Microsoft Ads Match Types: Search Terms and Negative Controls

#01What does a positive match type really control?

Treat a Search keyword as an input to matching, not a promise that only its literal text will trigger an ad. Microsoft's broad match can relate searches through meaning even when the exact keyword words are absent. Phrase and exact also use meaning and close variants rather than functioning as plain spreadsheet text comparisons.

This distinction matters when reviewing intent. An exact keyword can still require a query-quality review, while a broad keyword is not automatically careless if the business has suitable measurement, offers and boundaries. Choose a match approach based on the job, evidence and risk tolerance, not an assumption that one label always produces better customers.

Legacy syntax can mislead inherited accounts. Microsoft's documentation states that old broad-match-modifier keywords now serve as broad. The presence of plus signs in an old export is therefore not evidence that a historical modifier still narrows delivery as an operator remembers.

The Google match-types lesson provides useful conceptual background. Do not transfer every precedence rule or automated matching control. Microsoft settings, including any current AI Max features, determine the destination account's actual behaviour.

#02How do Microsoft negative rules differ?

Microsoft supports negative phrase and negative exact matching. A phrase negative blocks the specified sequence in the relevant search context with additional words possible; an exact negative blocks the specified sequence without additional words. These are not the same semantic expansion rules as positive phrase and exact keywords.

Microsoft explicitly notes that negative keywords do not automatically account for synonyms, plurals or misspellings. If the business needs to exclude related variants, review and add the relevant forms deliberately. Do not assume a single negative such as template covers every similar research query.

The examples are semantic review prompts, not universal exclusion lists. The same phrase can be irrelevant for one merchant and valuable for another. Maintain a permitted-query test set alongside the blocked-query set so exclusion quality is measured in both directions.

#03Which exclusion layer should you use?

Inventory account-level negatives, shared lists, campaign negatives and ad-group negatives. Their effects can accumulate. A reviewer examining one ad group may miss an account-level exclusion that already blocks its intended demand.

Microsoft does not support keyword-level negative keywords, even if an external interface permits a user to enter something that looks like one. Establish where the rule actually lives in Microsoft. An apparent control that never reaches the supported scope creates false confidence.

Choose the narrowest scope that expresses the commercial requirement. A merchant that never sells software may justify a broad business-wide exclusion family after review. A term irrelevant only to one product group should not automatically become an account-wide block that harms another offer.

Shared lists need ownership and association checks. Record the list's purpose, campaigns attached, review date and source. A generic agency list can include legacy assumptions from another client or market. Importing it without review is not a professional shortcut; it is an untested change to commercial reach.

#04How should you read the search-term report?

Keep query, keyword and match behaviour separate. The query is what the report records from a search. The keyword is an advertiser input. The match label explains part of the relationship but does not, by itself, tell you whether the search was commercially useful.

Microsoft's search-term report has coverage limitations and is not a complete census. The documentation warns that report totals can represent only part of broader impressions. Do not conclude that unlisted queries never occurred, or calculate a universal irrelevant-query percentage as though the visible rows were the entire population.

Review the available rows with cost, conversions, revenue where valid, destination and commercial outcome. A high-cost query with no recent conversions may deserve attention, but conversion lag and sparse evidence matter. A low-cost query can still deserve immediate exclusion if it concerns an offer the business cannot truthfully fulfil.

Use reason codes rather than a single good-or-bad flag. Categories such as unsupported product, wrong buyer, research intent, ambiguous term, misleading destination and valid low-evidence demand lead to different actions. Some need negatives; others need a clearer ad, a better page or more observation.

#05How do you avoid blocking useful demand?

Build a test matrix before broadening a negative. Include intended blocked searches, intended permitted searches and ambiguous searches requiring a human decision. Test singular, plural, reordered and additional-word versions where relevant, because negative semantics are not automatic language understanding.

For an illustrative Ledgerline free-download exclusion, useful permitted searches might include cable tray with free delivery. Blocking the word free globally would conflict with a genuine shipping proposition. A more specific exclusion family can address downloadable templates without suppressing that buying intent.

Review cross-product effects. A phrase irrelevant to a clamp-mounted tray could still matter to adhesive clips or planner accessories. Account-wide scope may save maintenance but increase false positives. If the team cannot explain why every affected product should reject the intent, choose a narrower layer or gather evidence.

Keep the change reversible and documented. Record the triggering query, intended effect, scope, variants reviewed and approval. After an authorised change, review both reduced irrelevant traffic and potential loss of useful demand. A fall in spend alone does not prove the negative improved the business.

#06What changes when Google Import is involved?

Microsoft documents that Google broad negatives import as phrase negatives. This is a mapping, not equivalent matching. A source exclusion policy can therefore behave differently even when the imported text looks identical.

Check shared-list content and associations after import. An import can merge or update list relationships, and later scheduled runs can restore source-controlled settings. A local repair may not survive if nobody changes the schedule's field ownership.

Create an import-specific negative audit with the original match type, destination match type, scope and intended query family. Test the most commercially consequential exclusions first. Do not attempt to rebuild source semantics by adding every imaginable variant indiscriminately; that can create a larger and more damaging block list.

The Google negative-keyword strategy is a useful planning reference, but the final acceptance test must use Microsoft rules. The destination's actual saved associations and settings, not the source spreadsheet, establish what is enforced.

#07How do you operate a sustainable query review?

Choose a cadence based on spend exposure, novelty and risk. A newly imported campaign with unfamiliar matching may need closer review than a stable product campaign, but there is no universal number of days that makes every query judgement valid.

Separate urgent issues from optimisation questions. A disallowed market, unsupported product or misleading claim can justify prompt corrective review. A valid query with uncertain economics should be evaluated with its conversion cycle and evidence quality, not automatically excluded after one expensive click.

Share useful language with product and content teams. Search terms can reveal missing compatibility explanations or demand for unavailable variants. Preserve the evidence and its limitations. Query data can guide a page brief without proving market size or guaranteeing that an AI engine will cite the resulting content.

#08How should a shared-list change be reviewed?

A shared-list request needs a blast-radius check. List every associated campaign and identify which offers could lose eligibility. Review the proposed text in each relevant language and retain examples of valuable searches that must remain possible. An exclusion justified by one cable-tray query should not automatically govern a planner campaign with a different use for the same word.

Use a two-person review for consequential changes where the organisation's process permits. The requester supplies the commercial reason and semantic tests; the reviewer challenges scope and false positives. Keep the approved version and the previous state in the change record so an authorised operator can reverse the change if evidence contradicts the hypothesis. This is a governance recommendation, not a new platform feature or a claim that Microsoft automatically validates the business meaning of every list.

#09Worked example: Ledgerline

Ledgerline sees reported searches about downloadable desk-planning templates alongside legitimate searches for its physical planners. An operator proposes an account-wide negative for free. The illustrative review finds that the store also advertises a real free-delivery offer under defined conditions.

The team rejects the blanket rule. It reviews specific template and download phrases, checks their variants and limits exclusions to the relevant campaign where appropriate. It also improves the ad's physical-product wording so customers understand the offer before clicking.

Walkthrough

1. Export available search terms with matched keyword, cost, destination and outcome context.
2. Mark unsupported downloads separately from valid physical-product research.
3. List all existing account, shared, campaign and ad-group negatives.
4. Draft specific phrase and exact candidates with blocked and permitted query examples.
5. Review the effect on free-delivery and product-compatibility searches.
6. Check whether an import schedule could restore a conflicting list or association.
7. Record the proposed change, reviewer and post-change evidence needed before claiming improvement.

The example does not claim an observed reduction in cost or increase in sales. It demonstrates a safer decision process and a clearer definition of exclusion success.

#10Common mistakes

Treating exact positives as literal-only controls leads teams to stop reviewing queries. Treating phrase negatives as semantic equivalents of positive phrase matching creates the opposite error: relevant variants remain unblocked while useful exact phrases may be suppressed.

Another mistake is adding a negative at the broadest available scope because it is easier to maintain. Scope should follow business relevance. Finally, avoid using one visible report row to make a claim about every search, every customer or the platform's algorithm. Report coverage and attribution limits remain part of the evidence.

#11Try it yourself

Choose one product campaign and build a negative-control worksheet without making changes. Include five illustrative blocked queries, five permitted queries and three ambiguous ones. For each proposed negative, state match type, scope, variants considered and the business reason.

Ask a colleague to find a valuable query that your proposed rule would block. Revise the rule or explain why the loss is justified. Then inspect any source-controlled import or shared-list relationship that could overwrite the decision. A complete exercise includes a protection plan, not just a longer negative list.

#12Quick quiz

1. Does Microsoft automatically expand negative keywords to every plural or synonym?

Answer: No. Relevant negative variants need deliberate review and configuration rather than assumptions based on positive matching.

2. Is a Microsoft keyword-level negative a supported control?

Answer: No. Microsoft documents supported account, campaign, shared-list and ad-group structures, not a keyword-level negative scope.

3. Why is a search-term export not a complete query census?

Answer: Microsoft applies reporting coverage filters, so visible rows and totals do not necessarily represent every query or impression.

#13What should your next review preserve?

Preserve the intent decision, scope, source evidence and permitted-query tests. An exclusion is successful only when it reduces unwanted eligibility without an unjustified loss of useful demand.

Previous: Microsoft Ads Keyword Research: Planner, Intent and Evidence · Next: Microsoft Ads Structure: Locations, Language and Distribution

Or return to the Microsoft Ads for Marketers track.

Related: Microsoft Ads Google Import: Mapping, Checks and Safe Cutover · Microsoft Ads Keyword Research: Planner, Intent and Evidence · Microsoft Ads AI Max and Copilot: Governed Paid Discovery

#14Common questions

Does exact positive matching mean literal text only? No. Microsoft positive matching can use meaning and close variants. Review reported queries and commercial fit even when the configured keyword is exact.

Do negative keywords include plurals automatically? No. Microsoft does not automatically expand negatives to every plural, synonym or misspelling. Review the relevant variants and permitted-query tests deliberately.

Can I inspect every search in the search-term report? No. Microsoft documents report coverage limits. Visible rows are useful evidence but not a complete census of queries or impressions.