Approval on Apple Ads is not a permanent state. It is a decision that can be revisited — and until this month, nobody outside Apple had a number for how often that happens.
On 13 August 2026, Apple filed its Digital Services Act transparency report for the App Store covering 1 January to 30 June 2026. Buried in it is the first hard figure anyone has published on App Store advertising enforcement: advertisements Apple removed after publication for breaching its advertising policies had already served 12,318,874 impressions in the European Union.
Those ads passed review. They ran. They spent budget. Then they were pulled.
That reframes something most acquisition teams treat as administrative. Ad review is a live dependency between budget and delivery, with a failure mode that arrives without warning on a surface where no comparable replacement inventory exists.
What Apple actually disclosed
A provenance note first, because it governs how confidently any of this can be quoted. Apple publishes its DSA transparency reports at a legal index on apple.com. Editions through August 2025 were HTML pages; the February and August 2026 editions are spreadsheet files only, so there is no current report page to quote directly. The figures below come from PPC Land and Noah Intelligence, both of which read the August 2026 spreadsheet and published on 23 August 2026, and which agree with each other on every figure we checked. They remain second-hand readings of a file rather than quotations from Apple.
| Metric | H1 2026 figure | What it covers |
|---|
| Impressions from ads removed after publication | 12,318,874 | EU App Store advertising only |
| Total enforcement measures | 11,761,465 | All App Store content, majority scam and fraud |
| Measures preceded by automated detection | 7,358,896 (62.6%) | All App Store content |
| Account terminations | 3,615,694 | Developers, Media Services customers and advertisers, combined |
| Content moderators | 687 (610 internal, 77 contracted) | All of App Store moderation, not ads-specific |
| Average monthly active EU recipients | 153 million | App Store |
Apple names four teams doing this work: App Review, Trust and Safety Operations, the Ad Certification Team, and the Intellectual Property Disputes Team. It states that “final decisions regarding content moderation are taken by humans” — worth holding alongside the 62.6% automated-detection figure. Automation initiates; humans confirm.
The scope is narrower than the headline suggests. This filing covers the App Store only; Apple News, Stocks, Maps and sports advertising surfaces are governed by a separate policy document and appear nowhere in it. That is a live gap, not a historical one: Apple Maps ads opened for booking in August 2026 with no enforcement disclosure of any kind.
→ WHAT THE FILING DOES NOT SAY
Every breakdown an advertiser would want is absent
The 12.3 million impressions are not split by advertiser, market, placement or ad format. The moderator headcount is not split out for the Ad Certification Team. Advertiser terminations are folded into the same bucket as developer and Media Services terminations, so they cannot be isolated. And the report says nothing about whether affected advertisers were credited for spend on ads Apple later decided should not have run.
The appeal numbers, and the asymmetry in them
The most useful numbers in the filing are the smallest ones, and almost nobody has picked them up.
828
Complaints on removal or visibility
32%
Of those outcomes changed
370
Complaints on account termination
Of 828 complaints about a removal or a visibility restriction, 236 produced a full reversal and 32 a partial one — 268 changed outcomes, roughly 32%. Of 370 complaints about an account suspension or termination, 8 were reversed: about 2.2%. Complaints about monetisation restrictions numbered zero, and so did reversals of them.
Read those rates next to each other and a policy posture becomes visible. Apple is willing to be argued out of a removal roughly a third of the time. It is almost never argued out of a termination. That should shape where escalation effort goes: contesting a pulled creative has a real hit rate, while contesting a closed account is close to a lottery. It also implies different evidentiary standards — a removal judges one piece of content, a termination judges a pattern.
One caution before anyone builds a process around 32%. These complaint counts are tiny against 11.7 million enforcement measures. Either the overwhelming majority of enforcement goes uncontested, or the complaint mechanism is not reaching the people affected by it. The filing gives no way to tell those apart.
What review looks like from the buyer’s seat
Apple’s published advertising policies are clearer than their reputation. The App Store policy sets out a prohibited list — hate, violence and harassment; profane or obscene content “including distasteful wording or phrases”; adult or graphic themes; tobacco and illegal drugs; guns, firearms, weapons and ammunition — and a restricted list where alcohol, sensitive imagery, pharmaceutical and medical content, real-money gambling, simulated gambling and dating are permitted subject to placement and region limits.
On process, Apple says: “Apple uses both manual and automated systems to determine if ad content is eligible to be displayed, on which placements, and/or which countries or regions… all determinations are reviewed by members of the content moderation team.” The separate video and audio policy is blunter still: “Apple reserves the right to modify these policies and reject, approve, or remove any ad content for any reason, at Apple’s sole discretion.”
What is missing from all of it is the operational layer.
| What you need to plan | What Apple publishes |
|---|
| Review turnaround time | Nothing. No SLA, no target, no queue estimate anywhere in the help documentation. |
| Appeal mechanism | No advertiser-facing appeal control. A policy support email address is the only documented channel. |
| Rejection reasons | Exposed in the Status column of the Ads dashboard and downloadable — per country or region, except app-icon rejections, which apply globally. |
| Remedy path | Fix the asset. For app icon, name or subtitle, that means shipping a corrected app version through App Store Connect, which triggers automatic re-review. |
| Post-approval stability | “Apple may also apply restrictions or limitations to your advertising.” No notice period stated. |
The Today tab documentation adds the sequencing detail most teams get wrong: “Today tab ads must be approved before they can run,” with status showing “Ad in review” until they are, and once approved the ad “will run right away unless you set its status to Pause.” Approval does not wait for your launch date.
Note also the asymmetry in scope. A rejected subtitle or screenshot is scoped to a country or region. A rejected app icon is global. That single line turns icon experimentation from a creative decision into a delivery risk across every storefront you buy in.
Apple already machine-reads rejection
Here is what undercuts any claim that ad review is inherently unknowable: Apple has already reduced a large part of it to enumerated, machine-readable codes exposed through the API.
The Apple Ads API documents an App Eligibility topic — “Check whether your app is eligible to promote in a campaign” — with a Find App Eligibility Records endpoint keyed on adam ID. Alongside it sits a topic called, unambiguously, Ad Rejection Reasons: “Review reasons for an ad rejection,” with endpoints to find creative rejection reasons and to fetch them by custom product page ID, plus objects including AppAsset, ProductPageReason and a data type called ReasonLevel, documented as “the level at which the system applies an ad rejection reason.” Apple’s description of the lifecycle is precise: “Ads require approval by Apple. While in review, your Ad is on hold. After approval, the AdServingStatus changes to RUNNING.”
The enumerated ProductPageReason codes tell you what the reviewer is actually looking at: app icon content and eligibility, app name language conflicts and content, subtitle language conflicts, subtitle content, subtitle pricing, offers or ranking claims, an app not being eligible “at this time,” a Product Page Optimization variant’s icon being ineligible, and a code named MIMICS_APP_STORE_TODAY_CARD — where Apple notes that “the phrases Game of the Day and App of the Day aren’t allowed.”
Every one of those codes describes a property of an asset you control, decided before delivery. None of them describes the case in the transparency filing — an ad that was approved, delivered, and then removed.
That is the gap worth naming. Pre-flight rejection is enumerable, inspectable and largely predictable. Post-publication removal is none of those things, and it is the one now quantified at 12.3 million impressions in six months in one region.
There is precedent for it operating at scale. Apple’s API release notes record that in Campaign Management API v4 release 49, “all previously rejected Today tab ad creatives were PAUSED and resubmitted for re-review” — a mass event advertisers had to notice and manually re-enable. If you are scoping the Platform API migration ahead of the January 2027 sunset, the rejection-reason endpoints belong in the read-path work, not the nice-to-have list.
The rejectable surface is about to get bigger
All of the above describes today’s account, where the reviewable creative surface is essentially your icon, name, subtitle, screenshots and custom product pages. That is about to expand considerably.
Apple announced on 8 June 2026 that advertisers will be able to design custom creative assets for Today tab and search results ads. The format ships alongside iOS 27 with a policy attached: every creative asset must meet a 4+ age rating regardless of the app’s own rating, with commercial prohibitions on pricing and discounts, website URLs, copyright symbols, unverifiable award claims, logos referencing competing platforms, and self-applied Apple recognitions such as Editor’s Choice or App of the Day.
Two structural changes from WWDC 2026 compound it. Assets now live in a centralised Asset Library in App Store Connect shared between your organic storefront and your campaigns, so a rejection is not contained to advertising. And assets can be submitted for App Review independently of an app update — good for iteration speed, and simultaneously far more review events per quarter than the old build-gated cadence produced.
Put those together with up to 70 custom product pages per app to attach assets to, and the number of independently rejectable objects in a mature account is heading for an order-of-magnitude increase. The apparatus described in the DSA filing does not obviously scale with it — and remember that 687 is the whole App Store moderation function, with no ads-specific share disclosed.
→ THE HONEST CAVEAT
No performance data exists for creative assets
The format has not shipped. Any claim you read that creative assets lift conversion rate by some percentage is extrapolated from adjacent formats or invented outright. What can be planned for now is the review workload and the rejection surface — not the lift.
Why an outage here costs more than elsewhere
Losing delivery on any channel is bad. Losing it on Apple Ads is a specific kind of bad, and the reason goes to why the channel is worth buying at all.
App Store search is where intent has already been declared. Apple’s own figures, published from data covering November 2024 through October 2025, put nearly 65% of downloads as happening directly after a search, with a conversion rate above 60% for ads at the top of search results. That is not a model inferring interest from behaviour elsewhere; it is a user typing what they want into the store they are about to buy in. It is why Apple Ads consistently buys higher-ARPU users than blended-inventory alternatives — the App Store audience monetises at a multiple of the alternative, and search placement intercepts declared intent rather than manufacturing it.
Which means Apple Ads inventory has no close substitute. If a Meta campaign is disapproved, the demand it was capturing is largely still addressable elsewhere at similar quality. If your top-of-search ad is pulled, the query is still being typed and the competitor in the second slot is still there — multi-slot search results have been global since the end of March 2026 — and there is no other way to be present at that moment. The cost of the interruption is not the paused spend. It is the downstream revenue delta between the users you would have acquired and the ones you acquired instead. On a channel whose entire thesis is user quality, that delta is the point.
Apple has little incentive to make any of this transparent. It discloses no advertising revenue line item; ads sit inside Services, which reported $30.739B in the June 2026 quarter, up 12% year over year, with CFO Kevan Parekh citing June-quarter records “in advertising, the App Store, Apple Care, music and video.” The only public sizing is estimate territory — eMarketer put US Apple ad revenues at $8.85B for 2026 in an April 2026 forecast, an estimate that should never be cited as a disclosure.
→ FREE 10-POINT AUDIT
Find out what happens to your account if an approval is reversed
We open your Apple Ads account, map every independently rejectable object across your campaigns and custom product pages, and show you where a single rejection would take delivery offline with no fallback. One hour. Senior strategist. No pitch.
Book my free audit →
Treating approval as a state, not an event
None of the above argues for paranoia. It argues for treating approval the way you would treat any external dependency with no SLA: monitor it, keep a fallback warm, and find out about changes from your own systems rather than from a revenue dip.
Practice 1 · MonitoringPoll serving status, do not eyeball the dashboard
The API exposes AdServingStatus. Anything that leaves RUNNING without someone on your side changing it should page a human the way a failed deploy would. It is a small piece of work that pays for itself the first time it fires.
Practice 2 · RedundancyNever let one asset carry a whole placement
Because rejection is scoped per asset — and globally for app icons — an account where every ad variation leans on the same hero screenshot has a single point of failure. Keep at least one approved variation per placement that shares no creative object with your primary.
Practice 3 · SequencingSubmit early, then pause deliberately
Approved ads begin serving immediately, so submit well ahead of launch and set status to Pause on approval. That decouples “cleared review” from “started spending,” and removes the temptation to submit late — which is what turns an unknown review window into a missed launch.
Practice 4 · EscalationContest removals; expect little from terminations
The 32% versus 2.2% split is the best available guide to where escalation earns a return. Keep the downloadable rejection reasons and a record of what changed and when, and raise policy questions promptly — a removal contested within days is a different conversation from one contested after a quarter.
Practice 5 · TerritoryAudit region-scoped rejections separately
Rejection reasons surface per country or region, so an ad healthy in your reporting aggregate can be dark in three storefronts. Reconcile serving status by territory rather than trusting a blended impression trend to reveal it.
What we still cannot know
The gaps here are large and mostly structural. We do not know how many advertisers those 12.3 million impressions represent, nor whether the ads ran for hours or months, nor whether any spend was credited. We do not know the Ad Certification Team’s size, throughput or turnaround, because the filing folds ad moderation into an App Store-wide headcount. And we do not know the base rate — how many ads were submitted, what share was rejected pre-flight, or how Apple’s post-publication removal rate compares to any other ad platform, none of which publish a comparable figure either.
The report is EU-only and App Store-only, so it says nothing about the US, nothing about Maps and nothing about Apple News. It also predates the creative assets format, making it a baseline taken immediately before the rejectable surface expands. No vendor fills the gap: MobileAction’s 2026 Apple Ads benchmark report and the wider 2026 benchmark set contain no approval-rate, rejection-rate or review-timeline data at all, and all describe 2025 or earlier auctions.
One earlier episode shows that moderation of paid placements is not hypothetical. In April 2026 the Tech Transparency Project reported that App Store search suggestions and sponsored ads were steering users toward face-swap and “nudify” apps, with roughly 40% of top results for several such queries returning them. Apple removed 15 apps the following day and gave six developers 14 days to comply. Paid placements were part of the finding, not an incidental detail.
Finally, an unresolved question of scope. Apple rewrote Section 6(a) of its advertising terms in July 2026, removing the word “Apple” from the description of where ads may be delivered and broadening “ad content” to include web pages and links. Whether that signals third-party inventory expansion remains speculation. If it happens, the content bar described here travels with it — and an enforcement disclosure currently scoped to one store in one region would need to travel with it too.
For how Apple’s controls and reporting compare with the other major iOS channel, see our comparison of Apple Ads and Google App Campaigns, where the same theme — what a platform lets you inspect versus what it decides for you — appears in a different form.