Free tool · Retention and conversion

Cohort retention calculator

Calculate point-in-time cohort retention: how many original users returned on your selected day or window.

Build your scenario

Runs in your browser. No signup.

Starting values are editable examples.

D7 retention
30%
Original users inactive in window
700

This is point-in-time retention, not rolling retention. Do not mix new users, different windows, or different activity events.

Free, no sign-up. Your numbers stay in your browser.

Updated October 30, 2026

What cohort retention is

Cohort retention measures how many people from the same group come back to the app after a given time. The cohort is the group that started together, for example everyone who installed in the same week. D7 retention is the share of that group that was active on day seven.

Measuring by cohort avoids a common confusion: if the app gains many new users, the number of active users goes up even if almost nobody comes back. By always following the same group, retention shows whether people keep using the app.

How to calculate it

Dn retention = cohort users active on day n ÷ original cohort size × 100

  1. Define the cohort: who installed or signed up in a period.
  2. Define what counts as active, for example opening the app and completing a valuable action.
  3. Count how many people in the cohort were active on the chosen day or window.
  4. Divide by the cohort’s original size and multiply by 100.

One cohort over time

An example cohort of 1,000 people, measured on three days:

DayActive from cohortRetention
D145045%
D730030%
D3018018%

The retention curve

Putting several days side by side reveals the cohort’s curve. It drops fast at first and then flattens. The point where it flattens shows the group of people who adopted the app; a curve that never flattens means almost nobody builds the habit. Continuing the example:

DayActive from cohortRetention
D3018018%
D6017017%
D9016516.5%

The cohort table

Putting several cohorts in one table, each row is a group and each column a day. Reading down a column, you see whether retention for the same day improves or worsens between weeks. The last cohort has not reached 30 days yet, so its D30 is blank.

CohortSizeD1D7D30
Week 11,00045%30%18%
Week 21,20043%28%17%
Week 390046%33%—

Point-in-time and rolling retention

This calculator measures point-in-time retention: who was active on that day or window. There is another form, called rolling retention, that counts who came back on that day or any day after. Rolling retention is always greater than or equal. Both are useful, but they cannot be mixed in the same comparison.

Comparing cohorts

Compare rates, not counts. A bigger cohort can have more active users and worse retention:

CohortSizeActive on D7D7 retention
Week 11,00030030%
Week 21,20033628%
Week 390029733%

Retention by acquisition channel

Splitting cohorts by the channel that brought the user shows which channels bring people who stay. A channel with cheap installs can bring users who disappear within days:

ChannelCohortActive on D7D7 retention
A1,00025025%
B50017535%
Compare channel CPI

How long to wait before measuring

A cohort only has D30 retention after it completes 30 days. Comparing an old cohort’s D30 with one that is only 20 days old mixes complete and incomplete data. Wait for each cohort to reach the day you are measuring before putting it in the comparison.

What counts as active

Define the active event before looking at the numbers and do not change it between cohorts. Opening the app is the simplest definition, but it can count someone who opened from a notification and left right away. A value event, such as logging a meal in a diet app, better shows who is really using the product.

Retention and revenue

In subscription apps, user retention and subscriber retention are different metrics. User retention shows whether people keep using the app; subscriber retention shows whether they keep paying. Both affect LTV, the value each customer leaves over time.

Calculate cohort LTV

Subscriber retention

For payers, the question is how many renew. In a group of 200 monthly subscribers, if 170 renew in the second month and 150 remain in the third, subscriber retention is 85% in month 2 and 75% in month 3. The flip side is churn: 15% left in the first month.

MonthSubscribersRetention from start
1200100%
217085%
315075%
Calculate subscriber churn

Split by platform and country

Users on iOS and Android, or in Brazil and the US, can have very different retention. Adding everything into one cohort hides those differences. Measure them separately and only combine when the numbers are similar.

What one point of retention is worth

An example for scale, with an assumption you should replace with your own numbers. In a weekly cohort of 1,000 people, raising D30 retention from 18% to 21% keeps 30 more people in the app. If 10% of active users subscribe at $29.90 per month, that is 3 more subscribers and $89.70 of monthly revenue for each week of new users. Over a year of cohorts, the improvement compounds.

Where retention usually drops

The biggest drop usually happens right at the start, between install and the first days. That is where onboarding has to show the app’s value. To study how apps that already make money handle that first experience, Noctral’s library shows each app’s onboarding screen by screen.

See the onboarding and paywall library

How to improve retention

  • Show the app’s value in the first session, before asking for sign-up or payment.
  • Lead people to their first valuable action during onboarding, not just to a welcome screen.
  • Use notifications that bring something useful, at the pace the app is used.
  • Fix what makes people leave early, by looking at where the first session ends.
Plan notification volume

Retention and DAU/MAU

Retention follows a fixed group over time. The DAU/MAU ratio looks at the whole base at one moment: how many of the month’s active users used the app on a given day. They complement each other: retention shows whether people stay, and DAU/MAU shows how often those who stayed use the app.

Calculate the DAU/MAU ratio

Where the numbers come from

  • Cohort size: the install or sign-up event in your analytics tool.
  • Active on the day: the active event, filtered to the cohort’s people only.
  • Window: the same day or interval for every cohort compared.

How to use the calculator

  1. Pick the cohort and the day you want to measure.
  2. Enter the cohort’s original size and how many were active on that day.
  3. Compare the result with the same day for other cohorts, using the same active definition.

Common mistakes

  • Counting new users together with the cohort’s.
  • Comparing cohorts with different active definitions.
  • Mixing point-in-time and rolling retention on the same chart.
  • Drawing conclusions from cohorts too small, where a few people swing the rate a lot.

See it in real data

Onboarding and paywall

Screens from apps that already earn, step by step

See the feature

Frequently asked questions

How do I use the cohort retention calculator?

Calculate point-in-time cohort retention: how many original users returned on your selected day or window.

What counts as active?

Define a product value event and window before analysis; do not change the definition between cohorts.

Why is my D1 retention low?

Most of the time, the first session did not show the app’s value in time: a long sign-up, permissions asked too early or a welcome screen with no action. Look at where people leave in the first session.

Which retention day should I track?

D1, D7 and D30 are the most used: D1 shows the first impression, D7 whether the habit started and D30 whether the app stuck. Pick the ones that fit how often your app is used.

Are user retention and subscriber retention the same thing?

No. User retention measures who keeps using; subscriber retention measures who keeps paying. A subscriber can use little and keep paying, and an active user may never pay.

Do I need an account and are my inputs uploaded?

No account is required. Calculations run in your browser. Entered numbers and text are not included in tool analytics events; only the tool ID, language, and action are recorded.

Want to go past the numbers and see the real market?

Noctral shows the estimated revenue, the ads and the paywalls of apps that already earn, so you decide with market data and not just projections.

See Noctral data