Apple changed how bids are set and how many ads a query can hold, five weeks apart, and delivered both changes inside campaigns that already existed. No migration, no prompt, no new campaign type to opt into.
That is the detail worth pausing on. When Google or Meta rebuild an auction, you get a deprecation notice and a migration deadline. When Apple expanded App Store search results on March 3, 2026, existing search campaigns became automatically eligible for the new placements — advertisers did not need to do anything. Which means the accounts that adapted and the accounts that didn't look identical in the interface, and different only in the payback report.
Here is what changed, why the second slot behaves differently than most people assume, and the structure we would rebuild an account into today.
01What changed, in five weeks
| Date | Change | Advertiser action required |
|---|
| Feb 26, 2026 | Maximize Conversions reaches general availability for all App Store advertisers | Opt-in per campaign |
| Mar 3, 2026 | Multiple search placements begin rolling out — UK first, then Japan | None |
| By Mar 31, 2026 | Multi-slot expansion completes globally | None |
| Mar 17, 2026 | Insights, an account-wide analytics engine, is announced | None |
| Jun 8, 2026 | Custom creative assets announced for Today tab and search results ads | Upload assets |
Two of those five arrived with no action required and no obvious in-account signal. That is the operating condition to plan around: on Apple Ads, the auction can change underneath a campaign that you have not touched in a quarter.
02The second slot is not a second chance
Up to two ads can now appear for a single search query, where previously there was one placement at the top of results. The intuitive read — “more inventory, so cheaper traffic” — has not held up, for three reasons.
Reason 1 · Auction dynamicsPosition is bought, not assigned
Placement is determined by auction dynamics including bid and relevance. You do not choose which slot you appear in, and you cannot bid specifically for the second one. A March 2026 analysis of 627 keyword observations found that bid outperformed relevance in 44% of cases — meaning that in close to half of contested queries, the higher bidder took the position regardless of the relevance signal. A well-targeted, highly relevant advertiser can be displaced by a less relevant one willing to pay more.
Reason 2 · New entrantsKeywords that were unwinnable are now winnable
The advertisers who benefit most from a second slot are the ones who could never take the single slot. Every keyword where an incumbent held position one uncontested is now a keyword where a challenger can buy visibility. Supply went up, but so did the number of bidders per query.
Reason 3 · Position value is not linearSlot two converts differently
Apple has cited conversion rates in excess of 60% for top-of-search placements across 175 App Store storefronts and 44 currencies. That figure describes the premium position. The second slot is a genuinely different product with a different conversion profile, and it is priced through the same cost-per-tap mechanism. If your blended conversion rate has drifted down since March while your cost per tap held roughly flat, the mix shift between positions is the first thing to check.
→ COMMON LEAK
Blended reporting hides a position mix shift entirely.
An account that used to win slot one 40% of the time and now wins some combination of slot one and slot two 65% of the time will show more impressions, more taps, a lower tap-through rate and a lower conversion rate — while every one of those numbers is an average across two placements that behave differently. Nothing in the account is broken. The report is just no longer measuring one thing.
03What Maximize Conversions actually optimises
Maximize Conversions replaced a rigid mechanism with a flexible one, and the difference is more interesting than “Apple added smart bidding.”
The deprecated CPA cap was a ceiling applied per query. Its failure mode was structural: it could not bid above the cap on a search that was genuinely worth more than the cap, so the most valuable queries in an account were systematically the ones it declined to compete for. Every advertiser who ever set a cap was, by definition, refusing their best traffic.
Target CPA under Maximize Conversions is an average held across a week, not a per-query limit. The system can pay well above target on a query it believes will convert and compensate elsewhere, with total spend bounded by the daily budget multiplied by 30.4 days. You set two things — target CPA and daily budget — and Apple adjusts individual bids in real time, with Search Match providing keyword coverage.
| CPA cap (deprecated) | Maximize Conversions target CPA |
|---|
| Applies to | Every individual query | The weekly average |
| Can exceed target on a single query | No | Yes |
| Keyword-level bid control | Yes | No |
| Keyword coverage | Your keyword list | Your list plus Search Match |
| Availability | — | Post-launch campaigns only; not pre-order |
| Negative keywords | Yes | Yes, via optional manual ad groups |
Apple's own operating guidance is specific and worth following literally: set a daily budget that permits at least five conversions per day, and leave the campaign alone for a minimum of two weeks before adjusting settings. Both rules exist because the system is fitting to your conversion signal. Starve it or interrupt it and the target CPA is fitting to noise.
5/day
The conversion floor. Below roughly five conversions a day, Maximize Conversions has too little signal to optimise against — and no amount of patience fixes a budget that cannot clear the threshold.
One published advertiser account reported getting “over 30 percent more installs with the same budget” from April–May 2025 testing, and Apple's own Toca Boca World success story from May 2026 credits a combination of multiple placements and Maximize Conversions campaigns with a 5× increase in global brand growth. Treat both as directional. They are single accounts, self-selected for publication, with no controlled comparison — and no credible independent Maximize Conversions study exists yet. That gap is real, and pretending otherwise would be the easiest way to mis-set your expectations.
04The account structure that survives both
The structural principle is the same one that has always separated accounts that scale from accounts that plateau, and both 2026 changes made it more consequential: never let placements or match types with different conversion economics share a budget or a target.
Layer 1 · BrandManual bidding, exact match, defended
Your own brand terms convert far above account average and cost far below it. This is exactly the traffic you want manual control over, and exactly the traffic where automation's willingness to overpay creates no value — you were going to win those users anyway. Keep brand on manual bidding, exact match, in its own campaign with its own budget. The brand-keyword overspend problem gets worse in a two-slot world, not better; we cover the diagnosis in the brand keyword overspend post.
Layer 2 · Competitor and categoryManual, exact, aggressive ceilings
High-intent, high-cost, and worth a ceiling set from your own payback arithmetic rather than a benchmark. This layer is where the second slot most changed the economics, because competitor terms are the ones challengers now buy into. Set the ceiling using the formula in the ARPU piece and hold it.
Layer 3 · DiscoveryMaximize Conversions, properly funded
This is where Maximize Conversions earns its keep — broad, unpredictable, long-tail demand where per-query manual bidding was never going to be accurate anyway. Fund it above the five-conversions-per-day floor or do not run it. A half-funded Maximize Conversions campaign is strictly worse than a manual one.
Layer 4 · NegativesOne shared exclusion discipline
Maximize Conversions campaigns still support negative keywords through optional manual ad groups, and with Search Match doing keyword coverage you need them more than before. Mine the search-term report weekly and push exclusions across layers. This is the single highest-yield recurring task in a post-March account.
→ FREE 10-POINT AUDIT
See which layer is quietly funding the others
We split your account into brand, competitor and discovery, show you the true cost and payback of each, and tell you which one to fund and which one to cap. One hour. Senior strategist. No pitch.
Book my free audit →
05Two ways budget saturates now
More inventory per query means more impressions available to the same campaign, which means daily budgets that were comfortable in February can exhaust before the day's highest-value hours.
Trap 1 · Early exhaustionThe day ends before your users wake up
A campaign that spent evenly across 24 hours pre-expansion may now clear its daily budget by early afternoon. If your highest-converting cohort searches in the evening, you have quietly stopped buying them — and because the campaign is spending its full budget and reporting an acceptable cost per install, nothing in the interface flags it. Check hourly delivery, not just daily spend.
Trap 2 · The 30.4 multiplierMonthly reality versus daily intent
Maximize Conversions bounds total spend at the daily budget multiplied by 30.4 days. Advertisers who think in daily terms and set an aspirational daily budget as a safety margin can find the campaign using that headroom on high-cost days. Set the daily budget to the number you actually intend to spend monthly divided by 30.4 — not to a ceiling you would be uncomfortable hitting.
06Brand defence in a two-slot auction
The clearest published illustration of what impression share is worth on brand terms comes from The Economist, which reported 44% download growth while maintaining 90% impression share on its branded keywords. That was a single-slot world. In a two-slot world the same defensive logic applies to a query that can now hold a competitor alongside you.
Three practical rules:
- Monitor impression share on brand terms weekly, not monthly. A competitor entering your brand auction shows up as an impression-share decline before it shows up as a revenue decline, and the gap between those two moments is your response window.
- Unify ASO and Apple Ads keyword strategy. These have historically sat with different owners on different tools. In a denser auction, organic position and paid position on the same query interact — treating them as separate functions leaves the interaction unmanaged.
- Do not defend everything. Brand defence has a ceiling set by the same payback arithmetic as everything else. Full impression share on a term with poor downstream retention is a cost, not a moat.
07The 14-day rollout sequence
If your account has not been restructured since February, do it in this order. The sequence matters more than the speed — every step depends on the baseline captured in step one.
| Days | Action | What you're protecting against |
|---|
| 1 | Export the trailing 90 days: cost per tap, TTR, conversion rate and day-30 payback by campaign, ad group and keyword | Losing the pre-restructure baseline |
| 1–2 | Check hourly delivery on every campaign at or near budget | Silent early exhaustion |
| 2–3 | Split brand, competitor and discovery into separate campaigns with separate budgets | Blended averages hiding the leak |
| 3–4 | Set per-layer bid ceilings from your own revenue per install, not benchmarks | One cap that is wrong for every cluster |
| 4–5 | Move discovery to Maximize Conversions — only if funded above five conversions/day | Automation fitting to noise |
| 5 | Set the daily budget as intended monthly spend ÷ 30.4 | The multiplier trap |
| 5–19 | Do nothing. Two full weeks, no setting changes | Resetting the learning period repeatedly |
| Weekly throughout | Mine search terms, push negatives, check brand impression share | Search Match drift and competitor entry |
| Day 19+ | Compare against the day-one baseline at the layer level, not blended | Judging the restructure on the wrong number |
→ THE FIX
The hardest step is day 5 through day 19.
Two weeks of not touching a campaign is genuinely difficult when the numbers wobble on day 8, and it is the step accounts skip. Every adjustment restarts the fit. If you cannot commit to the fortnight, keep the layer on manual bidding — a well-set manual bid beats a Maximize Conversions campaign that gets rescued every three days.
None of this is exotic. It is the same discipline that has always separated accounts that scale from accounts that plateau: keep economically different traffic in economically different containers, price each one against your own revenue rather than someone else's average, and give automation enough signal and enough time to be worth what it costs you in control. What changed in 2026 is only the penalty for not doing it — a denser auction punishes blended structures faster than a single-slot one ever did.