An Apple Ads campaign that will not spend is not a budget problem. It is a supply problem in the one place iOS intent is purchasable.
Why underspend here costs more than underspend anywhere else
Most channels have substitutes. If your Meta budget does not clear, the impressions were going to be interruptive anyway and the same user is reachable tomorrow at roughly the same price. App Store search results have no substitute. The inventory is a person who has already typed a query, on the storefront where consumer spending per download runs several times Google Play’s, and Apple reports more than a 60% average conversion rate for ads at the top of search results on data covering November 2024 through October 2025.
That is the whole reason Apple Ads sits at the top of the iOS acquisition stack. It is also why a quiet campaign is expensive in a way a quiet Meta campaign is not: the demand you failed to meet does not roll forward. Somebody else served that query, in a moment of declared intent, to a user who converts at better than one in two. There is no make-good.
So the first thing to reject is the framing that underspend is a budgeting inconvenience. In a channel where the price of a tap is set by an auction over scarce intent, unspent budget is a measurement failure with a revenue cost attached. The rest of this piece is about finding which failure you have.
What Apple actually documents
Apple’s complete published guidance on low delivery lives on a single Help page titled Tips for solving performance issues. Its remedies, in Apple’s own words: wait 24 to 48 hours for performance data to gather; note that the dashboard defaults to the last seven days and excludes the current day; review your app metadata in App Store Connect; keep the default audience settings to reach all eligible users; expand reach to new countries and regions; use multiple ad placements; increase your maximum cost-per-tap bid; and check your cost-per-acquisition cap to avoid missing impressions.
Read that last one again. The CPA cap is the control that Maximize Conversions replaced when Apple rolled the new bidding strategy out to all App Store advertisers in late February 2026. Apple’s troubleshooting page still tells advertisers to check it. That is not a gotcha; it is a signal about how much weight this documentation can carry. Treat it as a checklist of hygiene items, not as an explanatory model.
→ A SOURCING PROBLEM WORTH KNOWING
Not one Apple Ads Help page carries a date
Every page under ads.apple.com/app-store/help and /best-practices renders without a publication or last-updated timestamp. There is no changelog. It is therefore impossible to establish from the public record whether a given Apple page predates or postdates the February 2026 bidding change or the March 2026 inventory change. Every Apple citation in this piece is by access date: 26 August 2026.
The daily budget is a monthly cap
The single most repeated wrong diagnosis in this category is that a campaign burned its daily budget by lunchtime and stopped serving. Apple’s glossary says the opposite, plainly: a campaign’s monthly spend will not be more than the daily budget times the average number of days in a calendar month, calculated as 30.4, and on days with opportunities to get more downloads a campaign’s spend may exceed the daily budget.
So the daily budget is a pacing input to a monthly ceiling, not an intraday shutter. A campaign that spent 180% of its daily budget on a Tuesday has not malfunctioned, and a campaign that spent 40% on a Wednesday has not necessarily been throttled. The observable quantity that matters is the trailing 30-day total against budget × 30.4.
This matters because at least one widely read vendor guide, updated as recently as August 2026, still tells advertisers that relying solely on daily budget caps leads to budget exhaustion early in the day and causes ads to stop running. That contradicts Apple’s own documentation. When a vendor guide and Apple disagree on a mechanic, the mechanic is the thing to test, and the vendor is the thing to discount.
Account-level stops: the enumerated list
Apple documents one closed list of causes that move a campaign to On Hold. It is short, and it is the only part of this whole subject where a definitive answer exists.
| Stop | Where it lives | Time to fix |
|---|
| Campaign end date in the past | Campaign settings | Immediate |
| Lifetime budget reached | Campaign settings | Immediate |
| No valid payment method on the primary account | Account billing | Hours |
| Card declined, or a chargeback occurred | Account billing | Hours to days |
| Policy violation | Ad review | Unpublished — no SLA exists |
Two of those are worth separating out. The first is that eligibility gates sit upstream of everything: the app must be available on the App Store in a supported region, the Apple Account needs an email address rather than a phone number alone, the user needs an Admin, Legal, App Manager or Marketing role, and a newly published app can take up to 24 hours to appear as a promotable option.
The second is that policy violation is the only entry on Apple’s list with no published resolution time. Apple’s H1 2026 transparency filing disclosed 12.3 million impressions served by ads that were removed after publication, and there is no advertiser-facing appeal control and no review SLA anywhere in Apple’s documentation. If your campaign stopped and billing is clean, creative status is the next thing to check, not the last.
Supply-side stops: popularity, Search Match, match types
Once the account is clean, everything remaining is an auction outcome, and Apple provides no status string for auction outcomes. A campaign that is Running and spending nothing is telling you that not enough matched searches happened, or that your bid did not clear.
Start with popularity, and start by fixing the scale. Apple publishes search popularity on a 1 to 5 scale. That is the number in Apple’s glossary and in Apple’s reporting definitions: a relative indicator displayed as numbers from 1 to 5, with 5 being the most popular. The 5-to-100 and 0-to-100 popularity indexes that circulate in this industry are vendor constructions. SplitMetrics says so openly, describing its own 0-to-100 percentile range as derived from Apple’s underlying data without disclosing the conversion method. Other vendors publish an index without stating a scale at all. If your keyword list was built against a vendor index, you do not know its Apple popularity, and the two are not interchangeable.
Then check whether Search Match is carrying the campaign or has been switched off. Apple defines it as a feature that matches ads to search terms automatically, drawing on your App Store listing metadata, information about similar apps in the same genre, and other search data. It is on by default in Manage Bids search results campaigns and it is a requirement for the automatic ad group in Maximize Conversions campaigns. On thin keyword lists it is frequently the only thing generating impressions, which makes disabling it a common and invisible cause of collapse.
Negative keywords: the failure Apple never warns about
Here is a genuine asymmetry. Every practitioner guide worth reading warns that aggressive negative lists suppress delivery. Apple has never said so anywhere.
Apple’s negative keyword page is purely mechanical: the default match type is exact, negatives can be added at campaign or ad group level, and you can bulk upload a maximum of 5,000 negative keywords per ad group. There is no caution about over-negation on that page. The troubleshooting page does not list negatives among the causes of low impressions. Apple’s stated purpose for the feature, in its keyword best practices, is cost control — to avoid paying for taps that are unlikely to convert to downloads.
What Apple does confirm is the part that makes over-negation dangerous: if you are using Search Match, you may also want to consider negative keywords to ensure your ad does not appear for certain search terms. Negatives apply to Search Match traffic. So a broad exact-match negative list, accumulated over quarters by successive account managers, can quietly disable the discovery mechanism described in the previous section without producing any error state at all. Audit the list against the search term report before you touch a bid.
→ FREE 10-POINT AUDIT
Find out why your budget isn’t clearing
We open your Apple Ads account, separate the account-level stops from the auction-level ones, and show you where your negative lists and match types are suppressing the exact queries you are trying to buy. One hour. Senior strategist. No pitch.
Book my free audit →
Maximize Conversions: the undocumented failure mode
Apple made Maximize Conversions generally available in late February 2026. Its mechanic, in Apple’s words: Target CPA is the ideal amount you want to spend per install, and Apple’s auto-bidder uses that amount to calculate optimal bids for each search query. The target is pursued on a weekly basis, not per query.
Apple then publishes two soft constraints. It recommends a daily budget allowing at least five conversions per day, and it asks advertisers to let campaigns run at least two weeks before assessing impact. The strategy is search-results only and is designed for post-launch campaigns, not pre-order campaigns.
What Apple does not publish is what happens when you violate either constraint. There is no documented behaviour for a campaign sitting below five conversions per day, no statement about whether delivery is throttled during the two-week learning period, and no description of what an unreachably low Target CPA does to volume. Every confident explanation you have read about Maximize Conversions underdelivery is inference from practitioner experience. Some of it is probably right. None of it is documented.
Worse, Apple contradicts itself on the most consequential control. Apple’s Maximize Conversions best-practices page states that if you want more control, you have the option to manually set your own keyword bids at any time. Trade coverage of the rollout quotes the opposite: that you cannot add or manage keyword bids in Maximize Conversions campaigns. Both readings were live in August 2026. Before you diagnose a bidding problem, verify in your own account which of the two is true for you — and see the structural playbook for running this alongside manual campaigns.
Two ads per query, one impression-share metric
In March 2026 Apple expanded search results from one ad slot to two, starting in the UK and Japan on 3 March and reaching all markets by the end of the month, on devices running iOS and iPadOS 26.2 and later. Existing search results campaigns became eligible automatically, with no advertiser action required. Apple confirms that your ad may be shown in any of the available ad positions and that you cannot select or bid for a particular position. Pricing did not change.
This is where diagnosis gets genuinely hard, because Apple exposes exactly two impression-share metrics and neither has a position dimension you can slice.
2
Ad slots per search query since March 2026
2
Impression-share metrics Apple publishes
0
Reports with ad position as a dimension
0
Published studies of the spend effect
Apple offers Impression Share (First Ad Position) and Impression Share (All Ad Positions). The Impression Share report breaks out by day, app, app ID, country or region, and search term. Position is not among the dimensions, and the API’s impression-share endpoints return the same shape. You can subtract first-position share from all-positions share to estimate second-slot share, but that is a derived figure Apple does not sanction, and it tells you nothing about taps, conversion or cost by position. Note also that impression share is reported as an exact percentage only from 0 to 90%; above that it collapses to a single 91 to 100% band.
The consequence for this diagnostic is specific. If your delivery changed in 2026 and you cannot explain it, the second slot is a candidate explanation you cannot test. Vendors disagree about which direction it pushes: AppTweak predicted campaigns would hit daily limits earlier and stop serving later in the day, while Adapty predicted more available impressions and cost-per-tap falling over time for well-organised advertisers. Neither published data. As of late August 2026, no before-and-after study of multi-slot’s effect on an existing campaign’s spend exists anywhere in public, which is also why every published Apple Ads benchmark still describes a one-ad auction.
One study does exist on placement itself. In March 2026 an analysis of 132 keywords and 627 keyword-app auction observations in the UK App Store, scoring relevance with a language model, found the most semantically relevant app took the top position in only 43.9% of cases. Read carefully, that says bid strength does most of the ordering work once an app clears Apple’s relevance threshold. It is one market, one vendor’s data, an undisclosed scoring prompt and no confidence intervals — but it is the only public evidence on the question, and it argues against treating a low bid as safe.
The diagnostic sequence, in order
Order matters, because each step is cheaper than the one after it and rules out a whole class of cause.
| # | Check | Rules out |
|---|
| 1 | Campaign status, end date, lifetime budget, billing, ad review status | All documented account-level stops |
| 2 | Trailing 30-day spend against daily budget × 30.4 | The false budget-exhaustion diagnosis |
| 3 | Storefront availability, supported regions, app eligibility | Inventory that never existed |
| 4 | Search term report against the full negative list, both levels | Self-inflicted suppression |
| 5 | Search Match on or off; match type mix; Apple popularity 1–5 | Genuinely thin query supply |
| 6 | Impression share, first position and all positions, by search term | Bid clearing failure |
| 7 | Bidding mode, Target CPA level, conversions per day, days since change | Learning-period and auto-bid effects |
Steps one through four are answerable from the account in under an hour. Step five requires you to stop using a vendor popularity index as though it were Apple’s. Step six is the first point at which you are looking at an auction rather than a configuration. Step seven is where the documentation runs out, and where you should be explicit with stakeholders that you are reasoning past the evidence.
A quiet campaign with clean billing is not broken. It is a market telling you something about your bid, your queries or your list — and Apple has built almost no instrumentation for hearing which.
What nobody can tell you
Name these gaps to your stakeholders before someone else names them to you.
Apple publishes no dates on its Help pages and no changelog, so no page can be aged against the 2026 changes. Apple documents no underdelivery behaviour for Maximize Conversions at all, and contradicts itself on whether keyword bids are available in that mode. Apple has never published a warning that negative keywords suppress delivery, so the most widely repeated cause in the industry rests entirely on practitioner inference. There is no slot-level impression share, no position dimension in any report or API object, and therefore no way to attribute a delivery change to the second slot; second-slot share can only be derived by subtraction, and that derivation is ours, not Apple’s. No study anywhere quantifies multi-slot’s effect on an existing campaign’s spend. The one public auction study is 627 observations in a single market. And published cost benchmarks contradict each other by wide margins on the same year’s data, which makes any “your bid is too low” conclusion drawn from a benchmark a weak conclusion.
None of that makes the channel unmanageable. It makes it a channel where the honest diagnostic ends at step seven and says so, rather than manufacturing a cause. If you want the adjacent problems: the conversion surface your taps land on is about to gain a new testable asset, the reporting pipeline you will run this diagnostic through has a January 2027 sunset, and the alternative channel produces three non-agreeing install counts by design.