What is Search term?

A search term is the actual query. A keyword is the advertiser's request to appear for something like it.

For most of the history of search advertising those were nearly the same thing, because match types were literal. An exact-match keyword was triggered by that keyword. The report was a formality.

That has changed completely. All three match types now match on interpreted meaning, which means a keyword can be triggered by searches that share none of its words. The keyword list describes intent; the search-term report describes what happened. They are different documents and only one of them is evidence.

Every other view in a search account is a summary of things the advertiser chose. Keywords, match types, ad groups, budgets, bids — all of them are inputs.

The search-term report is the only output that says what the money was spent on, query by query. It is where an account discovers that a keyword is reaching an adjacent product, that a campaign is buying research traffic, that a competitor's name is being matched as a category, or that a term nobody thought of is converting better than anything on the list.

None of that is inferable from the keyword list, and none of it appears in a cost-per-conversion figure, which averages the good and the bad into one number that looks acceptable.

The report rewards a specific way of reading it, and it takes twenty minutes.

Sorted by spend, descending. Read from the top and stop when the amounts stop mattering. This finds where the money actually went, and in most accounts the expensive irrelevant queries are in the first page or two. Everything irrelevant here is a negative.

Sorted by clicks with no conversions. This finds the steady small leaks — terms that individually cost little and collectively cost a great deal, which the spend sort buries.

Both passes produce negatives. A third pass, sorted by conversions, produces the opposite: terms worth promoting into their own keywords.

The most valuable output of the report is not the negative list. It is the set of terms that convert.

A search term converting consistently through a broad or phrase keyword is traffic the account is buying with a generic ad and a generic page. Promoting it to an exact-match keyword in a group whose ad speaks to it directly does three things: it improves relevance, which improves rank, which lowers the price; it lets the landing page answer the query specifically, which improves conversion rate; and it moves that traffic out of a broader keyword whose figures were an average across a mixture.

This is the cycle that makes a search account better over time. Broader matching discovers, the report reveals, promotion structures, and negatives refuse. An account doing only the last part is defending itself and not improving.

Platforms withhold search terms below certain volume thresholds and terms that raise privacy considerations. A portion of spend therefore appears against no visible query.

The share varies by account and it is not negligible. It means the negative list can only ever cover what was seen, and that an account cannot fully audit its own spend from this report.

Two responses. A conversion-based bid strategy declines auctions it does not expect to convert, including searches nobody will ever see, which covers part of what negatives cannot reach. And watching the campaign's aggregate cost per conversion catches drift in the invisible portion that the visible rows do not explain.

Two habits change when an account moves toward broader match types.

The cadence has to increase. A tightly matched account could review monthly without much drift; a broadly matched one accumulates new terms continuously and needs weekly attention in proportion to its spend.

And the expectation has to change. Under broad matching, unfamiliar terms in the report are not a fault — they are the mechanism working. The question is not whether the account is reaching queries nobody wrote down, but whether those queries convert. Some will, and they are the discovery the setting was chosen for.

The report is also where an account finds out how much of its performance is its own name.

Brand searches convert at rates no other traffic matches, and when they are mixed into a campaign with generic terms they lift the campaign's averages substantially. An account that has never separated them may believe its prospecting is performing far better than it is.

Splitting brand into its own campaign is the fix, and the search-term report is how the extent of the mixing becomes visible in the first place.

A caution about acting on single conversions, because it is the most common way a negative list gets built badly.

A search term with one conversion from forty clicks and a term with one conversion from two clicks look very different in a sorted report and are statistically indistinguishable. Promoting the second to its own keyword, or protecting it from a negative, on the strength of one event is a decision made on noise.

The practical threshold is volume rather than a rule: act on terms with enough clicks that the absence of conversions means something, and treat everything below that as unresolved. A term with four clicks and no conversions has told you nothing yet.

The same applies in reverse to the spend pass. A term appearing once for a small amount is not worth a negative; a term appearing steadily for small amounts is.

The report earns a fixed slot rather than an investigation.

Read it on a schedule proportional to spend. Do the spend pass and the no-conversion pass. Add negatives for what is clearly wrong, promote what is clearly working, and leave the ambiguous middle alone until it accumulates enough volume to resolve.

Then check the campaign's aggregate cost per conversion against the previous period, because the withheld portion of the report is invisible and the aggregate is the only place its drift shows up.