Apple Ads campaign and keyword strategy

A useful search results structure in Apple Ads (formerly Apple Search Ads) separates known intent from discovery. Brand, category, competitor, and discovery campaigns answer different questions, require different bids, and should not compete for the same exact search terms. Structure is valuable only when it makes decisions clearer.

• Use four intent themes
• Separate exact match, broad match, and Search Match
• Use negatives to route or exclude demand
• Harvest search terms deliberately
• Make bid changes from a constraint diagnosis
• Budget for evidence without losing control
• Use evidence thresholds, not magic numbers
• Worked example: harvest without overlap
• How AdPilot supports the workflow
• Strategy checklist
• Next steps
• Official sources

Use four intent themes#

Apple’s campaign structure guidance recommends four common themes:

Campaign type	Intent	Typical setup
Brand	App or company name	Exact match, Search Match off
Category	Problem, use case, or app category	Exact match for proven non-brand terms
Competitor	Similar or alternative apps	Carefully selected exact terms
Discovery	Unknown relevant demand	Broad-match and Search Match ad groups

Brand traffic often behaves differently because the customer already knows the product. Keep it separate so inexpensive branded installs do not make generic acquisition look healthier than it is.

Category terms express what the customer wants to accomplish. Competitor terms can be relevant but may have weaker conversion and additional policy or brand considerations. Discovery buys information about queries you have not promoted as exact keywords yet.

Separate exact match, broad match, and Search Match#

Exact match offers tighter control while still allowing close variants under Apple’s matching system. Use it for terms with a clear hypothesis and a bid based on their expected value.

Broad match expands a keyword into related queries and variants. It can discover demand, but it can also mix different intents. Review the actual search terms rather than judging only the seed keyword.

Search Match lets Apple match the app automatically to relevant searches using App Store information and other signals. Keep it in its own discovery ad group so its results are visible and controllable.

A clean discovery campaign commonly has:

• a broad-match ad group containing useful seed keywords;
• a Search Match ad group with Search Match on and no targeting keywords;
• exact-match negatives for terms already owned by brand, category, or competitor campaigns.

Use negatives to route or exclude demand#

Negative keywords serve two jobs:

1. Routing negatives keep an exact term in the campaign where you intentionally bid on it.
2. Relevance negatives block queries that do not match the product or economics.

Exact negatives block the specified term without broadly excluding related demand. Broad negatives are stronger and should be used carefully because every included word must be present for Apple to block the search, regardless of order.

Do not create a negative only because a term has spent money. Check intent, taps, installs, lag, and sample size. A relevant query with limited evidence may need more observation; an obviously irrelevant query can justify action earlier.

Harvest search terms deliberately#

Keyword harvesting turns discovery evidence into controlled exact targeting:

1. Review search terms over complete days.
2. Confirm the term is relevant to the app and target market.
3. Check installs, CPI, conversion, and post-install evidence when available.
4. Add a promising term as an exact keyword in the appropriate campaign.
5. Add the same term as an exact negative in discovery to reduce overlap.
6. Set a bid supported by value and current auction evidence.
7. Re-evaluate after enough comparable data accumulates.

Keep a record of why the term moved. Harvesting everything with one install can create a large, noisy keyword inventory without improving control.

Make bid changes from a constraint diagnosis#

Increase a bid when the term is valuable, has room to capture more qualified volume, and is plausibly auction-constrained. Lower a bid when cost exceeds value and relevance remains strong enough to keep testing. Pause when the economics, relevance, or evidence no longer justify exposure.

Before changing a bid, review:

• serving and keyword status;
• current bid and ad-group default;
• impressions, taps, installs, and CPI;
• search-term intent and match source;
• product-page conversion;
• target CPA, payer CAC, ROAS, or payback;
• country, audience, and date-range comparability;
• whether another campaign is matching the same term.

Do not use a suggested or competitive bid as proof of profitability. It describes potential auction competitiveness, not customer value.

Budget for evidence without losing control#

A test budget should be large enough to observe meaningful behavior but small enough that a failed hypothesis is affordable. Daily budget, bid, traffic availability, and conversion rate jointly determine how quickly evidence arrives.

Define stop conditions before launch, such as:

• a maximum spend with no installs for a clearly measured term;
• a CPI materially above the accepted range after sufficient volume;
• a relevance failure visible in search terms;
• a serving or mapping problem that invalidates the test;
• a workspace or app exposure cap.

AdPilot’s limits are authorization guardrails, not a substitute for the test design. The maximum projected month can be higher than recent spend because it considers possible daily-budget exposure through month end.

Use evidence thresholds, not magic numbers#

There is no universal tap count that proves a keyword works or fails. Required evidence depends on baseline conversion, cost, value, variance, and the size of the decision.

Use stronger evidence for changes that are expensive or difficult to reverse. Consider:

• absolute taps and installs, not only percentage changes;
• whether the comparison periods have equal length and cohort age;
• how much one additional install would change CPI;
• whether tracking, mapping, or sync freshness is incomplete;
• whether product, pricing, creative, or seasonality changed;
• post-install quality when available.

AdPilot’s action evaluation labels low-volume results as inconclusive rather than forcing a success or failure verdict.

Worked example: harvest without overlap#

A discovery ad group matches the search term shared family calendar through broad match. Over 30 complete days it records 60 taps, 12 installs, $48 spend, and a $4 CPI. The term matches the product and compares favorably with the app’s $7 install target.

A disciplined action is to:

1. create [shared family calendar] as an exact keyword in the category campaign;
2. choose a starting bid supported by its current average CPT and target economics;
3. add shared family calendar as an exact negative in discovery;
4. record the source metrics and date range;
5. compare the exact keyword after a similar complete-day window.

If RevenueCat only provides blended app proceeds, do not claim this individual term produced a specific ROAS. Use install evidence for the keyword and app-level proceeds as supporting context.

How AdPilot supports the workflow#

AdPilot skills can review discovery terms, prepare keyword and negative-keyword changes, tune bids, diagnose under-delivery, and evaluate prior operations. The assistant is expected to load the matching skill before a multi-step task and to separate preparation from an explicit apply request.

Direct tools execute eligible requested changes immediately after authentication, workspace scoping, correctness checks, and limit enforcement. Changes are recorded with the Apple entity IDs, action ID, inputs, result, and rollback eligibility.

Try: Review discovery search terms for the last 30 complete days. Propose exact keywords to harvest and routing negatives, but do not apply anything. Include evidence and explain terms that need more data.

Strategy checklist#

• Keep brand performance separate from generic acquisition.
• Give discovery broad match and Search Match distinct ad groups.
• Route harvested exact terms out of discovery with exact negatives.
• Check actual search terms before changing a seed keyword.
• Tie bids to value and constraint diagnosis.
• Set budget and stop conditions before testing.
• Use complete days and comparable periods.
• Keep app-level profitability context separate from keyword attribution.
• Evaluate recorded changes instead of repeating them blindly.

Next steps#

• Apple Ads guide
• Apple Ads metrics and formulas
• AdPilot MCP skills for Apple Ads
• Direct operations, limits, and audit

Official sources#

• Apple Ads campaign structure best practices
• Understand Apple Ads keyword match types
• Use negative keywords in Apple Ads
• Apple Ads glossary