL insights enable you to directly query your data with SQL commands like SELECT, FROM, JOIN, WHERE, GROUP BY”
   • [claimed-docs] “It pulls data in from the tools you already use, lets you filter and reshape events as they're ingested, and sends the result wherever it ne…”
   • [claimed-docs] “Pipelines are built from Hog functions, which you can configure from a template, write yourself, or have PostHog AI write for you.”
   • [claimed-docs] “Combine everything in PostHog's context warehouse so that you, your agents, and your dashboard can query it directly.”

   proof ↗⚑ flag

   Mixpanel~partialclaimed7/10

   Mixpanel's Warehouse Connectors let you sync data from Snowflake, BigQuery, Databricks, Redshift, and Postgres into Mixpanel, and the Raw Event Export API (JSONL, recommended for high-volume) lets you pull events out for warehouse loading, satisfying both directions of the story at a basic level. However, continuous/automated export to a warehouse is not natively built-in as a managed reverse-sync pipeline—Mixpanel's own partner directory lists third-party Reverse ETL tools like Census for pushing Mixpanel data to warehouses, implying continuous outbound sync relies on partner tooling rather than a first-party feature. Missing for 10: evidence of a native continuous/scheduled export-to-warehouse pipeline (not just an on-demand API), and independent/hands-on confirmation that Warehouse Connectors and export API work reliably at scale.

   • [claimed-docs] “you can sync data from data warehouses like Snowflake, BigQuery, Databricks, Redshift, and Postgres to Mixpanel”
   • [claimed-docs] “With Warehouse Connectors, you can sync data from data warehouses like Snowflake, BigQuery, Databricks, Redshift, and Postgres to Mixpanel.”
   • [claimed-docs] “you can sync data from data warehouses like Snowflake, BigQuery, Databricks, Redshift, and Postgres to Mixpanel.”
   • [claimed-docs] “To export raw event data, use the Raw Event Export API. The events payload is returned as JSONL”
   • [claimed-docs] “The raw export API allows you to download your event data as it is received and stored within Mixpanel, complete with all event properties”
   • [claimed-docs] “To export raw event data, use the Raw Event Export API... This method is recommended for high volume exports.”
   • [claimed-docs] “You can export data from your project using a variety of methods, including lightweight CSVs generated from the UI, large volume raw data ex…”
   • [claimed-docs] “Census is the #1 Data Activation platform that unites revenue teams around a single source of truth in the cloud data warehouse. Our no-code…”
   • [claimed-docs] “SET CHANGE\TRACKING = TRUE; # or ALTER VIEW SET CHANGE\TRACKING = TRUE; \\\ Mixpanel will create and manage the necessary STREAM objects in …”
   • [claimed-docs] “SET TBLPROPERTIES (delta.enableChangeDataFeed = true); \\\ \\Considerations when using Mirror with Databricks:\\ \ Mirror is not supported o…”

   proof ↗⚑ flag