PostHog October 2026 Updates: Marketing Analytics, Data Warehouse, AI & MCP

In October 2026, PostHog introduced updates to its marketing analytics tools, advertising platform integrations, Data Warehouse, and AI agents. The company expanded the data available for import, added tools for analyzing AI applications, and simplified connections to external AI assistants through the Model Context Protocol (MCP).

October 2026 Product Updates

October 1: Batch export improvements. PostHog added support for the {data_interval_start} and {data_interval_end} parameters in HogQL queries used for Batch Exports. These parameters automatically limit exports to the required period. For example, a daily export can include only data from the latest interval instead of rebuilding the full report. Users can also apply custom HogQL settings to individual queries without changing project-wide settings.

October 2: Permission management through MCP. AI agents connected to PostHog through the Model Context Protocol gained the ability to create and modify access rules for users, roles, and individual resources. Changes require confirmation, and the feature is available starting with the Boost plan. In addition, MCP connections using renewable authorization tokens can now remain active longer without requiring users to reconnect.

October 2: Expanded AI Observability. PostHog added support for custom OpenAI-compatible API endpoints for automated evaluation of language model responses (LLM Evals) and testing in Playground. This allows teams to connect compatible AI services, including external gateways and regional providers, rather than being limited to the default integrations.

October 5: Email broadcasts in PostHog Workflows. The new Broadcasts interface allows users to create one-time and scheduled email campaigns for selected audiences. Teams can compose a message, check personalization against a specific recipient's data, send a test email, configure sending speed, and track conversion goals. The feature extends PostHog beyond analytics by enabling communications based on user data.

October 8: Nineteen new MCP tools. PostHog expanded the actions available to external AI agents. The new tools can launch and monitor bulk evaluations of AI model responses, retrieve daily analytics summaries, analyze statistics for automated processes, and work with product research findings. More analytics operations can now be handled through a connected AI assistant without manually navigating the interface.

October 15: Automated detection of underused product features. Self-Driving gained a new analytics agent that identifies product features available to users but rarely used. The agent examines product events and, when connected, source code context. It helps distinguish features with little actual demand from features users struggle to discover or use.

October 19: Google Ads change history in Data Warehouse. The Google Ads integration gained the ability to import advertising account change events into a separate table with incremental updates. This makes it possible to analyze which campaign settings changed, when changes occurred, and who made them. The records can be compared with advertising performance data to investigate the reasons behind changes in campaign results.

October 19: New Data Warehouse sources. PostHog added integrations with Auth0, Modal, Neo4j, Vimeo, Arcade, Scrunch, and other services. Six previously experimental sources, including Outbrain, Braze, Squarespace, Drip, Recurly, and Ramp, reached general availability. Synchronization reliability was also improved: when an import is interrupted, the system preserves progress more safely and reduces the risk of incomplete data.

October 22: Simplified MCP setup. PostHog Wizard can now automatically discover local development tools with AI agents, including Claude Code and Codex, and set up their MCP connection to PostHog. Browser-based clients, such as ChatGPT and Claude, have a separate connection process. This reduces the manual steps required to configure integrations.

October 22: New AI Evaluations templates. AI Observability expanded its templates for evaluating language model responses. In addition to Pass/Fail checks, templates now support numerical scoring and response categorization. Users can inspect examples of evaluations before running them. The tools are designed to test AI application quality and detect regressions after model or prompt changes.

October 23: WebMCP analytics. PostHog's JavaScript SDK added support for capturing WebMCP tool-call events directly in the browser. Calls are associated with a user, session, and Session Replay. Additional information about the model and the agent's intent can be captured, while sensitive values are sanitized. The feature helps teams understand how AI agents interact with a web application, which tools they call, and where errors occur.

October 23: More data sources for Self-Driving AI. PostHog's automated analytics agents gained access to 38 external data sources, including GitHub, Linear, Zendesk, Sentry, and Jira. Connections are configured through the Self-Driving setup interface. As a result, agents can use not only product telemetry but also information about errors, support requests, and development tasks.

October 23: Gemini Interactions API observability. PostHog AI Observability began capturing calls to Google's Gemini Interactions API, including streaming responses, tool usage, resource consumption, and supported media outputs. It supports calls executed directly during a user request. Background interactions and some agent workflows are not yet covered.

October 30: More data for Marketing Analytics. Data Warehouse integrations with Meta Ads, LinkedIn Ads, Ahrefs, and HeyReach gained additional reporting fields. New information includes video engagement, conversion and reach metrics, landing page URLs, daily website metrics, and campaign statistics at the sender level. This enables more detailed reports combining advertising data with events and conversions captured by PostHog.

October 30: Expanded Data Warehouse catalog. PostHog added hundreds of tables to existing integrations. The updates cover advertising platforms, CRM systems, email marketing, payment platforms, analytics services, and developer tools. The new tables expose previously unavailable data about audiences, conversions, advertising campaigns, orders, subscriptions, and customer interactions.

About PostHog

PostHog is a product and digital analytics platform combining user event tracking, web analytics, marketing reporting, conversion analysis, Session Replay, A/B testing, and feature flags. Its Data Warehouse consolidates data from external systems, AI Observability monitors applications built on language models, and Self-Driving AI supports automated research into user behavior. PostHog enables teams to connect advertising platform data with actions taken on a website or in an app, analyze conversion funnels and retention, and measure the performance of individual product features. It offers cloud infrastructure and tools for self-hosting selected components.