Quick answer
A negative keyword is a decision about which searches a campaign should stop serving. The difficult part is not finding words that look uncommercial. It is excluding the wrong task without excluding a relevant buyer who happens to use the same words.
This guide is for UK business owners and PPC managers reviewing Google Ads exclusions. It provides a policy for judging complete queries, choosing match type and scope, and testing examples that would challenge an exclusion. Use it before importing a shared list, expanding an existing negative or reviewing a campaign with unexpected gaps in relevant traffic.
- Use the narrowest justified exclusion. Base the decision on the complete query, the offer and the affected campaign, rather than a word in isolation.
- Separate irrelevance from weak performance. A relevant query without recorded conversions needs a different explanation from a request the business cannot fulfil.
- Test valuable counterexamples. Check whether the proposed negative would also block a plausible product-selection or service enquiry.
- Inspect every affected scope. A sensible exclusion in one campaign may be harmful in another campaign sharing the list.
- Record the reason and reversal route. Someone else should be able to understand, review and undo the decision.
Pause a bulk negative upload if the campaign scope, match types or buyer-query collisions have not been checked. A large list is not evidence of a well-controlled account.
What This Guide Does Not Solve
This is a search-query exclusion policy, not a complete bidding, attribution or account-restructure plan. It does not supply a universal blacklist or recommend excluding every query beginning with “how to”. It cannot establish profitability without suitable performance and business evidence.
The match examples below concern Search keyword matching. Do not apply them as a complete model of Display or Video exclusions. Google documents different behaviour for those campaign types. Campaign features and available controls should be confirmed in the account before implementation.
When to Pause or Escalate
- Pause performance-based exclusions when conversion recording is unreliable or the reporting period is too recent for the business’s normal decision cycle.
- Escalate a shared-list change when another campaign sells a product or service that the proposed exclusion could block.
- Stop automatic approval if a candidate removes an agreed protected query, even when the candidate also blocks many unwanted searches.
- Request an account review when the same intended query is blocked at several levels and the team cannot explain which restriction should remain.
Judge the Buyer’s Task Before the Wording
Classify the complete search according to the relationship between the task and your offer. Useful starting labels are relevant purchase, relevant buying research, outside the offer and uncertain. These are working labels for this method, not Google classifications.
A query asking how to choose a product can represent someone comparing purchases. A query asking how to manufacture the same product at home can represent a different task. The shared opening words do not settle whether either query belongs in a particular campaign.
The business model matters. “Repair” may describe the service a contractor sells, an application for a retailer’s materials or a request outside a replacement-only provider’s offer. Write down what the campaign is intended to attract before labelling the word itself irrelevant.
Ask the sales or product team to review ambiguous intent. The GOV.UK guidance on understanding user needs supports using evidence about users’ goals rather than treating internal assumptions as established needs. Here, that means preserving uncertainty until the task and offer can be compared.
Separate three reasons for excluding a query
Offer mismatch: the business does not supply the requested product, service or task. Confirm that the apparent mismatch is not simply unfamiliar customer vocabulary.
Campaign mismatch: the business serves the need, but the particular campaign is not intended to do so. This may justify a local exclusion, provided the wider account decision is deliberate.
Performance decision: the query is relevant, but the evidence supports limiting its spend under the campaign’s objectives. Document the evidence and review conditions. Do not relabel a relevant buyer as irrelevant merely to simplify the record.
Match Type Determines Which Queries the Decision Reaches
Google’s Search negative match types work differently from positive keyword matching:
- Negative broad: blocks a search containing all the negative’s terms, including when their order differs.
- Negative phrase: blocks a search containing the negative phrase in the same order, with additional words permitted.
- Negative exact: blocks the matching search without extra words; a longer query is not excluded by that exact negative alone.
Do not assume positive-keyword expansion applies. Google’s current guidance says synonyms and singular/plural forms need separate consideration, while casing and misspellings are handled automatically. Check current documentation and account behaviour for unusual strings rather than assuming a simple spreadsheet reproduces every matching detail.
“Not blocked by this negative” does not mean “the ad will show”. Other restrictions, targeting and auction conditions still matter. Keep that distinction visible in any internal test or decision record.
Choose Scope Before Approving the Exclusion
Define where the exclusion is justified. A campaign selling new equipment may have a different purpose from one advertising repairs. If a request is unwanted only in the new-equipment campaign, an account-wide rule would express the wrong business decision.
Shared negative lists apply changes to the campaigns using the list. Check the actual attachments, not just the list name. A list called “irrelevant searches” may contain decisions made for an old product range that are no longer appropriate elsewhere.
Account-level negatives apply across relevant Search and Shopping inventory. Reserve that scope for exclusions that remain valid across the affected business activity. Do not use the widest scope merely because it is easier to administer.
Google also supports campaign-level negatives in Performance Max, but those negatives apply to Search and Shopping inventory only. They are not a universal content-exclusion switch for every placement in the campaign. Confirm the available controls and the intended inventory before making changes.
Pause approval when a relevant query collides with the candidate. Narrow the term, match type or scope only if the revised rule expresses a supported decision. Otherwise retain the query or investigate further.
A narrower match type is not automatically the right answer. If the full query itself is commercially relevant, changing a broad exclusion into an exact exclusion still removes that query. The decision must be justified before its reach is adjusted.
Build a Counterexample Test Set
1. Capture the candidate and its evidence
Record the complete observed query, campaign, reporting dates, offered product or service, candidate negative and proposed match type. State whether the query is observed or invented for testing. Include the reason for exclusion in plain language.
Review conversion evidence with its limitations. “No recorded conversions” is an observation, not proof that no customer bought or enquired. Check the relevant conversion actions, reporting maturity and whether a known website or tracking issue affects interpretation. This is a reason to investigate, not a claim that hidden conversions must exist.
Google omits some low-activity queries from the search terms report for privacy. Treat the report as evidence of the queries it reveals, not a complete record of every search. Do not interpret the absence of a query as proof that it never occurred.
2. Identify queries the rule should protect
Include known relevant searches where available, important product or service phrases, and plausible buying questions. A protected query is one the team has agreed should not be removed by this exclusion; it is not necessarily a proven profitable query.
Challenge the candidate with a longer purchase query, different word order, a relevant use of an ambiguous word and a query belonging to another affected campaign. Test synonyms or singular/plural differences where they matter. The set should challenge the business decision as well as the matching logic.
Do not build the set only from searches the proposed rule already handles well. Ask a colleague to contribute the strongest reasonable buyer example that contains the excluded term. Record the intended offer alongside that example so the disagreement can be resolved.
3. Record expected exclusions and collisions
For every test query, document the intended outcome and whether the candidate is expected to block it under the stated match rule. Keep observed platform behaviour separate from a desk-based prediction. Label a predicted collision as predicted until it has been checked through an appropriate account review.
Use three decisions: approve when the reason and tested reach are supported; revise when a narrower or differently scoped candidate solves the problem; and investigate when intent or evidence remains unresolved. “Investigate” is a useful result, not a failure to complete the review.
4. Check the complete exclusion set
A new narrow rule will not protect a query if another existing rule still blocks it. Review relevant account settings, attached lists and direct campaign or ad-group negatives where applicable. Identify every rule that conflicts with the protected set.
Keep a copy of the current configuration and list attachments before changing them. Document which rule is being replaced and which restrictions deliberately remain. Do not assume that removing one visible negative has removed the entire conflict.
Worked Example: Why a “How To” Exclusion Can Be Too Wide
This is a fictional policy exercise, not a client result or an account test performed by KAP SEO Services. Assume a retailer sells reusable storage containers. Its campaign is intended to reach product buyers, including people comparing suitable container sizes. The retailer does not sell printable craft templates.
The proposal is to add how to as a negative broad keyword. The following table predicts the effect of that candidate alone using the stated Search match rules. It does not predict impressions, sales or auction eligibility.
| Full test query | Assumed buyer task | Blocked by this candidate? | Policy decision |
|---|---|---|---|
| how to choose storage containers for a small cupboard | Potentially relevant product-selection research | Yes: both negative terms appear | Collision with a protected buying question |
| how to make paper boxes free printable template | Craft-template task outside this retailer’s offer | Yes: both negative terms appear | Exclusion may be justified for this complete query |
| storage containers to buy and how sizes compare | Potential purchase and size comparison | Yes: word order does not protect it from negative broad | Second relevant-query collision |
| buy storage containers with lids | Direct product purchase | No: both terms are not present | Not blocked by this candidate alone |
The broad proposal should not be approved under the exercise’s stated objectives because it conflicts with protected buying research. A negative phrase using the same two words would still remove the first buying question. Changing match type without testing the actual counterexample would miss that problem.
A possible revision is a negative exact for how to make paper boxes free printable template, restricted to the relevant campaign. That revision addresses the complete unwanted request while leaving the other displayed queries outside that exact rule. It does not establish that every similar craft query will be excluded or that no other account rule affects the protected queries.
The example depends on its commercial assumptions. A shop selling paper-craft kits might judge the template query differently. A campaign deliberately limited to immediate purchase requests might choose a different research policy. Both require an explicit business decision, not a universal blacklist.
A defensible record would state: reject the broad candidate because of two relevant collisions; review the full craft-template query as a narrow exclusion; check all attached lists; record conversion evidence as not supplied; and assign a review owner. No percentage saving or revenue improvement can be inferred from this exercise.
Avoid Decisions That Look Efficient but Hide Uncertainty
Do not approve by list size. Adding hundreds of negatives may be easy, but volume does not demonstrate that the excluded tasks are unsuitable. Review themes and the most consequential collisions before importing a list.
Do not treat research language as proof of non-buying intent. A buyer can ask about selection, compatibility, dimensions or application before making a purchase. Decide whether the campaign and destination serve that task.
Do not invent a universal click or cost threshold. Relevant performance decisions depend on the objective, economics and available evidence. A fixed rule copied from another account does not establish the right decision for this business.
Do not repair a poor landing page with exclusions alone. If relevant queries consistently reach a page that cannot answer their needs, review the destination. Blocking the queries can conceal a customer-journey problem instead of resolving it.
Do not call every changed result an improvement. Lower spend can accompany lost opportunity. Review query relevance, qualified outcomes and the changes made elsewhere before attributing a movement to the negative policy.
Maintain the Policy as the Business Changes
Give each approved exclusion or coherent exclusion theme an owner, reason, scope and review trigger. Triggers can include a new service, a changed product range, a campaign split or evidence that a protected query is no longer appropriate.
For a release, record the date, previous configuration, changed entries, list attachments and expected effect. If a conflict appears, restore or revise the specific rule responsible. Avoid removing an entire established list merely because one entry is wrong.
Review outcomes after a period appropriate to the business’s conversion cycle and reporting coverage. Compare the unwanted queries being excluded with relevant activity that may have been constrained. A before-and-after comparison is descriptive evidence; it does not isolate causation when budgets, offers or other campaign settings changed too.
Keep expired decisions visible in the change history, with the reason they were removed. Otherwise the same negative can return in a later bulk upload and recreate the conflict. The policy should preserve the reasoning, not just the current list of words.
How to Get This Done
Prepare the campaign objectives, relevant search terms, current negatives, shared-list attachments and a protected-query set. Include the reporting dates and known measurement gaps. Ask for a decision record that distinguishes offer mismatch, campaign mismatch and evidence-based performance restrictions.
A Google Ads measurement and campaign review can assess targeting, conversion tracking and campaign priorities together. Agree whether the work covers recommendations, implementation and follow-up checks before changes are made.
The handover should explain what will be excluded, why the scope is justified, which buyer queries were tested and how the change can be reversed. Where the evidence cannot support a decision, keep the candidate in an investigation queue.
Summary
A negative keyword policy is strongest when every exclusion has a supported purpose and a tested reach. Judge complete queries, distinguish campaign needs, challenge candidates with relevant counterexamples and check existing restrictions. Preserve uncertain cases for investigation rather than turning missing evidence into a blanket exclusion.
Frequently Asked Questions
Can an informational-looking query still lead to a purchase?
It can express a relevant buying task, such as choosing a size or comparing products. Wording alone does not prove conversion potential. Assess the task, the offer and the evidence rather than excluding a whole opening phrase.
Should an uncertain term use a narrower exclusion?
Only when exclusion of the complete query is justified. A narrower rule reduces its reach, but it can still exclude a relevant query. Investigate first when the business case remains uncertain.
Can a shared list contain a rule that is wrong for one campaign?
Yes. A rule may suit the campaigns it originally served but conflict with another campaign attached later. Review both the entries and the current attachments whenever the list’s purpose changes.
Does a negative exact protect every longer buying query?
That exact negative alone does not exclude a longer query with extra words under the ordinary matching rule. Another negative or account restriction may still affect it. Test the complete applicable exclusion set.
Is the absence of a search term proof that it was blocked?
No. The search terms report is not exhaustive, and absence alone does not establish the cause. Review the relevant configuration and evidence without treating missing rows as proof of either successful exclusion or lost demand.
Are Your Negatives Blocking Relevant Buyers?
Ask KAP SEO Services to review the campaign objectives, proposed exclusions and buyer queries you want to protect.
