Google Ads search terms: the report that gives budget back

The search terms report is the one place in Google Ads where you literally see what the money bought. Not which keyword you configured — what the person typed before clicking. It's also where the most budget gets recovered for the least effort.

The difference that explains everything

A keyword is an instruction, not an exact filter. Except in exact match, Google decides which searches resemble that instruction closely enough to trigger your ad.

That "closely enough" test has widened over the years. Today a broad match keyword can trigger on searches that share intent according to Google's model but share no word with yours.

Sometimes that brings in excellent searches you'd never have thought of. Sometimes it's pure waste. The only way to know which is happening is to open the report.

What to look for on the first pass

Sort by cost, highest first. It's the fastest way to find the money.

Then read the first twenty or thirty lines one by one and sort each into three buckets.

Irrelevant: the search has nothing to do with what you sell. Goes to negatives.

Relevant but without intent: someone searching "how to do X for free" when you sell X. Depends on the business; in most, a negative.

Relevant with intent: the search you wanted. If it shows up often and converts, it deserves its own keyword and its own ad.

That third bucket is the part almost nobody works, and it's where the real gain sits — not in cutting the bad, but in giving specific treatment to the good.

Negatives worth having from day one

Some patterns show up in nearly every account and can be anticipated.

Free-intent words — free, no cost, trial — when the product is paid. Job words — job, vacancy, salary — when you aren't hiring. Learning words — course, tutorial, how to — when you sell the service done for them. Competitor names, unless you have a deliberate conquest strategy.

Building that list before launch saves the first weeks of exploratory spend, which are usually the most expensive.

Being careful with negative match types

A negative has a match type too, and that confuses a lot of people.

A broad negative blocks any search containing those words. If you add "free" as a broad negative, you also lose "free first consulting session", which might have been exactly your offer.

The safe practice is to add negatives in phrase or exact match when the term can appear in good contexts, and reserve broad for words that will never be relevant.

The mistake here is silent: the campaign simply stops appearing on searches that worked, and nobody notices because there's no alert for what didn't happen.

Where negatives live

They can be added at the ad group, the campaign, or in a shared account-level list.

The practical rule: what never serves the business goes in the shared list, applied to every campaign. What doesn't serve this campaign but serves another goes at the campaign level. And negatives between ad groups exist to route a search to the right group, not to block it.

Keeping one well-maintained shared list is among the things that most simplify a large account. Scattering hundreds of negatives across ad groups means that in six months nobody knows why a search stopped showing.

What the report doesn't show

A share of spend always stays without detail: Google withholds terms with very low volume. In small accounts that share can be substantial.

That's no reason to distrust the report, but it is a reason not to treat the sum of visible terms as total spend — and to pay closer attention to keywords whose spend the visible terms don't explain.

What changes with Performance Max

In Performance Max campaigns the detail level is lower: you see search term categories, not the full list, and the available control is more limited.

That doesn't cancel the review, but it changes its nature. Instead of adding negatives one by one, the work becomes about the signals you give the campaign: audiences, assets, account-level exclusions.

Worth keeping in mind when comparing campaigns: an account mixing standard search with Performance Max doesn't have the same diagnostic capability in both, and expecting the same level of control leads to frustration.

What to do about brand terms

One case that deserves an explicit decision: searches for your own company name.

They show up with excellent cost per conversion and very high conversion rate, which makes the account look better than it is. Some of those people were going to arrive anyway through organic results.

There's no single answer on whether to bid on your own brand — it depends on whether competitors are bidding there. What is mandatory is separating it into its own campaign, so it doesn't distort the read on everything else.

Turning a finding into a result

Finding a good term isn't the end. What produces results is what comes next.

If a term converts well, give it its own keyword and an ad that repeats that exact phrasing. The match improves the estimated quality, which lowers cost per click and improves position — same traffic, cheaper.

And if the landing page speaks to that specific topic instead of the generic offer, conversion rate usually rises noticeably. It's work, but work done once that pays every month.

The pattern that signals a bigger problem

Sometimes the report doesn't reveal a few bad terms — it reveals that the campaign is on the wrong topic entirely.

The signal is when most of the spend sits on searches that are technically related but commercially useless: people researching, comparing, looking for definitions, or looking for a cheaper version of what you sell.

That isn't fixed with negatives one at a time. It means the keywords chosen sit too early in the buying journey for the offer being advertised, and the fix is upstream: different keywords, or a different offer aimed at where those people actually are.

Adding forty negatives to a campaign with this problem produces a tidy report and the same bad results.

How often to review

On new campaigns or those using broad match, weekly — irrelevant spend accumulates fast.

On mature campaigns with exact match and a consolidated negative list, monthly is enough, and the review becomes more about discovery than cleanup.

What doesn't work is reviewing once at launch and never again. Search behavior changes, new products appear in the market, and a campaign cleaned well a year ago may now be paying for things that didn't exist back then.

If cost per click is the symptom, see high CPC on Google Ads. To understand what makes up quality, see Google Ads Quality Score. To run this audit by asking instead of exporting, see Google Ads prompts. And for the weekly read without exporting, see the paid traffic report.