Velocity Growth animated logo

We can't make retention predictable

By
Mukund Kabra

Retention looks unpredictable when different kinds of users are averaged together. Split users into cohorts by how they arrived and what they did first, and each cohort usually behaves far more consistently than the average.

Category:
Measurement
Reading time:
2
min read
Published on:
October 24, 2026
Resources
right Created with Sketch.
Measurement
right Created with Sketch.
We can't make retention predictable

We can't make retention predictable

Problem

One month retention holds, the next month it drops, and nobody can explain why.

That makes every plan fragile. You cannot forecast revenue, set a sensible acquisition budget or know whether last month's fix worked.

So the team swings between quick win-back campaigns and big product bets, without knowing which one to back.

Insight

Retention is rarely random. It is usually a mix of user groups that behave differently.

Users who came from a promotion, a referral or a search ad stay for different reasons and at different rates. When they are blended into one number, the number moves whenever the mix moves.

Seen cohort by cohort, most of the noise disappears and the real drivers show up.

How we approach it

We separate the groups first, then work on the ones that matter most.

  1. We build cohorts by source, first action and first-month behavior.
  2. We find the early signals that separate users who stay from users who leave in each cohort.
  3. We design lifecycle messages and product nudges for the cohorts with the most value at stake.

Retention becomes a set of numbers you can forecast and improve, cohort by cohort.

Tags:  
Growth Hiring Strategy; Organizational Design; Growth Talent; Team Scaling; Performance Alignment; Business Strategy
left Created with Sketch.
Previous
Next
right Created with Sketch.

Ready to scale profitably?

Let's discuss how to unlock sustainable growth without sacrificing unit economics.

Speak to Velocity
{"@context":"https://schema.org","@graph":[{"@type":"BlogPosting","headline":"We can't make retention predictable","description":"Why retention swings from month to month, and how cohort analysis makes it something you can forecast.","author":{"@type":"Person","name":"Mukund Kabra","url":"https://www.velocitygrowth.ae/about"},"publisher":{"@type":"Organization","name":"Velocity Growth","url":"https://www.velocitygrowth.ae","logo":{"@type":"ImageObject","url":"https://www.velocitygrowth.ae/vg-icon.png"}},"datePublished":"2026-01-24T04:08:22.575Z","dateModified":"2026-10-23T13:55:14.402Z","url":"https://www.velocitygrowth.ae/problems/cant-make-retention-predictable","mainEntityOfPage":{"@type":"WebPage","@id":"https://www.velocitygrowth.ae/problems/cant-make-retention-predictable"},"keywords":"growth hiring strategy, organizational design; growth talent; team scaling; performance alignment; business strategy","inLanguage":"en"},{"@type":"WebPage","name":"We can't make retention predictable","url":"https://www.velocitygrowth.ae/problems/cant-make-retention-predictable","speakable":{"@type":"SpeakableSpecification","cssSelector":["h1",".post-body > p:first-of-type"]}}]}