Product Analytics Team

Product Analytics Team

See what we're building
At a recent small team offsite in Cambridge, UK


What we're building

  • Data table exploration

    We're building an insights-like search functionality for everything qualitative. That means people, recordings, cohorts, events, and groups. This will enable you to search using simple sentences to easily find anything in PostHog. Additionally, you can trigger context-sensitive actions - such as cr...

    Project updates

    No updates yet. Engineers are currently hard at work, so check back soon!


Here’s what we’re considering building next. Vote for your favorites or share a new idea on GitHub.

Recently shipped

HogQL-powered insights

We've rewritten all our insights in HogQL and are releasing them, starting with trends.

This makes insights faster, more reliable, and enables advanced customization.


Q2 2024 Objectives

  • HogQL-based querying

    • Convert the remaining legacy queries to HogQL and release to public (Thomas, Julian, Marius)
      • Trends? Funnels? Paths?
      • Cohorts
    • Remove legacy querying backend (Thomas, Julian)
      • Convert filters to query (insights, notebooks, activity log, experiments)
      • Clean up or rewrite dashboardLogic
    • Missing Product Analytics features (Thomas, Julian)
      • Breakdowns (multiple) in literally everything
      • Make a list based on GitHub issues from customer requests…
    • Missing HogQL features (Tom, Marius)
      • Type system, JSON
      • Missing things when building funnels
  • Querying and processing performance (Thomas, Julian)

    • Global performance overview dashboards
      • Insights
      • Exports
      • Cohort recalculations
    • Query request tracing
      • Possibly query runner Python optimizations
      • Exports improvements
    • Identify top 5 query optimizations in terms of impact
  • Artificial Hog / Post Intelligence (Michael)

    • Ask a question to get a magical insight (aware of your taxonomy)
    • Figure out infra for upgrading queries and models
    • Product-wide framework for opting into sharing with OpenAI
  • Activation (side quest: Michael)

    • Michael to work with Growth to identify optimizations to getting started with Product Analytics

Q1 2024 Objectives

  1. HogQL & Data Exploration (Julian, Marius)
  • Convert ALL our insights to use HogQL as their base.
    • Most product analytics insights are built in a bespoke way using our "legacy" Python codebase. This code produces ClickHouse SQL as its output, and is hard to extend, as all queries mix both platform level concerns (e.g. how "group analytics" works) and query level concerns (most optimal query for this data structure). Adding new query types or new platform features has turned out to be a nightmater with so many loose parts to consider.
    • We've built the HogQL platform, which takes care of implementation details (e.g. where precisely are person properties stored) and lets users focus on writing queries. We're now rewriting all our insights to output HogQL instead of ClickHouse SQL. This will make the system more reliable, fix whole classes of bugs, and allow the end user to directly modify the generated HogQL SQL for further analysis.
    • The migrated queries are also a blocker for the removal of a lot of legacy frontend code, which is often causing bugs.
  • Improve the type system and get rid of assumeNotNull.
    • ClickHouse is notoriously cranky when it comes to NULL-s. Whenever it sees any unaccounted null anywhere, the entire row is discarded from analysis. We have several fields and properties, which are sometimes NULL, but mostly not. We have special handling for nullable fiels in the concat function, and in all compersions. For example (0 != NULL) feels like it should return true, but in ClickHouse it returns NULL. In HogQL it returns true due to our special handling.
    • This special handling in HogQL makes for better predictable queries, at the cost of some performance. We can optimise this further if we know a field to definitely not be NULL. Unfortunately this nullability information gets lost once fields pass functions and subquery boundaries. Upgrading the type system to account for this is the goal.
  • Write great docs for HogQL and data exploration nodes.
    • Make sure all Data Exploration Nodes, including HogQLQuery, are documented somewhere.
    • Can we build a live API playground?
  • Proactively monitor query performance.
  1. Product Analytics frontend (Thomas, Michael)
  • Clean things up now that we have PostHog 3000 and HogQL everywhere.
    • Migrate Insights in the database away from "filters" onto "query"
    • Purge legacy UI
    • Purge legacy python insights
  • Prioritise and work on any of the many ideas we have to make the app better:
    • Universal Exploration view
    • Reworked filtering experience
    • Query builder 3000
    • Notebook-first flow
    • Make Product Analytics as easy as Web Analytics
  1. BI (Tom)
  • Support non-event data sources on HogQL insights.
  • Build the new querying experience, and/or integrate it into the "Explore" view.


Who are we building for?


  • Primary Personas:
    • 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.
    • Product manager (ex-engineer type)
      • Supports the product teams (engineers, PMs, designers) to build the best products. They guide the product roadmap by speaking to customers and diving into the data.
      • Product managers are the power-users of analytics (further evidence in the data analysis of paying users). They have desire and the time to go significantly deeper into the data.
  • Limited focus:
    • Growth engineer
  • Not a focus but should be usable by:
    • Everyone in the product team (less technical PMs, designers)
    • Marketing
    • Leadership team

What types of companies?

The highest-performing product teams building the most loved products at high-growth startups. For more context on the company read about the ideal customer persona.

Jobs to be done

Product analytics is a wide tool which fulfills many job-to-be-done (non-exhaustive list):

  • Monitoring KPIs - how are the specific KPIs (product/team/company) doing? Are there any big changes, is everything going roughly in the right direction?
  • Insights into a new feature you've built - I've created a new feature and I want to make sure that it's being used successfully
  • Watching for errors and debugging - something went wrong (error gets trigger, product regression, drop in conversion), getting told it went wrong, debugging it, shipping a solution and making sure that fixes it
  • Conversion optimization - the growth team is monitoring how particular KPIs are doing, trying to come up with experiments, shipping features to try and improve these
  • Answering product strategy questions - which customers should we focus on, what are our most used/valued features. e.g. should we increase the pricing from X to Y? Which customers should we focus on?

You can broadly group the job-to-be-done of Product Analytics in PostHog as:

  • Creating: You have a specific query/dashboard in mind, you open PostHog to view it. E.g. creating a dashboard to Monitor KPIs, or creating the funnel for your onboarding flow
  • Consuming: you or someone else has made something in Posthog that you refer back to. E.g. Checking the dashboard you made to Monitor KPIs
  • Exploring: you're answering a broader open-ended question. E.g. If you're monitoring your KPIs and you see something not right - you then want to dive into understanding why


3 year goals

  • You can explore data across all insights and dimensions
  • You can trivially share any insight anywhere
  • Onboarding is as easy as a video game
  • Tight integration with developer workflows
  • No more complex than it is today
  • Using PostHog sparks joy
  • We support trillion event querying

Feature ownership

You can find out more about the features we own here