The ATT Problem We Actually Solved

The Consent Problem Isn't Going Away

Here's what happened: attribution networks like Branch and AppsFlyer had built their entire systems on IDFA, Apple's identifier for advertising. When ATT arrived, those systems either went blind or asked users for permission. The ask, of course, tanked opt-in rates. Most developers saw 10 to 15 percent of users grant permission. That meant 85 percent of your install data was just... gone.

We spent weeks talking to indie developers about this. The conversations were the same every time. "I can't spend £99 a month on attribution if I can only see 15 percent of my data." Or worse: "I'm not even going to bother measuring anything." That second group frightened me more than the first.

Attribution without consent wasn't a feature request. It was survival.

Deterministic Matching Gets You Part of the Way

We started with what we knew: deterministic attribution. If you use Apple's install validation APIs alongside SKAdNetwork and basic impression logging, you can map a good chunk of your installs back to their source without touching IDFA or asking for permission. Apple actually encourages this.

The problem is deterministic alone isn't enough. You lose install source visibility on organic traffic, SDK-less campaigns, and anything that doesn't fire through a measurable ad network. For indie developers doing performance marketing through networks like Rippl or running their own landing pages, that blindspot was crippling.

So we layered in probabilistic matching. Not fingerprinting in the old sense. Instead, we take the signals you can legitimately collect (carrier, timestamp, language, device type, regional locale, campaign impression data, user behaviour patterns) and run them through statistical models to infer install source with reasonable confidence. Combined with deterministic data, you get coverage.

Why We Didn't Go the Fingerprinting Route

I'll be honest about the road not taken. We could have gone hard on probabilistic fingerprinting. Collect more signals. Build a denser model. Some vendors have done this. But the more signals you collect, the closer you get to fingerprinting users without consent, which violates the spirit of iOS 14.5, even if it doesn't technically break the law.

That felt like a short-term gain for long-term pain. Apple is only going to tighten this. The iOS trajectory is clear: less data, fewer signals, more privacy by default. Building a business on fingerprinting felt like building on sand.

Instead, we built Attribr to work with the constraints as they actually exist. Minimal data collection. Open source methodology so developers can audit what we're doing. And honesty about confidence levels. Our dashboard tells you which installs are high-confidence deterministic, which are probabilistic, and how much data you're actually seeing. No false certainty.

The Rippl Connection Changed Everything

Here's where the architecture got interesting. Rippl, our community-driven performance marketing platform, already had a direct integration point with developers. When a Rippl promoter drives an install, we know that install happened because of that specific promoter. There's no guesswork.

We built Attribr to sit directly on top of that channel. If a user came from a Rippl promoter, we have a direct line of sight. If they came from elsewhere, we use deterministic and probabilistic matching to make the best inference we can. The result is a three-layer system: direct attribution where we can prove it, deterministic attribution where Apple's APIs give us visibility, and probabilistic inference for everything else.

For indie developers on Rippl, that third layer closes the loop. You're not paying enterprise pricing for attribution you'll never fully see. You're paying £29 a month for data you can actually act on.

What This Means in Practice

An indie developer integrated Attribr last month. Took three lines of Swift code. After a week, they had visibility into where 88 percent of their installs were coming from. Not the 15 percent they'd get with ATT consent. Eighty-eight. They sent us a message: "This is the first time I've actually understood my own user acquisition since iOS 14.5 landed."

That's the specificity we were after. Not enterprise-grade attribution at indie-friendly prices. Something actually purpose-built for the constraints you're actually facing. A 50KB SDK. Zero third-party dependencies, which means no fingerprinting libraries hiding in your dependency tree. Sub-50ms launch overhead, because we're not collecting half the internet on startup.

The dashboard shows you cohort retention at day 7, 14, and 30. You see which install sources keep users around and which don't. You can spot a user acquisition channel that drives high-quality installs versus one that drives installs that disappear in 48 hours. And if you're using Rippl, you can trace specific installs directly to the promoter who drove them.

The Honest Part

I need to be clear about what Attribr is not. We're not trying to replace AppsFlyer or Adjust. Those tools do things we don't do. They've got deeper fraud detection. They've got integrations into hundreds of ad networks. They're built for teams spending six figures a month on user acquisition.

Attribr is built for the developers who can't afford those tools. The studios making £100,000 a month in revenue, trying to understand where the next installs should come from. The solo developer who shipped a game on the side and suddenly it's making money. The small team that doesn't have a growth budget but wants to grow smart.

For that scale, the ATT reality is different. You need attribution that works without consent because most of your users won't give it. You need something lightweight that doesn't drag down your app launch. You need pricing that fits a small team's budget. And you need to understand your retention, because low-quality installs are worse than no installs.

We built Attribr for that reality. Not around the future we wish we had. Around the iOS 14.5+ world we're actually living in.