Which UA campaigns are cannibalising organic installs?

What cannibalisation means

Cannibalisation is when a paid campaign takes credit for installs that would have happened organically, so attributed installs go up while organic installs go down by about the same amount.

Cannibalisation: paid attribution capturing users who were already going to install. The installs are real; the campaign just did not cause them.

It is not fraud and it is not an attribution bug. Last-touch attribution credits the last ad touched, and people who are already looking for your app touch a lot of ads on the way. This article on when paid installs cannibalise organic growth describes the pattern from the attribution side.

The usual suspects

Brand search

Someone searches the App Store or Google for your app by name. Your paid result sits above your organic listing; they tap it. The ad gets the install, but the search itself shows the person already wanted the app. The three users walkthrough on the start page follows one such person, Ana, through her week.

• High attributed conversion rates and very low CPIs are typical, which makes brand campaigns look like the best channel on the dashboard.
• The open question is what a competitor bidding on your name would take if you stopped. That is worth testing rather than assuming.

Retargeting

Retargeting: showing ads to people who have already visited, installed or used the app. This glossary entry on retargeting covers the basics.

Retargeting audiences are, by design, people most likely to act anyway. Re-engagement and reinstall campaigns often report the highest ROAS of any channel for that reason. Some of it is real (lapsed users nudged back); some is credit for users who would have returned.

Broad campaigns during busy weeks

During festivals, sales or a press moment, organic interest rises. Prospecting campaigns running at the same time collect credit for part of that rise, and they look unusually efficient.

How do you identify which campaigns are cannibalising organic?

A four-step check you can run on your own data, before any formal test:

1. Plot total, organic and attributed installs by day for the last three to six months, with campaign start and stop dates marked.
2. Look for mirror images. Organic dipping in the same week a campaign starts, by a similar amount to the campaign's attributed installs, is the classic sign. The cannibalisation lab draws exactly this shape on sample data.
3. Check total installs. If paid rose by 1,000 a day and total installs rose by 200, roughly 800 of the paid installs were relabelled organics. If total rose by 900, the campaign is mostly incremental.
4. Rank the suspects. Brand terms first, then retargeting, then anything that started at a seasonal peak.

This check needs one data source that sees organic and paid installs on the same basis. An MMP such as Linkrunner, AppsFlyer or Adjust shows organic and attributed installs side by side by day, which is the view to start from.

Sample data: a made-up app to show the method, not a benchmark.

A made-up example: a weekday organic baseline of 2,000 installs. A push is credited with 1,500 installs a day, but only 40% of those are caused by it. The dashboard shows organic falling to 1,100 a day (a 45% drop) while total installs rise by only 600 a day (30%).

Confirm with a test

The patterns above are clues. Seasonality, store featuring or a competitor's move can produce the same shapes, so confirm before cutting spend.

• Brand search: pause brand terms in some countries or regions for two to four weeks; keep them running elsewhere. Compare total installs from search, not paid installs.
• Retargeting: use a holdout on the retargeting audience if the platform supports it; otherwise pause in matched regions.
• Read the test on total installs. The lift test reader handles holdout results; the test types guide covers geo set-up.

What to do with the answer

• High cannibalisation, low competitor pressure: cut or cap the campaign and watch organic recover.
• High cannibalisation, competitors bidding on your name: keep a defensive budget, but judge it on incremental installs, not attributed CPI.
• Low cannibalisation: the campaign is earning its credit. Check whether it can scale before the incremental share starts to fall.

For a reference point on how large the organic share typically is for apps of your size, AttributionBenchmarks has a short read on what a good organic install share looks like. The acting on results guide turns the outcome into a budget move.