Apple Ads: relevance gates you in, but it doesn’t keep you on top (ConsultMyApp)
A practical summary of ConsultMyApp’s Apple Ads auction analysis: semantic relevance looks like an eligibility filter, then bids and predicted performance reshuffle positions. The takeaway is not ‘bid harder’, it’s ‘structure discovery and negatives so you’re not paying for nonsense’.
Original post (source): ConsultMyApp - “How Does the Apple Ads Auction Actually Work?” (March 10, 2026)
ConsultMyApp ran a neat, data-driven attempt at answering the question Apple never really spells out:
Once you’re in the Apple Ads auction, how much does relevance actually decide your position?
They used an Apple Ads auction dataset (UK), then applied a semantic relevance scoring approach (their goal: “would a human expect this app to solve this need?”) to compare relevance order vs actual ad position.
The one-line lesson
Treat relevance like a gate, not the ranking. Once you clear the gate, bid and predicted performance do most of the ordering.
What stood out
- The most relevant app didn’t win #1 most of the time. Their headline stat is that the “most relevant” app landed in slot #1 in only ~44% of keywords in their sample.
- Search Match can be extremely permissive. Their examples show ads showing up for keywords that are hard to justify by intent alone, which matches what most practitioners see when they leave discovery too open.
- Broad and Search Match are discovery tools, not steady-state acquisition. Apple’s own docs say this, but the data makes it feel more real: discovery expands the candidate set, and then the auction can push weird winners to the top.
Why this matters (practically)
If you treat Apple Ads as “relevance wins”, you’ll misdiagnose problems:
- you’ll call bad traffic a “creative problem” (when it’s actually a discovery leakage problem)
- you’ll call CPC inflation “competition” (when it’s sometimes conquest + permissive expansion)
- you’ll over-index on metadata tweaks for terms where the issue is simply bidding/structure
Tiny win (30-60 minutes)
- Pull your last 14 days of search terms for Search Match and broad.
- Tag each query as: core intent, adjacent intent, or nonsense.
- Move the proven core terms into exact match campaigns, then negate them back out of discovery.
That one loop usually saves more budget than any “rewrite the subtitle” sprint.
Want help with ASO?
If you want this implemented for your app, check out our services - or run your workflow in APPlyzer.