Google AdsA search term with no recorded conversions is a question to investigate. It is not automatically a negative keyword. The useful distinction is between a search your business cannot serve and a relevant search for which the evidence is still incomplete.
This workflow uses an imaginary paid workforce-scheduling app. It sells software to businesses, offers a trial and has no spreadsheet-template product. Download the search-term review sheet to document the decision before making changes in an advertising account.
Export a report with its context
Open the search terms report for the campaign you are reviewing. Keep the date range, campaign, ad group, search term, matched keyword where available, cost and relevant conversion measures together. Record the conversion action and reporting window outside the export if those details are not included.
Google's search terms documentation distinguishes the user's actual query from the keyword that matched it. Some low-activity queries are omitted for privacy. Your export therefore is not a complete transcript of every search that consumed spend.
Do not compare yesterday's queries with a mature month's sales without accounting for conversion delay. Write a review cutoff based on the time your own customers take to convert. The qualified-lead guide helps define which downstream event belongs in that review.
Give each term a reasoned status
Use three statuses: relevant, irrelevant or unresolved. Add a plain-language reason that someone outside the media team could understand. “Bad CPA” is a result description; “we do not offer payroll certification courses” explains a mismatch.
| Example query | Initial status | What to verify |
|---|---|---|
| employee scheduling software | Relevant | Does the page explain the scheduling product? |
| free employee scheduling trial | Relevant | The business offers a trial; do not block it merely for containing “free” |
| staff rota spreadsheet template | Potentially irrelevant | Confirm there is no template offer or planned template landing page |
| scheduling app for restaurants | Unresolved | Check whether restaurants are supported and whether the page addresses them |
These are example decisions for this hypothetical business. A company that uses a free template as a legitimate acquisition offer should reach a different conclusion about the third row.
Choose the smallest justified exclusion
For Search, Google supports negative broad, phrase and exact match. Negative matching differs from positive keyword matching, and close variants are not automatically covered. Consult the current negative keyword rules before approving a list.
Consider the proposed negative free. It would be a much broader business decision than excluding one irrelevant query about a downloadable template. In this example, it risks removing people who are actively looking for the available trial.
Write both the proposed negative and its match type in the worksheet. Add the intended scope: ad group, campaign or an applicable shared/account list. A term that is irrelevant to one product line might be valuable to another, so avoid turning a local decision into an account-wide exclusion without checking the other offers.
Review conflicts before applying the batch
Read the candidate exclusions alongside the positive keywords, actual offer and active landing pages. Have the campaign owner review ambiguous brand names, research queries and phrases related to future products. Keep the exact previous list or a dated export so the owner can identify what changed.
Google's report-based negative keyword instructions explain the account controls. This article's worksheet prepares the choices; it does not connect to an account or apply an exclusion. Use the interface preview and applicable scope shown in your account as the final check.
For a small batch, record an approver, the application date and the intended effect. “Exclude searches for training courses from the software campaign” is more reviewable than “improve traffic quality.” Avoid mixing an exclusion batch with unrelated changes to bids, the conversion goal and the landing page if you need to interpret the result.
Check the effect without inventing savings
At the agreed review date, compare the relevant traffic and business outcome under consistent definitions. Record whether desired searches disappeared, whether newly visible queries suggest another mismatch, and whether conversion quality changed. Missing queries in a privacy-filtered report cannot establish that all unwanted traffic has stopped.
Do not report historical spend on excluded queries as guaranteed future savings. Demand, auction conditions and matching can change. Report the actual configuration change and subsequent observed spend separately.
Attach the sheet to the Google Ads launch checklist or carry it into a supported AdRails review. The next person should see the search evidence, proposed exclusion, scope and approval in one place.
Sources checked October 14, 2026. The query classifications are illustrative, not a universal negative keyword list.