Search Match Query Expansion: How to “Promote” Winners and Clamp Down on Losers in Apple Search Ads

If you’re running Apple Search Ads with Search Match enabled, you’re already relying on Apple to find queries that look similar to your keyword set. The downside: you don’t get direct per-keyword control over which queries are showing your ads. Over time, that can make results feel like you’re paying to explore—rather than paying for performance.

A good way out is to treat Search Match like a research engine: let it discover, then “promote” the best queries into your own Exact keywords and “clamp down” the rest with negatives.

The core idea: promote discovered queries, don’t just judge Search Match

Search Match runs automatically and can surface a mix of query intent. Instead of asking, “Is Search Match profitable?” (it often starts noisy), ask two better questions:

• Which actual queries are producing installs with reasonable conversion (taps→installs) and acceptable CPI/CPA?
• Which queries are wasting taps (high TTR, low installs) or producing installs that don’t turn into revenue?

Once you can name those queries, you regain control using manual keywords and negatives—while still letting Search Match do new discovery.

Step 1: Put Search Match in its own ad group (so you can contain the damage)

This matters because Search Match is essentially “shared discovery.” If you mix it into the same ad group as your Exact/Broad manual keywords, it becomes harder to interpret changes.

Do this:

• Create one ad group whose keywords are only Search Match terms (the Discovery/Search Match keyword type).
• Keep your manual ad groups (Exact/Broad) separate.
• Use the same country and ad group structure discipline so you’re comparing like with like.

Why: if you later add negatives and see improvements, you want to know they’re actually coming from Search Match being constrained—not from something else.

Step 2: Wait for enough signal (then evaluate by query, not by vibes)

Use the search terms / query insights report for your Search Match ad group and filter to the period where delivery is stable.

When deciding whether a query is a “winner,” look at the chain:

• Taps → Installs conversion rate (installs divided by taps)
• CPI (cost per install)
• ROAS (only after your revenue pipeline maps installs to purchases)

A practical rule: give each query time to get enough taps. If a query only got a handful of taps, the CPI/ROAS will be basically noise.

A useful heuristic (so you don’t overreact)

• If a query has high taps but low installs, it’s often misaligned with your actual app promise (bad landing/product fit, unclear value prop, or the query is too top-of-funnel).
• If a query has reasonable installs but weak ROAS, it may be attracting the wrong audience for your offer (wrong feature emphasis, wrong subscription intent, or pricing friction).

Step 3: Promote winners into manual Exact keywords

For the queries that clearly outperform, you want to stop relying on Apple’s automatic expansion for them.

How to promote:

1. Pick the top subset of queries (not everything).
2. Create a new manual ad group (or add into an existing manual ad group dedicated to “promoted winners”).
3. Add each winning query as an Exact keyword.
4. Start with a cautious bid based on what Search Match is effectively paying for those queries (you can use your CPI/CPT patterns as a guide, but don’t assume you can copy one number perfectly).
5. Keep promotions limited per batch (more on batching below).

Outcome: if the query was good, Exact ensures you don’t accidentally broaden into lower-intent variants.

Step 4: Clamp down on losers with negatives (targeted, not reckless)

Negatives are where you regain efficiency. But don’t carpet-bomb—be surgical.

How to decide what to negative:

• Strong negative candidates: queries with lots of taps but very few/no installs, or queries where installs are happening but later revenue is consistently poor.
• Weak negative candidates: queries with minimal traffic (too few taps to trust) or queries that are only slightly underperforming.

How to apply negatives:

• Add negatives to the Search Match ad group that is generating the noise.
• If you have query-level insight, use it to target the exact phrasing that appears irrelevant.

Tip: your goal is to reduce spend on obviously misaligned intent, not to eliminate every exploration possibility.

Step 5: Use a batch workflow so you can tell what worked

If you promote 40 queries and add 200 negatives in one afternoon, you won’t know what caused any improvement (or decline).

A safer workflow:

• Run Search Match normally for discovery.
• On a schedule (e.g., weekly), do one batch:
  • promote, and
  • clamp with negatives,
  • and leave bids alone for the same batch.
• After the batch, watch for the next reporting window where delivery resumes.

You’re basically running a controlled experiment with a living keyword set.

Step 6: Keep “discovery” alive without mixing your controls

One trap: if you constantly clamp Search Match too aggressively, it can stop finding new winners.

A balanced approach:

• Keep Search Match active for ongoing discovery.
• Only promote the best queries into Exact.
• Add negatives mainly for the most clearly irrelevant queries.

This way, your manual side becomes more stable over time while Search Match keeps expanding your query universe.

Step 7: Verify your attribution chain before you conclude ROAS changes

Because Apple attribution is install→purchase mapping via AdServices (resolved within ~24 hours), your “revenue from ASA installs” depends on your analytics layer (often RevenueCat or similar).

Before you decide that a batch of promotions/negatives changed ROAS:

• Confirm your reporting reflects the revenue mapping you expect (install → purchase).
• Make sure you’re not comparing net vs gross or excluding refunds/trials in your interpretation.
• Use a consistent time window so you’re not reacting to early purchase patterns that later settle.

Common mistakes (and what to do instead)

• Mistake: Judging Search Match by delivery totals.

  • Fix: evaluate by the actual queries surfaced, using conversion chain metrics.

• Mistake: Promoting everything that has installs.

  • Fix: promote only queries that are clearly efficient (not just “not horrible”).

• Mistake: Adding negatives based on tiny sample sizes.

  • Fix: prioritize negatives with meaningful tap volume or consistent poor conversion.

• Mistake: Changing bids at the same time as keyword set changes.

  • Fix: batch keyword/negative changes first; adjust bids only after you can interpret results.

Optional: how to structure your campaigns as you grow

As your promoted winners pile up, you’ll benefit from a structure like:

• Campaign 1 (Search Match discovery): one ad group, Search Match keyword(s), negatives used to control waste.
• Campaign 2+ (manual winners): one or more ad groups with Exact keywords you promote.

This keeps analysis clean and prevents “organic” improvements from hiding inside a mixed setup.

Closing takeaway

Treat Search Match as discovery, not as your final strategy. Promote the queries that consistently convert and monetize into Exact keywords, and clamp down on the queries that waste taps (or fail to earn revenue) using targeted negatives—done in small batches so you can actually measure impact.

If you’re already pulling your ASA + revenue data into your workflow, this is exactly the kind of change set that AdsBuddy can help you prioritize (you approve and apply everything yourself).