Last quarter I ran influencer campaigns with five creators. Combined reach: about 800,000 followers. Total purchases I could actually trace back to them: 23. That's not a typo. Eight hundred thousand followers, twenty-three purchases.
The thing is, one of those five creators drove 19 of those 23 purchases. She had 12,000 followers. The creator with 400,000 followers drove exactly one.
If I'd measured ROI the way most people do, by looking at views, likes, and "engagement rate," I'd have renewed the big account and dropped the small one. I'd have got it completely backwards.
Why the standard metrics are useless
The influencer marketing industry runs on metrics that don't connect to money. Reach, impressions, engagement rate, CPM. These numbers describe how many people saw something. They say nothing about whether anyone opened their wallet.
For mobile apps, this gap is even worse. Someone sees an influencer's post, taps a link, gets sent to the App Store, installs the app, opens it, uses it, and then maybe makes a purchase. That's a long chain. "Engagement rate" measures the first link in that chain and ignores everything after it.
I've seen influencer reports that proudly state "14,000 impressions at 4.2% engagement" as if that means something. What it actually means is about 588 people liked or commented. How many of them installed the app? How many purchased? The report doesn't say, because the tools they're using can't tell you.
What ROI actually means here
ROI for influencer marketing on a mobile app is straightforward arithmetic:
Revenue from attributed purchases minus what you paid the influencer, divided by what you paid.
If you paid a creator £200 and they drove 15 purchases at £2.49 each (£37.35 gross, roughly £31.75 after Apple's 15% cut), your ROI is negative. You lost about £168. That's useful information.
If another creator charged £50 and drove 30 purchases, netting you about £63.50 after Apple's cut, you made £13.50 profit. That's a 27% return. Not spectacular, but positive, and you've now got 30 new users who might tell their friends.
The hard part isn't the maths. It's getting the purchase attribution data in the first place.
The attribution problem
Here's where it gets tricky for apps specifically. When someone taps a link to the App Store and installs your app, there's a disconnect. The App Store doesn't pass referral data through to your app. Apple designed it that way on purpose, for privacy.
So you can't just stick a UTM parameter on a link and expect it to follow the user through the App Store and into your app. Web attribution works that way. App attribution doesn't.
You need something that bridges the gap. When a user taps the influencer's link, you store that context. When they later open your app, you match them back. When they make a purchase, you connect the dots.
This is what mobile measurement platforms (MMPs) like AppsFlyer and Adjust do. It's also what LinkOwl does, but without the enterprise pricing that makes those tools pointless for indie developers.
Setting up purchase-level attribution
The workflow with LinkOwl looks like this:
- Create a unique tracked link for each influencer
- The influencer uses that link in their bio, story, or post
- When someone taps it, LinkOwl records the click and redirects to the App Store
- Your app has the LinkOwl SDK, which handles attribution at install time
- When a purchase happens, your payment provider (RevenueCat, Superwall) fires a webhook to LinkOwl
- LinkOwl matches the purchase back to the original click
Now you have a direct line from "this influencer's link was tapped" to "this user bought the premium tier." That's the data you need for real ROI.
Running the numbers properly
Once you have purchase attribution, calculating ROI per influencer gets specific. Here's what to track:
Cost per purchase. You paid the influencer £150. They drove 8 purchases. That's £18.75 per purchase. If your app charges £2.49 one-time (netting you about £2.12 after Apple's cut), each purchase from this influencer cost you roughly 9x what it earned. Bad deal.
Break-even threshold. Work backwards from what you paid. If the deal was £150 and each purchase nets £2.12, you need 71 purchases to break even. Did the influencer's audience look likely to generate that? Probably not, in hindsight.
Revenue per follower. Divide total revenue attributed to the influencer by their follower count. This sounds crude but it's surprisingly useful for comparing creators. A nano-influencer generating £0.005 per follower will usually beat a macro-influencer generating £0.0001 per follower, and it helps you decide where to spend next time.
Payback period. Some influencer posts keep driving installs for weeks. Others spike on day one and die. Check your attribution data over 7, 14, and 30 days. A creator whose audience trickles in over a month has a different value than one who spikes and fades.
What I learned from actually measuring this
After three months of proper attribution tracking across about a dozen influencer deals, some patterns became obvious:
Nano-influencers (1K-10K followers) in a specific niche converted better than anyone with a general audience. Every time. The conversion rate difference was sometimes 10x.
Video posts outperformed static posts for driving installs, but the installs from static posts converted to purchases at a higher rate. My theory: video attracts casual browsers. Image/carousel posts attract people who read the caption and make a deliberate decision to try the app.
Paying a flat fee per post was almost always worse than offering a revenue share or per-install bonus. When creators have skin in the game, they make better content and they actually reply to comments asking about the app.
Re-posting worked surprisingly well. One creator posted about my app three times over two months. The third post drove more purchases than the first. Familiarity builds trust with their audience.
The tools question
For indie developers, enterprise MMPs don't make financial sense. AppsFlyer starts at several hundred dollars a month. Adjust is similar. Branch focuses more on deep linking than purchase attribution. These tools are built for companies spending thousands per month on user acquisition.
If you're running a few influencer deals at a time and your app makes less than £1,000 a month, you need something lighter. LinkOwl charges 5p per attributed purchase with no monthly fee. If an influencer drives zero purchases, you pay zero. The cost only exists when revenue exists.
Getting started
Measuring influencer ROI properly takes about 15 minutes of setup:
- Sign up at linkowl.app and add your app
- Drop the SDK into your project (Swift or React Native, about 3 lines of code)
- Hook up RevenueCat or Superwall via webhook
- Create tracked links, one per influencer
- Wait for data
After your first influencer campaign with proper tracking, you'll have actual numbers. Revenue per influencer, cost per purchase, ROI percentage. From there, decisions about who to rebook and how much to pay become obvious.
The influencer marketing industry has a measurement problem. Everyone talks about reach and engagement because those numbers are easy to get. Purchase attribution is harder to set up but it's the only number that tells you if you made money. Everything else is decoration.