Three Charts That Actually Tell You What's Happening

The problem with guessing at retention

When you're running an indie app, you make decisions on fragments of information. You know your overall retention number. You know your total installs. But you don't know which channels are feeding you users who stick around, and which are feeding you users who open the app once and vanish.

That's not a small thing. A 25% day-7 retention rate could mean you're shipping a product nobody wants, or it could mean you're burning budget on traffic sources that attract the wrong users entirely. Those are two completely different problems with completely different solutions.

The cohort + funnel + retention dashboard in Attribr exists because I kept hearing the same frustration from developers who had integrated the SDK: they could see where installs came from, but seeing retention trends across traffic sources required spreadsheets, manual lookups, or worse - spreadsheets nobody updated.

What the cohort chart actually shows you

Cohorts are just fancy buckets. You create a group of users (say, 'users who installed via Rippl on 15 January') and then you watch what happens to that group over time. Do 30% of them come back on day 2? Do 8% come back on day 30?

In Attribr's dashboard, you see this visualised. Each row is a cohort (grouped by install date or traffic source). Each column represents a day post-install. The colour gradient tells you the retention percentage at a glance. A week of dark reds means something broke and users stopped coming back. A week of steady ambers means your onboarding is working, but something falls off later.

What makes this useful for indie scale is that you're not drowning in features. You pick a time range, optionally filter by traffic source (organic, Rippl, a specific network), and you see the pattern immediately. No toggles for cohort size, no export-to-dashboard buttons, no four-layer menu. Just the chart.

The funnel chart: where users actually drop

Retention tells you if users come back. Funnels tell you where they stop coming back, or never arrive in the first place.

You might define it like this: install > app opens > user completes onboarding > user makes a purchase. Or: install > day 7 check-in > day 30 check-in. Whatever your app needs. The funnel chart shows you the percentage of users who make each step.

I built the funnel view into Attribr because retention alone is incomplete. You could have high 7-day retention but 0% conversion to purchase. That tells you something different than high 7-day retention and 30% conversion. One says your product is sticky but your monetisation is broken. The other says both are working.

More importantly, you can break this down by install source. Maybe your Rippl cohort converts at 22%, but your organic cohort converts at 8%. That's the kind of signal that changes where you spend next month's budget.

Retention charts: the long view

Day 7 and day 30 retention matter. They're the moments when real behaviour surfaces. A user who opens your app and never returns on day 7 isn't coming back. Period. A user who comes back on day 30 is part of the core.

The retention chart in Attribr's dashboard stacks these windows. You see day 7, day 14, and day 30 retention rates across your traffic sources in one place. If you're on the Growth or Pro plan, you also see fraud signals overlaid, so you know whether a retention drop is real user churn or a sign that a traffic source is sending bots.

What I hear from developers most often is relief. Relief that they're not managing retention in their head anymore, or pulling numbers from three different places to cross-reference. It's one chart. Specific. Actionable.

The moment it all connects

The three charts work together. You notice day-30 retention is dropping. You flip to the cohort view and see it's specifically the cohort from two weeks ago. You check the funnel and see that cohort has high day-7 retention but low conversion. That tells you something changed in your ad network's targeting or in your product's engagement between day 7 and day 30.

That's not coincidence. That's diagnosis. And diagnosis is only useful if the data is clean. Attribr's SDK handles install attribution across deterministic and probabilistic matching, works without requiring ATT permission on iOS 14.5+, and integrates in three lines of code. If your attribution is wrong, your dashboard is just a pretty lie. We built the SDK to be accurate first, because the charts are only as good as the data underneath them.