Why we built a 50KB attribution SDK (and what that means for your app)

The problem every indie developer faces

When you ship an app, you need to know three things: where did this install come from, is the user still around on day 7, and is the marketing spend actually working? These are not hard questions. But answering them with the existing tools means choosing between paying thousands per month or building your own tracking. Most people never build their own. They just guess.

The major attribution platforms serve enterprise teams. Their SDKs are heavy, their dashboards are overwhelming, and their pricing assumes you're spending six figures on user acquisition. For a studio shipping 10,000 or 20,000 installs a month, it feels absurd to invest like that. So you don't measure. You launch. You hope.

We wanted to break that pattern.

Starting small, staying small

The 50KB figure isn't marketing. It's a constraint we chose. We knew that if Attribr felt heavy, indie developers would skip it entirely. So we started with the minimum viable set of features: answer the three questions, nothing more. No session tracking. No event-level analytics. No graph database. Just install source, retention cohorts at days 7, 14, and 30, and a way to see which Rippl promoters drove which installs.

Zero third-party dependencies meant no surprise conflicts with your other tools. Sub-50ms launch overhead meant you'd never see a measurable slowdown in your app's startup time. This wasn't about being clever. It was about respecting your app's performance.

Three lines of code to integrate (Swift or Kotlin, your choice) because anything longer and developers simply wouldn't do it. We tested this. When the integration was five lines, some people skipped it. Three lines? Almost everyone got it working on the first try.

How attribution actually works without asking for permission

iOS 14.5 changed everything. ATT permission became the gate between you and knowing where installs came from. Most attribution platforms responded by building fingerprinting on top of probabilistic matching. Attribr works differently.

We use deterministic matching first: if the install came from a link you created in our dashboard, we know exactly where it came from because we built the link. No guessing, no probabilistic tricks. If it came from an ad network, we check their postback data. If it came from organic search or word of mouth, we use probabilistic signals to make a best guess about source. The combination is remarkably accurate without requiring ATT permission or building invasive fingerprints.

The Rippl bridge is where this gets interesting for community-driven growth. If your users came via a Rippl promoter (someone sharing your app within their community), Attribr ties that install back to them. You see exactly which promoter drove which users, and you can track whether those users stuck around. For studios using Rippl for community marketing, this is the missing piece every other platform forgot to build.

What happens on day 7, 14, and 30

Retention is where most indie developers stop measuring. An install count tells you nothing. A day 7 retention number tells you whether your game is actually engaging. We built the dashboard to show you cohorts straight away: how many people who installed on Monday are still active on that Monday a week later, two weeks later, a month later.

The charts are simple. Funnel view if you want to see how many users reach certain milestones. Retention curves to spot when people drop off. None of it requires a data science degree to understand. You look at the dashboard on launch day and you know if you've built something people want to use.

For the Pro plan, we added fraud signals and ad-network roll-up. Fraud signals flag installs that smell wrong (someone clicking an ad 500 times in 10 minutes). Ad-network roll-up means if you're running on multiple networks, you see them aggregated by source, not as seventeen separate line items. The Pro and Business tiers are built for people scaling beyond hobby projects, where that level of detail starts to matter.

Why size matters, even when nobody talks about it

Every kilobyte in your SDK is kilobytes not spent on content, gameplay, or features. Every millisecond of launch overhead is battery drain and user frustration. These things sound small until you ship to millions of devices.

We've seen teams using other attribution platforms where the SDK alone accounted for 3 to 5 MB of bundle size. They added it for attribution, but then they spent engineering time compressing assets and cutting features elsewhere. That's backwards. Your attribution tool should make your job easier, not harder.

The other constraint we built in: no bring-your-own-SDK middleman. You're not waiting for us to update a wrapper around some third-party service. We run the servers, we own the data flow, and that's why we can keep it genuinely lightweight. You integrate once and it just works.

Who this is actually for

We built Attribr for developers shipping between 1,000 and 100,000 installs per month. That's the sweet spot where you care about your metrics but you're not running a 50-person growth team. The Free plan covers 1,000 installs per month. If you ship a viral hit and hit 25,000 installs, Growth plan is £29 per month. If you're scaling to 100,000 monthly installs, Pro is £99. Beyond that, you talk to us.

We're not trying to be Adjust or AppsFlyer for enterprises. We're trying to be the thing those companies forgot about: the tool that works for everyone else. The developers who want to measure growth without spending like a Fortune 500 company.