Your keyword list describes what you intended to target. The search-terms report shows the actual searches that caused ads to appear and received enough activity to be reported.
That distinction matters. A sensible keyword can admit searches for different services, locations, customer types, or purposes. The standard report also does not expose every individual query, so treat it as useful evidence rather than a complete transcript.
A non-converting term is not automatically waste. A platform conversion does not automatically prove that a search produced desirable work. The owner question is narrower: Does this search plausibly represent work we want, and what should happen to it?
| What the search reveals | Typical evidence | First action |
|---|---|---|
| Work you want | Right service, customer, location, and situation | Keep and review the resulting leads |
| Clearly out of scope | Wrong service, location, audience, or purpose | Exclude or correct targeting |
| Distinct valuable intent | Relevant demand, but a materially different need or expectation | Give it separate targeting and messaging |
| Relevant but poorly served or unclear | The ad or page misses the intent, tracking is unclear, data is thin, or lead quality is unknown | Investigate before excluding |
Use four actions:
- Keep
- Exclude
- Separate
- Investigate
What “worth paying for” means#
A search term is worth continued consideration when it could lead to work the business actually wants. Conversion status is one piece of evidence, not the definition.
Ask:
- Do we provide this service?
- Do we provide it in this location?
- Does this look like a plausible customer?
- Does the search describe work we want more of?
- Can the likely work support its acquisition cost?
- Did the person receive an ad and page appropriate to the search?
- Can we tell whether the resulting inquiry was real and useful?
Relevance asks whether the search fits the business. Conversion status asks whether the platform counted an action. Lead quality asks whether a real person reached the business with a useful need. A relevant search can fail to convert because the message, page, next step, or measurement is weak. An irrelevant search can still produce a counted action that nobody would want to buy again.
Review repeated patterns before isolated oddities#
Start with the searches that can materially affect a decision:
- searches consuming the most meaningful spend;
- recurring clearly irrelevant patterns;
- recurring commercially valuable patterns;
- searches producing leads that someone actually reviewed;
- unexplained spend that does not fit those groups.
One strange query should not receive more attention than a recurring pattern consuming material budget.
Group terms by the need they express, not merely by shared words. Several differently worded searches may describe the same service situation. Conversely, two searches containing the same service name may imply different customer types, locations, urgency, or next steps.
Google Ads may also provide grouped search-intent insights that summarize themes and subthemes, including some low-volume activity not shown as individual queries. Use those groupings as supporting evidence, not a replacement for reviewing material searches, spend, and actual leads.
Keep, exclude, separate, or investigate#
Keep#
Keep a search when it describes the right work for a plausible customer in a location you serve, and the available lead evidence does not show a meaningful mismatch.
Imagine a service business reviewing a recurring set of searches for urgent repair, routine maintenance, jobs, and repairs in a town outside its service area. The urgent-repair terms may be worth keeping when they produce the kind of calls the business wants. “Keep” means the intent fits the campaign's commercial job, not that the term is exempt from future lead review.
Exclude#
Exclude a search because it is commercially out of scope, not simply because it failed to convert during a limited period.
In the same account, repeated job searches and out-of-area repairs express demand the campaign cannot serve. Exclude them or correct the underlying location targeting because the intent is out of scope, not merely because they failed to convert.
Do not turn those examples into a universal negative-keyword list. “Free,” for instance, could signal research in one account and a legitimate service expectation in another. Judge the search in context.
An exclusion may belong across the account when it is never useful, or within one campaign or narrower segment when another part of the account should remain eligible. Negative keywords prevent certain searches from triggering ads; they do not behave exactly like positive-keyword matching. Choose the level and match deliberately, then verify the effect.
Separate#
Separate a recurring theme when it represents valuable demand but needs a materially different treatment.
Separate only when a recurring, valuable theme represents materially different intent: another service, customer type, eligibility condition, next-step expectation, proof need, or lead result. Similar words do not settle the question. “Emergency repair” from a homeowner and from a commercial property manager may name the same service while requiring different availability, proof, and intake.
Separation may mean different targeting, a campaign or ad group, ad language, offer, or landing page. A separate page is not automatic; the customer intent must be genuinely distinct.
Investigate#
Investigate when the search is plausible but the available evidence cannot support a confident decision.
Suppose the example account shows a few searches for an unfamiliar repair phrase in the correct location. The work may fit, but there is little data and no reviewed lead result. Excluding the phrase now could remove useful demand before anyone understands it.
Check the ad shown, destination, location setting, cost, actual inquiry, and whether a counted conversion was a real lead. Investigation should end with a clearer classification or a specific repair.
When a relevant search is poorly served#
A search can fit the business and still perform poorly because the current ad and page answer a broader or different question.
An urgent-repair search may receive a routine-maintenance ad. A fitting ad may lead to a broad page. The page may fit while its form asks questions appropriate to a large planned project. In each case, the search itself can still be valuable.
Do not punish the search for a message problem. Compare the recurring valuable group with the ad and destination it actually received. The search-to-lead offer-alignment guide shows how to trace that path and fix the earliest mismatch before expanding the campaign or creating more pages.
Likewise, a platform conversion does not establish value when the resulting call is spam, a job seeker, an existing customer seeking support, or unwanted work. Connect the search theme to real lead quality and business fit.
A repeatable search-term review#
Use a compact process:
- Choose a date range with enough activity to review.
- Start with searches consuming meaningful spend.
- Mark each material term as keep, exclude, separate, or investigate.
- Group repeated searches that express the same intent.
- Compare valuable groups with their ads and destination pages.
- Review whether the resulting leads were real and desirable.
- Fix the largest repeated mismatch and review again.
The right review interval depends on account volume. A low-volume local account may need more time before patterns are credible. A higher-volume account may reveal repeated mismatches sooner. Do not impose a calendar that creates decisions from thin evidence.
Stop when every material pattern has a defensible keep, exclude, separate, or investigate action. The output is not a universal negative list; it is one focused correction backed by repeated searches and reviewed lead results.