Apple Search Ads best practices: judge keywords by trials
Apple Search Ads best practices for subscription apps: four campaigns, bids, custom product pages and a weekly routine judged by cost per trial.
Run Apple Search Ads as four separate campaigns, and judge every keyword by its cost per trial start, not its cost per install. Picture a growth lead on Monday morning: one keyword has the cheapest taps in the account and zero trials after two weeks.
A cheap install that never starts a trial is waste, however good it looks in the ad dashboard. This setup works for subscription apps and AI tools: structure, measurement, bids, custom product pages and a weekly review.
What Apple Search Ads is now
Apple renamed the product in 2025. Its news page, in an item dated April 14, 2025, says "Apple Search Ads is now Apple Ads." The reason Apple gives: "Our ad offerings have grown beyond search." The rest of this post uses Apple Ads, the name on Apple's own help pages.
Apple Ads now covers four App Store placements. Apple's placement guide lists Today tab ads, Search tab ads, search results ads and ads on product pages. Only search results campaigns use keywords. The other three placements have no keyword targeting.
Search results remain the main placement for most subscription apps. Apple says "Almost 65% of downloads happen directly after a search." That figure covers the App Store worldwide in 2022. A searcher names the job they want done, and a keyword lets you buy that intent.
You do not control where your ad lands on the page. Apple says "your ad can appear either at the top of their search results or further down the page." It adds that "you can't select or bid for a particular ad position."
Open your account this week and list the placement each campaign runs on. Keep keyword work inside search results campaigns, and treat the other placements as separate tests.
The four-campaign structure Apple recommends
Apple's campaign structure guidance splits search results into four campaigns. Each campaign holds one kind of intent, so its numbers mean one thing.
- Brand: "keywords that relate to your app or company name".
- Category: non-branded keywords that describe your app's category or what it does.
- Competitor: "keywords for apps that are similar to yours".
- Discovery: "reaching a wider audience and mining for popular search terms to add as keywords".
Brand, category and competitor run on exact match, and Apple tells you to "make sure Search Match is off". Discovery gets two ad groups. One runs broad match keywords with Search Match off. The other has no keywords and Search Match on.
Then add every brand, category and competitor keyword to discovery "as exact match negative keywords". Teams often skip this step. Without it, discovery buys your brand term at discovery bids, and two campaigns split the same searches between their reports.
Set match types with care, because mistakes are hard to undo. Apple's match type page calls broad match "the default match type for new keywords in Apple Ads". It also says "Once they've been added, match type can't be changed." A keyword added as broad by accident has to be paused and added again as exact.
Exact match trades reach for precision. Apple says your ad may see fewer impressions with exact match, while "your tap-through rates (TTRs) and conversions on those impressions may be higher". For keyword-level reading, that precision is the point.
Check every keyword in your brand, category and competitor campaigns this week. Move anything on broad match, and switch off Search Match in those three campaigns.
What to measure for a subscription app
In Maximize Conversions, Apple's automated bidding, a conversion is an install. The Maximize Conversions page defines target CPA as "the ideal amount you want to spend per install." Trials and subscriptions are not part of that number. A keyword can hit its install target and still bring no paying users.
Trial starts and paid users come from your own data. Apple's AdServices framework gives your app an attribution token. You or your MMP send it to Apple's attribution server, which returns the attribution record. Apple's documentation lists campaign, ad group and keyword IDs in that attribution data. Apple also states that "AdServices does not engage in tracking as defined under the App Tracking Transparency framework."
Subscription tools can join that keyword data to revenue. RevenueCat's Apple Ads integration feeds charts that measure "subscription, trial, revenue, lifetime value, and conversion data", split by campaign, ad group and keyword. An MMP can do the same job if it receives your subscription events.
Pick one cohort rule and keep it everywhere. Group installs by the week they happened. Count trials started within 7 days of install, and paid users at day 30. Then read three numbers per keyword: cost per install, cost per trial and cost per paid user.
Your own past cohorts are the only fair yardstick for cost per trial. The public benchmarks later in this post stop at installs.
A keyword's trial rate also depends on what happens after the install. Onboarding and the paywall model decide whether a motivated searcher starts a trial. For Pixel+, an AI photo and video editing app, Kreto ran ads across Google UAC, Meta and Apple Search Ads. The same project redesigned onboarding and the paywall. Read a weak keyword against those later steps before you cut it.
Connect AdServices attribution to your MMP or subscription tool this week. Then build one table with a row per keyword: spend, installs, trials within 7 days and paid users at day 30.
How to set bids and target CPA
Your max CPT bid is a ceiling, not a price. Apple says the bid "acts as a price ceiling" and that the "actual price you end up paying may be less than your max CPT bid amount."
Apple also advises you to "apply your most aggressive max CPT bids to the highest-performing keywords." For a subscription app, highest performing means the lowest cost per trial. Cost per install comes second.
Maximize Conversions sets bids for you, "maximizing conversions, at a target CPA you set." Because the conversion is an install, the system looks for cheap installs. Apple recommends a daily budget that allows "at least five conversions per day" and a run of "at least two weeks" before you assess. Apple says CPA caps will soon be unavailable, so plan around target CPA.
Derive the install target from the trial cost you can afford. Say you can pay $30 per trial, and 20% of a campaign's installs start a trial within 7 days. Your target CPA is then $6 per install. Recheck the trial rate at the end of the test window. If the cheaper installs start fewer trials, lower the target or return to manual bids.
What the 2025 benchmarks show
AppTweak's Apple Ads benchmarks give a rough reference. The figures are medians for search results campaigns in the US, January to December 2025, from about 2,800 advertisers tracked by AppTweak. The table shows apps only. AppTweak's "generic" type is what Apple calls category.
| Campaign type | Median CPT | Median TTR | Median tap-to-install rate |
|---|---|---|---|
| Brand | $2.05 | 25.3% | 73.5% |
| Generic (category) | $2.75 | 8.5% | 59.8% |
| Competitor | $3.26 | 3.8% | 50.1% |
| Discovery | $1.16 | 4.2% | 44.5% |
Discovery has the cheapest taps in the set and the lowest share of taps that become installs. Its median CPI still sits near brand, while competitor installs cost the most. AppTweak says the dataset "contains no revenue, ROAS or lifetime value," so none of these medians tell you what a trial costs.
Compare your CPT per campaign type with the table this week. Then sort your keywords by cost per trial and raise bids only on the top of that list.
Custom product pages matched to keyword intent
A searcher who types a specific job wants to see that job on the first screen. Custom product pages let you show it. Apple lets you build "up to 70 additional versions" of your product page. Apple Ads can use them as ad variations in search results ad groups.
Each ad group gets one page. Apple's help page says "Each ad group can only have one active custom ad." You can set pages up "without submitting new app versions." Group keywords by intent so one page fits all of them. A photo app might split background removal, photo restoration and AI portraits into three ad groups.
Two sources report a lift, measured differently. Apple says "Developers see a 2.5 percentage point increase on average when referring people to a custom product page." Apple gives no date or method.
AppTweak's US 2025 data shows a gap of 3.2 points: "Apps using a CPP convert taps to installs at 58.7% versus 55.5% on the default page." The samples and methods differ, and neither source describes a controlled test. Apple does not name its metric, while AppTweak's gap is tap-to-install rate. Read both as a hint of a few points, not a forecast.
Most apps still skip them. AppTweak found that "61% of apps and 74% of games ran no custom product pages in Apple Ads in 2025." Judge each page by trials within 7 days as well. A page that promises a feature the paywall then blocks can lift installs and lower trials.
Pick your largest category ad group this week. Build one custom product page for its intent and attach it as that ad group's ad.
A weekly routine for Apple Ads
Most of the work is one review a week, done the same way each time. Kreto runs Apple Ads campaigns as part of its performance ads work. The steps:
- Open the search terms report for discovery and sort by spend.
- Promote relevant search terms with steady installs to category or competitor as exact keywords. Add each one to discovery as an exact negative the same day. As exact keywords, they get their own trial data.
- Add exact negatives for terms that do not match your app's job, whatever they spend.
- Update the keyword table for every cohort that has reached its window: trials within 7 days, paid users at day 30.
- Set bids by cost per trial. Raise below target, hold near target, lower above it.
- Pause a keyword once it has spent two to three times your target cost per trial with no trial.
- Log each change and its reason, so next week starts from facts.
Keep the order. Search terms come first, because new exact keywords change what discovery spends on next week.
Put a fixed 45-minute slot in your calendar for this review, starting this week.
Frequently asked questions
How much does Apple Search Ads cost?
Keyword campaigns charge per tap, at or below your max CPT bid. In AppTweak's US 2025 medians for apps, CPT ran from $1.16 for discovery to $3.26 for competitor campaigns. Your category and country will move that number. Apple also offers Apple Ads Basic, with a "Cost-per-install pricing model", search results only and no keywords.
Is Search Match worth turning on?
Yes, in the discovery campaign. Apple's glossary says Search Match "allows ads to be matched automatically to search terms". In discovery, that finds searches you would not have guessed. In brand, category and competitor campaigns, it adds unplanned searches and muddies the keyword data.
How long should a keyword run before you judge it?
Apple asks for at least two weeks before you assess a Maximize Conversions campaign. For a single keyword, wait for volume too. Judge it once spend reaches two to three times your target cost per trial. Low-volume keywords can take a month.
Bid on what happens after the install
An Apple Ads keyword is worth what its users do after the install. Give each keyword its own row, join that row to trials and paid users, and fix the funnel step that leaks before you raise a bid.
If you want help structuring Apple Ads around trial and revenue data, talk to Kreto. We can go through your keyword table and funnel with you on a discovery call.