Open any Apple Ads help page and look for a date. There isn’t one — not a publication date, not a last-updated stamp, not a changelog.
That sounds like a documentation nitpick. It is actually a measurement problem with money attached. Apple changed its bidding model in February 2026 and its search inventory in March 2026. If you cannot tell whether a page was written before or after those changes, you cannot tell whether the guidance on it describes the product you are buying. And when two Apple pages state different numbers for the same thing — which happens more often than the industry has noticed — there is no mechanism at all for deciding which is current.
What follows is a file, not an argument. Every entry was checked against Apple’s live pages on September 4, 2026. Where something has changed since we last looked, we say so, including one case where Apple quietly corrected itself and left no record of having done it.
01No Apple Ads page carries a date
Every page checked under the Apple Ads help tree and the best-practices tree displays no publication date, no last-updated date and no changelog. The glossary, the reporting definitions, the budget pages, the campaign-creation pages, the placement and bidding best practices — none of them. The single date-like signal is a site-wide copyright footer reading “Copyright © 2026 Apple Inc.,” which is identical on every page and therefore says nothing about the content above it.
There is one buried tell. Some embedded image filenames carry dates — the recommendations best-practices page serves art stamped 2025-01-30 — but those are invisible to a reader and prove only when an image was produced, not when the prose was last touched.
Apple’s developer documentation does better: the Platform API publishes availability strings, which is how we were able to date endpoints in the API migration piece. The advertiser-facing help tree has no equivalent.
→ THE PRACTICAL RULE
Cite Apple Ads Help by access date, never by publication date
There is no publication date to cite. Any citation implying one is manufacturing provenance Apple did not provide.
02Three visitor counts became two, silently
Apple’s advertising site has been carrying multiple, incompatible answers to the simplest question about its own audience.
On the Apple Ads home page and on the App Store advertising page: “Over 850 million people visit the App Store every week,” footnoted to “App Store, worldwide, 2026.”
On the ad placements best-practices page, in the Today tab section: “Promote your app where more than half a billion visitors start their journey to discover new apps every week.” No footnote. No source. No date.
Those two are still live and describe the same population with a gap of roughly 350 million people — and the smaller figure is the unsourced one.
Now the part that makes the dating problem concrete. As recently as late August 2026, Apple’s Advanced page carried a third figure: “over 650 million people go every week.” As of September 4 it reads 850 million. Apple corrected one of three inconsistent numbers and published no notice, no changelog and no dated revision. The only reason anyone can say the edit happened is that someone recorded the earlier wording. That is a fair description of the entire epistemic situation: Apple’s documentation is a live document with no version history, and the industry’s only defence is its own note-taking.
03Four headline figures, four vintages
Apple footnotes its main advertising claims properly, which is more than most platforms do. Read the footnotes together, though, and the four numbers underwriting the case for App Store search turn out to come from four different eras.
| Claim | Apple’s footnoted data window | Age today |
|---|
| Over 850 million weekly App Store visitors | Worldwide, 2026 | Current |
| Over 60% conversion for ads at the top of search results | Nov 2024 – Oct 2025 | Closed 11 months ago |
| 95% of downloads after an ad tap occur within a minute | Aug 1 – 31, 2024 | A single month, 2 years old |
| Almost 65% of downloads happen directly after a search | Worldwide, 2022 | 4 years old |
The 2022 figure is the one to watch, because it is the one most often repeated as though it were current — including in Apple’s own 2026 announcements. On Apple’s Advanced page a second 2022-vintage footnote covers both the 65% claim and the statement that “70% of App Store visitors use search to discover apps,” alongside a commissioned GWI survey from January to February 2022.
None of this makes the underlying thesis wrong. Search-led discovery on a store whose users out-spend the alternative by a wide margin is still the structural reason Apple Ads buys the highest-value iOS users — the derivation is in the ARPU piece. But a four-year-old share-of-discovery figure should be presented as a four-year-old figure, and a one-month sample should be presented as a one-month sample.
04The same claim, footnoted here and bare there
A smaller entry, and a clean illustration of how these things propagate.
The App Store advertising page states “Almost 65% of downloads happen directly after a search,” footnoted to 2022 data. The ad placements best-practices page states “Nearly 65 percent of downloads happen directly after a search” — different wording, same statistic, and no footnote at all.
A practitioner who lands on the second page has no way to learn the figure is four years old. When they quote it in a deck, the vintage disappears permanently. Multiply that across an industry and you get the situation we found while assembling the benchmarks comparison: figures circulating with more confidence than their sources ever claimed for them.
There is a related orphan worth flagging. A widely repeated claim that more than 85% of App Store visits result in a download could not be located on any Apple property — not the advertising home page, not the App Store page, not the Advanced page, not the placements best practices. If you have it in a deck, it did not come from Apple.
05Apple against Apple on keyword bids
This is the entry with the most direct operational cost, and it has now got worse rather than better.
Apple’s Maximize Conversions best-practices page: “And if you decide you want more control, you have the option to manually set your own keyword bids at any time.”
Apple’s own campaign-creation help page describes the same strategy as using Search Match to find popular terms “but without the need to manage individual keyword bids.”
PPC Land’s February 26, 2026 rollout coverage puts it flatly: “you can’t add or manage keyword bids in Maximize Conversions campaigns.”
When we first logged this, it read as Apple versus a vendor. It is now Apple versus Apple: one Apple page offers a control that another Apple page says is unnecessary and a third party says does not exist. All three are live. The account structure implications are not trivial — whether you can hand-price a term inside an automated campaign determines whether you need a parallel manual campaign at all, which is the fork we mapped in the multi-slot playbook.
Where Apple’s own pages disagree about what a control does, the only reliable documentation is your own account.
06A superseded control, still in the fix-it page
Apple’s sole page on low delivery — the one an advertiser reaches when campaigns are not spending — still instructs readers to “check your cost-per-acquisition (CPA) cap.”
A cap and a target are different instruments. Apple’s campaign-creation page describes the current control as a target CPA, attached to Maximize Conversions, which arrived in February 2026. The manual bidding best-practices page — the page governing the mode where a cap would have lived — does not list a CPA cap as an available setting at all; it mentions CPA only descriptively, in a sentence about a keyword delivering taps at an acquisition cost higher than your target.
So the troubleshooting page sends advertisers to check something that the campaign-creation page and the bidding page do not describe. We used this as the epistemic frame for the delivery-diagnostics piece, and it still holds: the page most likely to be read under pressure is the one carrying the stalest instruction.
One correction to our own earlier record, in the interest of not doing to vendors what we are criticising. We previously logged AppTweak’s guidance as contradicting Apple on daily budgets — specifically the claim that budget exhaustion stops ads by midday. AppTweak’s campaign-structure guide, updated February 17, 2026, now states the correct model: monthly spend is calculated on the average 30.4-day month, and daily spend may exceed the daily budget on high-opportunity days. AppTweak fixed it. Apple has not fixed the CPA cap.
→ FREE 10-POINT AUDIT
Find out which of these are costing you money
We open your Apple Ads account and check it against what the platform actually does today — not what an undated help page says it does. Bidding mode, impression share, budget behaviour, placement mix. One hour. Senior strategist. No pitch.
Book my free audit →
07Impression share and Rank, described two ways
Apple’s reporting definitions page publishes two distinct metrics. Impression Share (First Ad Position): “The number of times your ad was shown in the first position for a specific search term, out of the total number of searches on that search term, in the same App Store countries and regions.” Impression Share (All Ad Positions) is the same sentence with “any ad position” substituted. Both are noted as available only in predefined Impression Share reports and on the Recommendations page.
The Recommendations best-practices page describes something else entirely: a single, position-blind metric, “displayed as a percent range.” One Apple page gives you two precise position-split ratios; another gives you one banded number.
Rank suffers the same split. The reporting definitions page defines Rank against impression share with no scale attached. The recommendations page bounds it: “Rank is displayed from 1 to 5 or >5, with 1 being the highest rank.” Same metric, one page carrying the scale you need to interpret it.
Underneath both is the structural gap we have flagged before and that nothing has closed: Apple publishes no dimension exposing which position an individual impression occupied. Second-slot share exists only as a subtraction Apple does not sanction, and it returns no taps, no conversions and no cost by position.
08The conversion figure that predates the second slot
This is the entry that matters most, and it is not a typo or a stale bullet. It is an evidence gap sitting under the platform’s single most-quoted number.
Apple sells up to two ad positions per App Store search query. That rollout began in the UK on March 3, 2026 and completed globally by the end of that month. Six months later, Apple’s own placement documentation has not been updated to acknowledge it. The ad placements page and the campaign-creation page both describe search results advertising without stating how many ads can appear per query; create-campaigns says only that when a customer types a relevant query including one of your keywords, “your ad could appear.”
Meanwhile Apple’s headline conversion claim remains scoped, in Apple’s own words, to “ads at the top of search results,” measured November 2024 to October 2025. That window closes five months before the second slot finished rolling out worldwide.
So the number every advertiser cites to justify top-of-search spend describes an auction in which there was only a top. Apple has not restated it, and no vendor has filled the gap: every public Apple Ads benchmark — Adapty, AppTweak, SplitMetrics, MobileAction, Singular, Business of Apps — still samples calendar 2025 or earlier. Checked again in September 2026, there is still no published dataset of any kind describing the two-ad auction. Tap-through rate is the metric most exposed to the change, and as section 07 shows, Apple ships no reporting that would let an advertiser diagnose a TTR decline as second-position delivery rather than weak creative.
None of that means the conversion figure has fallen. It means nobody — inside or outside Apple’s published record — can currently say whether it has.
09How to read an undated page
A short method, since the condition is not going away.
First, check whether the page names a control that was introduced or retired in 2026. Maximize Conversions, target CPA, multi-slot search results and the Insights engine all landed between February and March 2026. A page that mentions none of them and describes CPA caps is probably older than the product.
Second, prefer pages that carry footnotes over pages that do not. Apple’s marketing pages footnote data windows; its best-practices pages frequently restate the same claims bare. Where the two disagree, the footnoted one is at least auditable.
Third, treat the developer documentation as the more reliable half of Apple’s output. The Platform API publishes availability strings and enumerated objects; the advertiser help tree publishes prose. When we needed to establish what Apple had actually shipped for Maps, the answer came from the documentation, not the marketing — as set out in the Maps mechanics piece.
Fourth, and least comfortable: keep your own dated copies. Section 02 exists only because someone wrote down what a page said before Apple changed it.
10What this file cannot settle
It cannot tell you which of two contradictory Apple pages is current, because no mechanism exists to establish that. It cannot tell you when the 650-million figure changed, only that it did. It cannot tell you whether the over-60% conversion rate has moved since a second advertiser started appearing on the same queries, because Apple has not restated it and no independent dataset covers the period.
It also cannot tell you that Apple is being careless. A large documentation estate drifts; the more interesting fact is the absence of the ordinary apparatus — dates, changelogs, revision notes — that would let anyone outside Apple manage that drift. Google Ads help pages carry last-updated stamps. Apple’s do not.
The practical consequence for anyone spending on this channel is narrow and worth stating without drama. Apple Ads remains the place where declared intent is intercepted at the moment of highest value, and that argument does not depend on any single figure above. But the numbers you use to size the opportunity should be dated, and roughly half the ones in general circulation cannot be. We will keep this file current as pages change — and if you spot an entry we have missed, or one Apple has quietly fixed, it belongs here. The companion piece on the creative assets review window runs into the same problem from the other direction: a format with a confirmed trigger, no published date, and no review commitment attached to it.