You're running ads, sending emails and posting links, and installs are coming in — but which link drove which install, and did those users convert? That's mobile attribution. Here's how it works and how to set it up without adopting a heavy measurement SDK.
The attribution chain
Attribution connects four things: click → install → first open → in-app events (sign-up, purchase). The goal is to carry the source (campaign, channel, ad click id) all the way from the first tap to revenue, so you can see which links and campaigns actually pay off.
Deterministic vs probabilistic matching
- Deterministic: the same link/click id is present at click and at first open (e.g. via deferred deep linking). Most accurate.
- Probabilistic: when a deterministic id isn't available, match on signals like IP + platform within a short window. Coarser, privacy-friendlier.
iOS: SKAdNetwork
On iOS, Apple's SKAdNetwork (SKAN) reports a coarse conversion value by campaign instead of per-user data. You map in-app events to a conversion value (0–63) and read installs back through SKAN postbacks — the privacy-preserving way to measure paid iOS installs.
Ad networks: click IDs + postbacks
Paid campaigns attach a click id (gclid for Google, fbclid for Meta). Capture it at click time, and when an attributed install or purchase happens, send a server-to-server postback to the network so its optimizer learns. Add UTM parameters for your own analytics on top.
How to do it without a heavy SDK
lynkily captures UTMs and click ids at click time, does deferred deep linking, and exposes a lightweight REST attribution API: your app calls /api/v1/attribution/init on first open to record an attributed install, and /api/v1/events to record sign-ups, purchases and revenue — which get tied back to the original link and campaign. lynkily can forward attributed events to Google Ads and Meta (Conversions API) and return the SKAN conversion value to set on iOS. It's REST-based rather than a large native SDK (native SDKs are on the roadmap).
// after a purchase
fetch('https://lynkily.com/api/v1/events', {
method: 'POST', headers: {'Content-Type':'application/json'},
body: JSON.stringify({ app:'yourapp', name:'purchase', value: 9.99, currency:'USD' })
});
Do you need a full MMP?
If you run heavy paid UA across many networks with fraud concerns, a full MMP may be worth it — see do you need an MMP?. If you mainly need to know which links and campaigns drive installs and conversions, a REST-based approach is simpler and cheaper. Build your campaign URLs with the free UTM builder.