The Install Question iOS 14.5 Made Harder (and How We Solved It)
In spring 2021, I watched indie developers panic. Apple's privacy update meant no more IDFA by default, and the tracking workarounds everyone relied on vanished overnight. One developer emailed me asking a simple question: if I can't track installs the old way, how do I know if my marketing actually works? That question shaped everything we built into Attribr.
The problem was real, and most solutions were expensive
The standard answer was always Branch, AppsFlyer, or Adjust. Those platforms are powerful, but they come with three catches. First, they cost more than many indie studios can justify. Second, they're built for enterprise campaigns with massive budgets to absorb the complexity. Third, they're so feature-rich that you end up paying for functionality you don't need.
We watched indie developers choose between two bad options: either skip attribution entirely and guess which channels were driving installs, or swallow the cost and integrate a heavyweight SDK that felt like overkill for an app hitting a few thousand installs a month. Neither felt right.
The real problem wasn't that attribution was impossible after iOS 14.5. It was that everyone had built their solutions around IDFA, and when Apple removed the easy path, nobody wanted to explain the harder one.
Deterministic matching tells you the install source directly
Here's the core of how we answer the question without asking for ATT permission. When someone clicks an ad or promo link, they land on a page or email with a unique parameter attached. That parameter travels with them into the app. Attribr reads it during launch, matches it against known sources, and tells you exactly where that install came from. No guesswork. No probability calculations. Just a clean signal.
We call this deterministic matching, and it's the fastest, most reliable method available. It works with every major ad network, with your own landing pages, with email campaigns, with Rippl promoters running performance marketing. The SDK is only 50KB, so it doesn't bloat your app size. It adds less than 50 milliseconds to launch time. Three lines of Swift or Kotlin to integrate it. Most developers have it working in an afternoon.
The catch is that deterministic matching only works when a parameter actually makes the journey. If someone sees your app in the store and installs it directly, there's no parameter. That's where the probabilistic piece enters the picture.
Probabilistic matching fills the gaps without fingerprinting
Probabilistic matching sounds mysterious until you understand what it actually does. When a deterministic match fails, we look at the install timestamp, the device locale, the app version, and the ad network's own server-to-server signals. We cross-reference those against recent campaigns and ad spend. We don't build a device fingerprint or track users across apps. We simply answer the question: given this install's metadata, which campaign is most likely the source?
It's not perfect. Probabilistic matching carries more uncertainty than a clean parameter. But combined with deterministic results, it gives you a much clearer picture than blind guessing. And crucially, it works without ATT permission. You're not asking users for consent. You're not storing identifiers that Apple's privacy labels would flag.
The two methods work together. Deterministic handles your precision installs. Probabilistic catches the rest and narrows down the noise. You're left with a cohort of installs you can actually trust.
Retention tracking answers the question that attribution skips
Here's what most attribution tools miss. They tell you an install happened. They tell you the source. But they don't tell you if that user is still active a week later. For indie developers and small studios, that silence is deafening. An install from a cheap traffic source isn't a win if the user churns on day two.
Attribr tracks retention at 7, 14, and 30 days. When a user opens your app within those windows, we log it against their original install source. Now you can answer the real question: which sources are driving installs that actually stick around?
We surface this in the dashboard with cohort and retention charts. You see your install sources ranked not by volume but by how many users from each source came back. It changes how you think about marketing. A campaign that drove 500 installs but a 20% day-7 retention rate might be worth less than a campaign that drove 100 installs with 60% retention. Most attribution platforms leave that calculation to you. We do it automatically, because we know that's what matters to people running small operations.
The Rippl connection is something you won't find elsewhere
One of the stranger discoveries we made while building Attribr was that indie developers using Rippl (a community-driven performance marketing network) had zero insight into which individual promoters were driving their installs. Rippl matches app creators with marketers who run paid campaigns in exchange for a cut of the install revenue. It's a smart system. But Rippl's reporting was silent on the promoter level.
We built Attribr with a direct bridge to Rippl's infrastructure. Now, when a Rippl promoter drives an install, Attribr connects that install back to the specific person who ran the campaign. You see your Rippl revenue and cost per install broken down by promoter. If one marketer's installs have genuinely better retention than another's, you know it immediately. You can double down on what works and adjust where it doesn't.
This integration exists nowhere else, because we built Attribr with the indie and small-studio reality in mind from the start. We weren't trying to replicate enterprise tools. We were trying to answer the questions that actually matter to people bootstrapping apps with performance marketing budgets, not corporate campaign spends.
The numbers stay real because the SDK stays simple
I could list features, but what matters is this: Attribr starts free for up to 1,000 installs a month. Growth plan runs £29 a month for up to 25,000 installs. Pro is £99 for 100,000 installs. We don't charge by user tracking or by data events. We charge by install volume, because that's what developers actually care about forecasting.
The simplicity isn't accidental. When a tool has zero third-party dependencies, weighs 50KB, and adds sub-50ms overhead, it means we're not running hidden background processes or batching data to third-party servers. Your privacy labels stay clean. Your app size stays lean. Your launch time stays fast. Those things matter more than marketing language does.
We've spent the past two years listening to indie developers ask how other attribution tools work. What they usually mean is: where's the catch? With enterprise tools, the catch is usually complexity you don't need or pricing that scales beyond reason. We decided the catch should be none at all. You integrate it, you get your answers, you move on.
iOS 14.5 didn't kill attribution. It killed the easy way. If you've been skipping it because the standard tools felt too heavy or too expensive, or if you've been guessing at marketing performance because you didn't think there was a real alternative, what's actually stopping you from finding out which channels are working?