How to Build a Retention Chart in PostHog for a SaaS

To build a retention chart in PostHog, use the native Retention insight: pick the event that marks a new user and the event that marks a return, and it produces the cohort grid automatically. For a supporting trend, compute weekly active users in HogQL. Retention is the truest measure of whether your product actually sticks — conversion gets people in; retention is whether they stay.

Build the chart with the native Retention insight

Retention curves are a cohort calculation that the native insight is built for, so start there:

1. In PostHog, open Insights → Retention.
2. Set the cohortising event to what makes someone a new user — signup_completed.
3. Set the returning event to what counts as coming back — any $pageview, or better, a core action like project_created.
4. Choose a weekly granularity for a SaaS product.

PostHog produces a grid: each row is a weekly cohort, each column is weeks-since-signup, and each cell is the share of that cohort still active. Reading it is simple — you want the numbers to flatten across columns rather than fall to zero, which means a stable core keeps returning.

Picking the right returning event matters. "Any pageview" overstates retention (people who log in but get no value still count); a genuine core action gives you the honest curve.

A supporting signal: weekly active users in HogQL

The retention grid shows cohort behaviour; a weekly-active-users trend shows the overall trajectory. This HogQL counts unique active users per week:

SELECT
  toStartOfWeek(timestamp) AS week,
  uniq(person_id) AS weekly_active_users
FROM events
WHERE timestamp > now() - INTERVAL 12 WEEK
GROUP BY week
ORDER BY week

Illustrative sample output:

week	weekly_active_users
2026-04-06	1,240
2026-04-13	1,310
2026-04-20	1,395

Rising WAU with a flattening retention curve is the healthy combination: you're adding users and keeping them. Rising WAU with a retention curve that still drops to zero means you're filling a leaky bucket — growth that masks a retention problem.

How to read retention for a SaaS

• Flattening beats high. A curve that settles at 30% and stays there is healthier than one that starts at 60% and slides toward zero.
• Compare to your own cohorts. Retention varies enormously by product; last quarter's cohorts are your real benchmark, not an industry number.
• Tie drop-off to onboarding. Most retention is won or lost in the first session — if week-1 retention is weak, your activation step is the place to work.

From measurement to action

Retention tells you whether the product delivers value; if it doesn't flatten, no amount of top-of-funnel conversion will save the numbers. Once your curve is stable, conversion work compounds. If you'd like the conversion side handled for you — the highest-impact fix found and shipped as a Pull Request weekly — that's what Velyr does.