How to Build a Retention Analysis in PostHog
Acquisition metrics tell you how many users showed up. Retention metrics tell you how many stayed. For most products, improving retention by five percentage points has a bigger impact on growth than doubling signups, because retained users compound — they generate revenue, refer others, and deepen their usage over time. Yet retention is the metric most teams measure least rigorously.
PostHog's retention analysis tools — retention tables, lifecycle charts, and cohort breakdowns — give you the infrastructure to measure retention properly without a data warehouse. This guide walks through setting up retention analysis that answers three questions: what percentage of users come back, when do they stop coming back, and what behaviors predict long-term retention.
Step-by-step guide
Define your retention events and time windows
Retention measurement requires two events: the 'start' event that begins the clock (typically first login or signup) and the 'return' event that counts as a retained visit. The return event should be a meaningful action, not just a pageview — logging in, performing the core action, or viewing a key screen. Define your retention window: day, week, or month depending on your product's natural usage frequency.
- Choose your start event (signup, first login, or onboarding completion)
- Choose your return event (should indicate the user got value, not just visited)
- Select your retention window: use weekly for products used multiple times per week, monthly for less frequent products
Create a basic retention table in PostHog
Navigate to PostHog's Retention section and configure a retention table. Set the start event, return event, and group users into weekly cohorts by their start date. PostHog will generate a triangular table showing what percentage of each cohort returned in week 1, week 2, week 3, and so on. The first column shows cohort size; subsequent columns show the retention curve.
- Go to Insights > New Insight > Retention
- Set 'First event' to your start event and 'Returning event' to your return event
- Set the period to 'Week' and the number of periods to 12 for a three-month view
Analyze the retention curve shape
Look at the retention curve for your largest cohort. A healthy curve flattens after the initial drop — users who survive week 2-3 tend to stick around. If the curve keeps declining without flattening, you have a leaky bucket: users try the product but never form a habit. The week where the curve flattens (if it does) is your 'retention inflection point' — getting users past that point is your activation challenge.
- Identify the steepest drop-off week (usually week 1 to week 2)
- Check if the curve flattens and at which week
- Compare to industry benchmarks: B2B SaaS should see week 4 retention above 30% for weekly usage
Segment retention by user properties
Break down the retention table by user properties: plan type, company size, acquisition channel, and onboarding completion status. This reveals which segments retain best and worst. You might find that users who completed onboarding have 50% week-4 retention while those who skipped it have 15%. That gap is your highest-leverage improvement opportunity.
- Add a breakdown by 'plan type' and compare retention curves
- Break down by 'onboarding completed' (true/false) to quantify onboarding impact
- Create separate retention tables for each key segment to track improvements over time
Set up lifecycle analysis to track user state transitions
PostHog's Lifecycle chart shows users in four states: New (first time performing the event this period), Returning (performed the event this period and last), Resurrecting (returned after being dormant), and Dormant (performed the event last period but not this period). This dynamic view complements the static retention table by showing flows between states.
- Go to Insights > New Insight > Lifecycle
- Set the target event to your core return action
- Set the period to 'Week' and look at the balance between Returning and Dormant users
Build a retention dashboard and set alerts
Create a PostHog dashboard with: the retention table (overall), retention tables segmented by your top two properties, the lifecycle chart, and a trend of weekly active users. Add an alert that fires when week-1 retention drops below your threshold for two consecutive cohorts — this catches retention regressions early before they compound into churn.
- Create a new dashboard and pin all four retention insights
- Set up a subscription to receive the dashboard via email weekly
- Configure an alert on week-1 retention dropping below your target (e.g., 40%)
Common mistakes
Using pageviews as the return event
A user who lands on your marketing page via a Google search is not 'retained.' The return event must represent meaningful product engagement — logging in, performing the core action, or spending meaningful time in the app. Pageviews inflate retention numbers and mask the real health of your product.
Drawing conclusions from tiny cohorts
A retention table for a cohort of fifteen users is noise, not signal. You need at least fifty to one hundred users per cohort for the percentages to be statistically meaningful. If your weekly cohorts are smaller than this, use monthly cohorts instead.
Only measuring overall retention without segmentation
Aggregate retention hides the reality that some segments retain at 60% and others at 10%. Without segmentation, you cannot target improvements at the lowest-retaining group. Always break retention down by at least plan type and onboarding completion status.
Tips
Track 'unbounded retention' (users who return at any point after day N) alongside 'N-day retention' (users who return specifically on day N) — unbounded is more forgiving and often more representative of real usage patterns.
Create a 'Power User Retention' table that uses a high-frequency return event (e.g., performed core action 3+ times) to measure retention of your most engaged segment.
Compare retention curves before and after major product changes to quantify impact — overlay the pre-change and post-change cohorts on the same chart.
Use PostHog's path analysis alongside retention to understand what retained users do differently from churned users — this reveals the behaviors worth incentivizing.
How Vantage helps
Vantage integrates retention data as context when generating PRDs for growth or engagement features. If your PostHog retention shows a steep drop at week 2, Vantage surfaces this pattern during requirement generation so the PRD specifically targets the activation gap that causes early churn.