Cohort Retention

What Cohort Retention Actually Shows

Cohort retention groups users by the date (or week, or month) they signed up, then tracks what percentage of that group is still active at fixed intervals afterward: day 1, day 7, day 30, and so on. Instead of asking "how many users do we have," it asks "do the users we get actually stick around," which is a far more honest question for an early-stage product.

A "cohort" is simply a batch of users who share a start date. "Retention" is the share of that batch still doing meaningful activity later. Put together, cohort retention turns a single vanity metric (total signups) into a curve that reveals the real health of your product.

Why It Matters for Early-Stage Founders

Aggregate metrics lie. If you look at total weekly active users, growth can mask a leaky bucket: you might be adding 500 new users a week while losing 480 old ones, and the top-line number still goes up. Cohort retention exposes this immediately because it isolates each signup group and shows exactly where they drop off.

For founders, this matters in three concrete ways:

• It's the clearest signal of product-market fit. If a meaningful share of users from every cohort keeps coming back on their own, without you nudging them, you have something people want.
• It tells you whether growth spend is worth it. Paid acquisition into a product with poor retention just buys you a faster churn cycle.
• It shows if the product is improving. Comparing cohort curves month over month tells you if new features and onboarding changes are actually helping people stay, not just guessing.

Investors ask for retention curves specifically because they can't be gamed the way "total users" can. A flattening curve is one of the strongest pieces of evidence you can put in a pitch deck.

How to Calculate It

Pick a cohort (e.g., everyone who signed up in the first week of March), pick an activity definition (e.g., "logged in" or "completed a core action"), and pick your time intervals (day 1, day 7, day 30, day 90 are common).

Formula:

Cohort Retention (Day N) = (Users from the cohort active on Day N / Total users in the cohort) x 100

Example:

Say 200 users sign up for your app in Week 1 (this is your cohort). You track how many of those same 200 people log in during each following week:

• Week 1 (signup week): 200 active (100%)
• Week 2: 90 active (45%)
• Week 4: 60 active (30%)
• Week 8: 48 active (24%)
• Week 12: 44 active (22%)

Plotted, this forms a retention curve that drops sharply at first, then flattens. The flattening point matters more than the initial drop: it represents your "resilient" user base, the people who found real value and are likely to stay long-term.

Run this same analysis for every weekly (or monthly) cohort, lay the curves side by side, and you can see whether newer cohorts retain better than older ones, which is the clearest proof that your product (or onboarding) is actually improving.

Reading the Curve: What Good Looks Like

There's no universal "good" number since it varies wildly by product category, but the shape matters more than the absolute number:

• Smiling curve (retention dips then rises): Rare and excellent. Usually seen in social or network products where value compounds over time.
• Flattening curve (drops then levels off): Healthy. Signals a durable core user base has formed.
• Curve that keeps declining toward zero: Bad. The product isn't sticky; you're renting users, not keeping them.

For consumer apps, a day-30 retention of 25-40% is often considered decent; for B2B SaaS with weekly-use products, 60%+ month-over-month retention within an engaged segment is a stronger benchmark. Always benchmark against your specific category rather than generic numbers.

Common Mistakes

• Using signups instead of activated users as the cohort base. This dilutes the data with people who never really tried the product. Cohort on "activated" users when possible for a cleaner signal.
• Picking a weak definition of "active." If "active" just means opening the app, you'll overstate retention. Tie it to a core action that reflects real value (posting, completing a task, making a purchase).
• Only looking at blended retention. Averaging all cohorts together hides whether things are getting better or worse. Always compare cohort by cohort.
• Giving up too early. Retention curves take a few weeks or months of data to show a real pattern; judging from day 1 and day 7 alone is premature.

The Bottom Line

Cohort retention is the single best early-warning system for whether you've built something people want to keep using. Before you pour money into acquisition channels or plan a splashy launch on welaunch.sh, make sure your retention curve is flattening, not falling. Growth on top of weak retention just accelerates how fast you find out the product isn't sticky.