Behavioral Cohorts and Retention Grids in Vexo

An earlier post covered Vexo's overall retention widget — a single curve reporting how many users came back on Day 1, 7, 14, and 30. That one number is a great health check, but it flattens everyone into one line. It can't tell you whether last week's cohort retains better than the one before it, or whether users who hit your aha moment stick around longer than users who never did. The retention grid answers exactly those questions.

Everything below is live today, and it builds directly on the shared filter context from the segments post — no new setup beyond the SDK you already have.

The retention grid: cohorts down, periods across

A retention curve is one line. A retention grid is a table. Each row is a cohort — a group of users bucketed by when they first showed up — and each column is a period measured from that cohort's start. The cell where they meet is the percentage of that cohort still active in that period. Read a row left to right and you watch one cohort decay over time; read a column top to bottom and you compare the same period across cohorts.

Every row starts with the cohort's label and its size — how many users it contains — followed by the period cells. Period 0 is the cohort itself, so it's always 100% and always the darkest cell in the row; each cell after it shades lighter as the return rate drops. Hover any cell and Vexo tells you the raw count behind the percentage — how many of the cohort returned in that period. The color isn't decoration: it's a heatmap, so a healthy product shows deep color holding across the columns and a leaky one fades to pale within a period or two.

Daily, weekly, or monthly

One switch changes the time base of the whole grid. Vexo ships three granularities, each with a period window tuned to how far that lens is worth reading:

The column headers follow the granularity — D0, D1, D2… in daily, W0, W1… in weekly,M0, M1… in monthly — so it's always clear what one step across the grid means.

Behavioral cohorts: group by what users did

By default the grid's cohorts are time-based: everyone who first appeared in a given day, week, or month. That's the classic model. But the more useful question is often behavioral — do the users who did X retain better than everyone else? That's what the cohort event field is for.

Type a custom event name into the box above the grid and the cohorts change from “all active users” to “users who did that event.” The rows still bucket by time and the periods still measure return rate — but now every number describes the population that took the action you care about. Leave the field empty and you're back to retention across all active users. It's one input, and it converts the grid from “who came back?” into “do the people who completed_checkout come back more than the people who didn't?”

import { customEvent } from 'vexo-analytics';

// The events you already track are the cohorts you can slice by.
customEvent('completed_checkout', { plan: 'pro' });

There's nothing new to instrument. Any custom event you already send is available as a cohort definition — the grid reads from the same event stream that powers your funnels and trends.

It respects the filters you already set

Like the rest of Vexo's product-analytics views, the retention grid reads from the global filter context — date range, app version, operating system, country, and your custom-event filters. Apply a saved segment and every cohort in the grid narrows to that population at once. So you can layer the two: a behavioral cohort (“users who finished onboarding”) inside a segment (“on the version we shipped last week, in the US”), and read how that group retains week over week — without rebuilding the audience.

When there isn't enough history yet, the grid says so plainly: cohorts need at least one full period — a full day, week, or month depending on the granularity — before there's anything to show.

How to actually read it

1. Scan down a column to see whether a given period's retention is improving cohort after cohort. This is your release scorecard.
2. Scan across a row to see how fast a single cohort decays — where the color falls off is where you lose people.
3. Set a cohort event to test a hypothesis: pick the action you think drives retention and see if its cohorts hold color longer than the all-users baseline.
4. Apply a segment to ask the same questions of one slice — a version, a country, an acquisition source — instead of the whole user base.

Getting started

The retention grid is live in your app's analytics today. If you're already sending events with the Vexo React Native SDK, open the retention grid, pick a granularity, and — when you want a behavioral cut — type in a cohort event. New to Vexo? Create a free account and connect your app in a few minutes; see the docs for the full setup.

Frequently asked questions