SSubida Skills

Esc to close · results from all notes, skills and people

Mobile apps›User acquisition

Apple Ads measurement pitfalls — AdServices isolation, match rate, storefront country, CPP comparisons

Thomas Petit's list of ways Apple Ads numbers mislead — AdServices attribution can't be compared with SKAN-measured channels, MMP-vs-Apple install "match rate" (80–90% is good), App Store storefront ≠ user location (up to ~30% non-US installs on US campaigns), and custom product page results that look 10× better only because they serve a different audience.

Updated 2026-10-24·1 source·user-acquisitionapple-adsattributionadservicescustom-product-pagesmeasurement

Apple Ads measurement pitfalls

TL;DR

  • AdServices (Apple’s own attribution) doesn’t rely on ATT consent or SKAN. That makes it useful, but Apple Ads results “live in isolation”. Never compare them with SKAN- or fingerprint-measured channels, and never blend in brand.12
  • Track match rate = MMP-attributed installs ÷ Apple-reported installs. 80–90% is doing great. Monitor it, because it shifts, and on non-search placements it looked very different.3
  • Country = App Store storefront, not IP location. On US campaigns, users from Saudi Arabia, Iraq and Nigeria using the US store can be ~30% of installs, bought at US prices.4
  • CPP vs default page comparisons are biased. CPPs served only on newer iOS versions reach a different audience. “CPP spends 10× more at half the CPA” isn’t a valid win. Compare CPPs with each other, before/after.56
  • Apple’s headline stats (“65% of downloads follow a search”, 70% → 43% of visitors search) are sales points, mostly driven by big-brand searches. Don’t plan around them.7

Details

Market-share illusions. A 2022 Branch/Financial Times figure claimed Apple went from 5% to 58% of installs after ATT. That was only attributed installs: other channels moved to unattributed SKAN. Petit’s estimate: ~$8–10B Apple search ad revenue vs ~$100–150B total app ad spend, so under 10%.2

Non-search placements (Today tab, search tab, product pages): don’t buy them for direct-return performance, and be careful how you measure them. Some advertisers (Glovo, Google) spend heavily there for other reasons.8

Interface gaps (2023): no post-install data in Apple’s UI, even though Apple has purchases and retention in App Store Connect. Third-party Apple Ads tools (faster UI, MMP integration, configurable graphs) pay off for big or multi-app accounts. Small single-app accounts can live with Apple’s UI. SplitMetrics’ extension shows search popularity, which is exponential, so 38 vs 47 matters.910

Disagreements / caveats

  • Adapty (2026) markets user-level deterministic Apple Ads attribution and says Apple Ads users show higher LTV than other channels (Optimising Apple Ads keywords for revenue, not installs). Petit’s isolation point says to read such cross-channel comparisons cautiously.
  • Talk recorded July 2023. Apple has since renamed Search Ads to Apple Ads and added placements; specifics (iOS version gates, UI) may have changed.

See also

Footnotes

  1. Hard truth about Apple Search Ads | Thomas Petit | ASO Conference 2023 @ 25:49 — Thomas Petit ↩

  2. Hard truth about Apple Search Ads | Thomas Petit | ASO Conference 2023 @ 27:20 — Thomas Petit ↩ ↩2

  3. Hard truth about Apple Search Ads | Thomas Petit | ASO Conference 2023 @ 28:50 — Thomas Petit ↩

  4. Hard truth about Apple Search Ads | Thomas Petit | ASO Conference 2023 @ 30:21 — Thomas Petit ↩

  5. Hard truth about Apple Search Ads | Thomas Petit | ASO Conference 2023 @ 18:40 — Thomas Petit ↩

  6. Hard truth about Apple Search Ads | Thomas Petit | ASO Conference 2023 @ 19:35 — Thomas Petit ↩

  7. Hard truth about Apple Search Ads | Thomas Petit | ASO Conference 2023 @ 02:25 — Thomas Petit ↩

  8. Hard truth about Apple Search Ads | Thomas Petit | ASO Conference 2023 @ 17:05 — Thomas Petit ↩

  9. Hard truth about Apple Search Ads | Thomas Petit | ASO Conference 2023 @ 05:37 — Thomas Petit ↩

  10. Hard truth about Apple Search Ads | Thomas Petit | ASO Conference 2023 @ 07:27 — Thomas Petit ↩