Roadmap

Goals

Q4 2026 objectives

Last quarter we kept every product alive but moved none of them to a new level. This quarter we focus on fewer bets, pair on shared surfaces, and take what we have to GA – live, priced, and out of beta. We build MCP-first – if an agent can't use it, it isn't done – but keep the UI for users who still need it.

Take AEO and SEO to GA – Jordan MryyanJordan Mryyan

• Show the full crawled → cited → converted funnel per page and page group
• Get more crawler data in: one-click log drains (Vercel, Cloudflare, Netlify) to capture non-JS AI crawlers
• Alert on crawl and citation spikes or cliffs, and on pages that agents read but that error or load slowly for them
• Self-driving: turn content gaps and crawl errors into concrete fixes (FAQ, llms.txt, structured data, robots rules)
• Price it and ship it to everyone

Take agent experience to GA – Lucas RicoyLucas Ricoy

AI browsers (ChatGPT, Claude) now use customers' websites on behalf of people. Customers need to see how that goes:

• Detect agent sessions by default in posthog-js, so every customer gets agent vs. human traffic with zero config
• Compare agents and humans on the same pages: paths, conversion, and where agents get stuck
• Self-driving: flag flows that agents fail to complete (forms, sign-up, checkout) and suggest fixes
• Price it and ship it to everyone

Make web analytics agent-native – Lucas RicoyLucas Ricoy

If agents become the main way people use PostHog, web analytics has to work great without our UI:

• Let the wizard set up web analytics end to end, including a conversion goal
• Measure MCP answers to web, marketing, and AEO questions with evals, and fix what's weak
• Ship ready-made prompts: a web analytics dashboard as an artifact, and a scheduled agent task that watches your site and flags problems

Take ad performance to GA – Javier BahamondesJavier Bahamondes

Ad spend is business data, and PostHog should be the best place to correlate it with everything else customers send us:

• Attribute from start to finish: connect every source – paid ads, SEO, AEO, and agents – to revenue and new customers, using conversion goals and revenue events
• Give customers a clear view of their ad optimization data that they can use from their own agents through the MCP, including write tools to set up sources and conversion goals
• Self-driving: spot underperforming campaigns and suggest budget and setup changes that the customer's agent can apply
• Finish the pricing review and remove the beta label – customers and agencies are waiting on it before they commit

Cancel Vitally – Arthur Moreira de DeusArthur Moreira de Deus

• CS, onboarding, and sales run 100% on Customer Analytics and we cancel Vitally in October
• Improve reliability and monitoring for Account creation, property syncs and data ingestion
• Built at least one self-driving loop: from signal triage to surfacing it for the user

House cleanup

• Heatmaps and the main web analytics dashboard: no new features this quarter. Ship the remaining heatmaps SDK change and the paywall comms, then remove the heatmaps beta label.
• Performance: roll out Sessions v3 to fix one-off queries for the largest teams

Q3 2026 recap

• Marketing Analytics: organic insights, setup fixes, and MCP suggestions for sources and conversion goals shipped behind flags – not self-serve, priced, or GA yet
• Customer Analytics: on track to replace Vitally in October, with performance parity and team workflows migrated one by one
• Heatmaps: paywall, toolbar gating, and page-specific monitoring shipped
• Self-driving: scouts produce concrete code changes in dogfood (bot definitions, web vitals with Lighthouse, CSP), but not for customers yet
• Agent and bot analytics not fully released – non-JS crawler capture still blocked on server-side ingestion
• Precomputed improvements released to everyone: 85%+ of queries under 200ms
• Accounts list load fixed; Marketing Analytics dashboard p95 not improved

Q2 2026 recap

• Activation: reversed the decline in Web Analytics activation
• Signals, Skills & MCP: Marketing Analytics AI and MCP integration shipped, but read-only and lightly used so far
• Agent and bot analytics: building blocks and docs merged, working everywhere HogQL works (12 teams and ~15M events over 21 days organically) – full release pending
• Marketing Analytics: PMF investigation nearly wrapped, performance improved (p95 ~23s to ~15s), drill-down navigation shipped (channels to sources to campaigns, UTM and ad group breakdowns)
• Precomputed improvements ready to release to everyone (3–30x speedups)
• Arthur joined the team to lead the Customer Analytics / Vitally replacement effort
• Virtual HogQL properties and Google Search Console integration shipped

Q1 2026 recap

• Pre-aggregated tables shipped to 50% of teams
• Session explorer performance improvements
• Filters and UX revamp across Web Analytics
• Heatmaps inherited and stabilized with many fixes
• Marketing Analytics: improved ROAS, added new sources, ongoing support post-GA
• Agent/bot basic segmentation shipped, browser detection deferred to Q2
• Installation Health v2 partially shipped

Q4 2025 recap

• Marketing Analytics launched as GA with strong user feedback
• Web Analytics integrated with PostHog AI
• Shipped Web Analytics components as insights
• Pre-aggregated tables made available for selected teams
• Query concurrency and performance improved
• Many bug fixes and UX improvements across Web Analytics

Handbook

Who are we building for?

Personas

• Primary Personas:

  • Technical Founder
    • First technical person at the company, responsible for setting up the initial landing page.
    • At an early stage, wants tools that initially cost little to nothing, but won't need to be replaced as the company grows.
  • Product engineer
    • These are the engineers building the product. Normally full-stack engineers skewing frontend or frontend engineers.
    • Product engineers have more limited time. Need to quickly get high-quality insights to inform what they are building and assess what they've shipped.

• Not a focus, but should be usable by:

  • Other engineers
  • Marketing
  • Everyone in the product team (PMs, designers)
  • Non-technical leadership

What types of companies?

• We are building web analytics for high-growth startups that have PMF. Our product should be useful for them right from the start.

Jobs to be done

Web analytics is an opinionated tool, that will help people:

• Measure and improve the effectiveness of their acquisition strategy
• Measure and improve the effectiveness of their conversion funnels
• Understand who is using their website
• Understand how their website is being used
• Create simple-to-medium complexity combinations of these (e.g. how likely are people who came in from this ad to convert, compared to everyone else)

Additionally, it should be:

• Easy to set up
• Easy to use
• It's ok if that means we don't cover every use case, as people who need more can use product analytics

Roadmap

5-year vision

Web analytics as the command center Web analytics becomes the central command hub for everything that happens on the web. It unites acquisition, performance, and behavior data in one intelligent view. It is the 80/20 dashboard for all related PostHog products, connecting directly to experiments, sessions, and insights. No longer a passive reporting tool, it’s where teams take action.

From reactive to proactive Analysis turns into action. Web analytics doesn’t just highlight regressions; it solves them. When conversion rates drop or core web vitals slip, it can open a GitHub PR, create a task, or launch a follow-up experiment automatically.

A self-optimizing web Everything related to the web lives here: marketing spend, attribution, churn risk, SEO/AEO/GEO, and performance metrics, all in one continuous feedback loop. PostHog AI connects the dots across marketing, performance, and behavior, finding opportunities for improvement, generating new assets or variants, and running experiments autonomously.

From chart-drilling to conversations Instead of drilling into dashboards, users simply chat with web analytics. They ask questions, get clear explanations, and receive ready-to-apply actions, all in natural language. The focus shifts from just analyzing data to acting on insights.

Analytics for the agent era As AI agents and LLMs become real web users, PostHog helps teams measure, understand, and optimize for them. Web analytics will distinguish between agent traffic and human traffic, paving the way for the area and empowering teams to tailor their experiences for both human and AI audiences.