The Complete Guide to Measuring User Retention (Cohorts, Curves & Benchmarks)

User retention is the percentage of users who keep coming back to your product over a given period. It's the single clearest signal of whether what you've built delivers lasting value — acquisition tells you people will try your product, but retention tells you whether it's worth keeping.

Most teams measure retention badly, or not at all. They track a single headline number, watch it wobble, and can't explain why. This guide covers how to measure retention properly: the formula, the difference between retention and churn, how to read a retention curve, how cohort analysis turns a flat number into something actionable, and what "good" actually looks like.

What is user retention?

User retention measures how many of your users remain active over time. If 1,000 people start using your product this month and 400 are still active next month, your retention rate is 40%.

Retention is the inverse of churn. Where retention counts who stayed, churn counts who left — and the two always add up to 100% for a given period. A product with 40% monthly retention has 60% monthly churn. Both describe the same reality; which one you lead with is usually a matter of whether you want to celebrate loyalty or confront loss.

How to calculate retention rate

The basic retention rate formula is:

Retention rate = ((Users at end of period − New users acquired during period) ÷ Users at start of period) × 100

You subtract new users so you're measuring whether existing users stuck around, not masking churn with fresh acquisition. For example: you start a month with 1,000 users, acquire 300 new ones, and end with 1,100 total. Your retained users are 1,100 − 300 = 800, so retention is 800 ÷ 1,000 = 80%.

The matching churn rate formula is simply:

Churn rate = (Users lost during period ÷ Users at start of period) × 100

A few related metrics worth knowing:

• Net revenue retention (NRR) measures retained revenue including expansion (upgrades, cross-sells) minus contraction and churn. Above 100% means your existing customers grow in value even if you add no one new — the holy grail for subscription businesses.
• Gross retention strips out expansion, showing the pure floor: how much you keep before any upsell.

Why a single retention number lies to you

A headline retention rate hides more than it reveals. "60% monthly retention" could describe a healthy product or a dying one depending on when users leave and which users you're looking at. To see the truth, you need two tools: the retention curve and cohort analysis.

Reading a retention curve

A retention curve plots the percentage of a group of users still active against time since they signed up — day 1, day 7, day 30, and so on. The shape of the curve is what matters:

• A curve that drops steeply and keeps falling to zero signals a product that isn't forming habits. People try it and abandon it.
• A curve that drops then flattens into a plateau is the healthy pattern. The plateau — the "smile" — represents your core of habitual users who've found lasting value. A flattening curve, even at a modest level, beats a higher curve that's still sliding downward.

The first job of retention measurement is to find where your curve flattens, and how high. That plateau, not the day-1 number, is your real retention.

Why cohort analysis changes everything

A cohort is a group of users who share a starting point — typically the week or month they signed up. Cohort analysis tracks each group's retention separately over time, instead of blending everyone into one average.

This matters because an aggregate number can stay flat while the underlying reality shifts dramatically. If your January cohort retains at 50% but each new monthly cohort retains a little worse, the blended average can look stable for months while your product quietly rots. Only cohorts expose this.

Cohort analysis answers the questions a single number can't:

• Are recent users retaining better or worse than older ones? This tells you whether product changes are working.
• Which acquisition channels bring users who stay? Cohort by source and the cheap channel that churns fast loses to the expensive one that sticks.
• Did a specific release move the needle? Compare cohorts from before and after the change.

When you can group users dynamically — by behavior, plan, source, or any property — cohorts stop being a reporting exercise and become the engine of your retention strategy.

What is a good retention rate?

The honest answer: it depends entirely on your product category, business model, and how you define "active." A daily-use social app and an annual tax tool have completely different healthy curves, and comparing them is meaningless.

Rather than chase someone else's benchmark, anchor on three principles:

1. Your curve should flatten. A plateau at any level is healthier than a curve still declining. Flattening means you've found a population for whom the product is genuinely sticky.
2. Newer cohorts should retain at least as well as older ones. Improving cohort-over-cohort retention is the clearest sign your product is getting better.
3. Define "active" honestly. Retention measured on a trivial action (opening the app) flatters you; retention measured on a meaningful action (completing the core workflow) tells the truth. Pick the action that represents real value and measure that.

The most useful benchmark is your own past. Improvement against your previous cohorts beats any industry average.

From measuring to improving

Measurement is only the setup. Once you can see where in the lifecycle users drop off and which cohorts struggle, you can act:

• Fix the early cliff. Most products lose the largest share of users in the first few days. A sharp day-1-to-day-7 drop points to an onboarding or activation problem — users aren't reaching the moment where the value becomes obvious.
• Intervene with the right message at the right time. If a cohort tends to go quiet around day 10, that's where a well-timed nudge, in-app guide, or relevant notification earns its place — not a generic blast, but a contextual prompt tied to where the user actually is.
• Double down on what sticky cohorts have in common. If users who adopt a particular feature retain far better, getting more users to that feature early becomes a concrete, measurable goal.

How Countly fits

Measuring retention well requires three things most setups lack: cohorts you can define dynamically, retention curves you can actually read, and the ability to act on what you find without exporting to a separate tool.

Countly brings these together. Dynamic cohorts let you group users by behavior, source, or any property and watch each group's retention over time. Retention reports show you the curve and where it flattens. And because engagement tools — journeys, messaging, surveys — live in the same platform, the gap between seeing a drop-off and doing something about it closes. For teams that need to keep that behavioral data in their own infrastructure, Countly supports that too, so your retention analysis never depends on handing user data to a third party.

Frequently asked questions

How do you calculate retention rate?Subtract new users acquired during the period from your total users at the end, divide by the number of users you started with, and multiply by 100. This isolates whether existing users stayed rather than letting new acquisition mask churn.

What is the difference between retention rate and churn rate?Retention measures the percentage of users who stayed; churn measures the percentage who left. For any period they sum to 100% — 70% retention means 30% churn.

What is a good user retention rate?There's no universal number; it varies by product type and business model. The better test is whether your retention curve flattens into a plateau and whether newer cohorts retain as well as or better than older ones.

What is a retention curve?A graph showing the percentage of a user cohort still active over time since signup. A healthy curve drops initially then flattens, indicating a stable core of habitual users.

What is cohort analysis in retention?Grouping users by a shared starting point (like signup month) and tracking each group's retention separately, so you can see how retention changes for newer versus older users instead of relying on a single blended average that can hide decline.

What is net revenue retention?The percentage of recurring revenue retained from existing customers over a period, including expansion from upgrades and minus losses from churn. Above 100% means existing customers grow in value on their own.

‍