Skip to main content
Solved

Cohort Descriptions - Adjust vs RevenueCat & Late Converters Issue

  • October 7, 2026
  • 5 replies
  • 25 views
Forum|alt.badge.img

We use RevenueCat and Adjust together, but their cohorting logic is different.

1. User count mismatch:
Adjust counts each reinstall as a new install, while RevenueCat keeps the user tied to their original first_seen_time. So Adjust can show significantly more users in the short term.

What is the recommended way to reconcile these two systems and make user counts comparable?

2. Delayed conversions:
We cohort users based on first_seen_time. However, a user may not start a trial or pay during their original cohort and then return months later and become a paid user. In this case, looking only at the first_seen cohort can make us miss when the actual conversion happened.

Would it be better to keep first_seen as the acquisition cohort but create a separate conversion cohort based on the first trial/paid event? How do you usually handle this?

Best answer by Crudyg123

For acquisition decisions, I’d keep late converters tied to their original acquisition cohort and measure its results at a consistent age, such as 30, 60 or 90 days. Their payments still happen later; you’re linking that revenue back to the spend that acquired them.

If you mean CAC per paying customer, calculate acquisition spend for that cohort divided by the unique customers from that same cohort who have paid by the chosen cutoff. For example, $1,000 spent with 100 payers by day 30 gives $10 CAC. If another 25 first pay by day 90, it becomes $8, assuming no additional acquisition cost. Spend per acquired user stays unchanged. Dividing this month’s spend by everyone who first pays this month would mix different acquisition cohorts.

In RevenueCat, New Customers with Realized LTV / Customer lets you see revenue accumulating as those users eventually convert. Match the customer population to the spend you’re comparing:
https://www.revenuecat.com/docs/dashboard-and-metrics/charts/cohort-explorer

For budget allocation, I’d compare campaigns at the same age and use older, comparable cohorts to estimate the later conversions for newer ones. Keep that forecast separate from actual results. Adjust’s cohort maturity and cumulative metrics matter here:
https://help.adjust.com/en/article/how-cohorts-work

I’d set the CAC target around expected value and a payback period you can afford, allowing for store fees, taxes and servicing costs. Late conversions can improve eventual returns while still leaving you waiting months to recover the spend. If paid retargeting brings users back, include that extra cost separately so it doesn’t look like free conversion uplift.

5 replies

Forum|alt.badge.img+5
  • Crudyg123
  • Active Helper
  • October 7, 2026

I’d keep both views. The acquisition cohort tells you how users acquired in a given month eventually convert, while the conversion cohort tells you when they actually start a trial or pay.

RevenueCat’s Cohort Explorer already has this split: Initial Conversions groups customers by their first conversion, including a free trial, and New Paying Customers groups them by their first payment. Someone acquired in January who first pays in April can appear in January’s acquisition cohort and April’s paying customer cohort.
https://www.revenuecat.com/docs/dashboard-and-metrics/charts/cohort-explorer

For the count mismatch, check the Adjust metric first. Adjust distinguishes reinstalls from qualifying redownload installs. Redownload installs count toward Installs and can start a new cohort depending on your settings, so every reinstall isn’t necessarily a new user.
https://help.adjust.com/en/article/redownloads

I’d compare the same date range and population, separate new acquisitions from returning users, and check whether your RevenueCat App User IDs stay consistent across reinstalls. Otherwise you’re comparing install events on one side with customer identities on the other, and the totals won’t necessarily match.
https://www.revenuecat.com/docs/customers/identifying-customers

    Forum|alt.badge.img
    • 525b040404040
    • Author
    • New Member
    • October 8, 2026

    Thank you for your response! Our main concern is actually on the acquisition side. When setting a target CAC, we previously didn’t account for “late converters,” since the acquisition spend had already been incurred, while the revenue from these users only arrived later, creating a timing mismatch.

    Could you elaborate on your answer specifically from an acquisition and marketing decision-making perspective? For example, how should we account for these late converters when evaluating CAC and making acquisition or budget allocation decisions? 

    Forum|alt.badge.img
    • 525b040404040
    • Author
    • New Member
    • October 8, 2026

    I’d keep both views. The acquisition cohort tells you how users acquired in a given month eventually convert, while the conversion cohort tells you when they actually start a trial or pay.

    RevenueCat’s Cohort Explorer already has this split: Initial Conversions groups customers by their first conversion, including a free trial, and New Paying Customers groups them by their first payment. Someone acquired in January who first pays in April can appear in January’s acquisition cohort and April’s paying customer cohort.
    https://www.revenuecat.com/docs/dashboard-and-metrics/charts/cohort-explorer

    For the count mismatch, check the Adjust metric first. Adjust distinguishes reinstalls from qualifying redownload installs. Redownload installs count toward Installs and can start a new cohort depending on your settings, so every reinstall isn’t necessarily a new user.
    https://help.adjust.com/en/article/redownloads

    I’d compare the same date range and population, separate new acquisitions from returning users, and check whether your RevenueCat App User IDs stay consistent across reinstalls. Otherwise you’re comparing install events on one side with customer identities on the other, and the totals won’t necessarily match.
    https://www.revenuecat.com/docs/customers/identifying-customers

    Thank you for your response! Our main concern is actually on the acquisition side. When setting a target CAC, we previously didn’t account for “late converters,” since the acquisition spend had already been incurred, while the revenue from these users only arrived later, creating a timing mismatch.

    Could you elaborate on your answer specifically from an acquisition and marketing decision-making perspective? For example, how should we account for these late converters when evaluating CAC and making acquisition or budget allocation decisions? 

    Forum|alt.badge.img+5
    • Crudyg123
    • Active Helper
    • Answer
    • October 8, 2026

    For acquisition decisions, I’d keep late converters tied to their original acquisition cohort and measure its results at a consistent age, such as 30, 60 or 90 days. Their payments still happen later; you’re linking that revenue back to the spend that acquired them.

    If you mean CAC per paying customer, calculate acquisition spend for that cohort divided by the unique customers from that same cohort who have paid by the chosen cutoff. For example, $1,000 spent with 100 payers by day 30 gives $10 CAC. If another 25 first pay by day 90, it becomes $8, assuming no additional acquisition cost. Spend per acquired user stays unchanged. Dividing this month’s spend by everyone who first pays this month would mix different acquisition cohorts.

    In RevenueCat, New Customers with Realized LTV / Customer lets you see revenue accumulating as those users eventually convert. Match the customer population to the spend you’re comparing:
    https://www.revenuecat.com/docs/dashboard-and-metrics/charts/cohort-explorer

    For budget allocation, I’d compare campaigns at the same age and use older, comparable cohorts to estimate the later conversions for newer ones. Keep that forecast separate from actual results. Adjust’s cohort maturity and cumulative metrics matter here:
    https://help.adjust.com/en/article/how-cohorts-work

    I’d set the CAC target around expected value and a payback period you can afford, allowing for store fees, taxes and servicing costs. Late conversions can improve eventual returns while still leaving you waiting months to recover the spend. If paid retargeting brings users back, include that extra cost separately so it doesn’t look like free conversion uplift.

    Forum|alt.badge.img
    • 525b040404040
    • Author
    • New Member
    • October 8, 2026

    Truly appreciated! We also share weekly LTV predictions with our marketing team and set “CAC Targets” based on these predictions. This creates another issue for us. We currently take the number of paid subscribers from RevenueCat’s first_seen cohort for a given week and compare it with the acquisition spend incurred during that same week.

    However, when we look at the actualized revenue, it appears to be higher because there is a group of users that we never accounted for in our original predictions. These users did not become paid subscribers within their original first_seen cohort, so we effectively classified them as “missed users”  users we spent money to acquire but who did not convert within that cohort.

    When we aggregate our LTV predictions and backtest them against actual revenue, we therefore see that we are systematically missing these late converters.

    Given this, what would be the best way to incorporate late converters into our LTV predictions and CAC targets while still keeping our acquisition performance measurement accurate?

    Terms & ConditionsAccessibility statement