Analytics for React Native: from install attribution to retention

Web analytics assumes a page and a URL. Mobile has neither — it has installs, sessions, screen views, and the awkward gap between an ad click in a browser and a first app open. Here’s what to measure in a React Native app, and why install attribution is the part most tools get wrong.

Why mobile analytics is different

On the web, the journey is continuous: click, land, browse. On mobile, the user taps an ad in one app, gets sent to a store, installs, and opens your app minutes later — a hard break that erases the usual referrer. Screen views replace page views, and retention matters more than any single session.

Install attribution: the hardest part

Attribution answers a simple question — which campaign drove this install? — that’s deceptively hard on mobile because of that store-to-app gap. Doing it deterministically (a real match, not a probabilistic guess) is what separates real attribution from an estimate.

Android referrer vs iOS pasteboard

The two platforms need different mechanisms, and Nohmo handles both in the SDK:

• Android — the click ID rides along in the Play Store install referrer, which Google Play delivers on first open. A 100% deterministic match, with no GAID or fingerprinting.
• iOS — the App Store has no referrer, so a brief interstitial writes the click ID to the system pasteboard; the SDK reads and clears it on first open, with GAID/IDFA and IP matching as fallbacks.

Retention and uninstalls

Acquisition is only half the story. What you really want to know is whether people come back. Nohmo tracks D1, D7, and D30 retention, detects uninstalls, and reports reinstalls — so you can see not just how many installed, but how many stayed.

One SDK, one package

The React Native SDK ships in the same nohmo package as the web SDK, with a Babel plugin that auto-captures taps and screen views — no per-screen wiring. Install attribution, retention, crash capture, and custom events all come from one integration, for iOS and Android alike.