Most of my Google Ads spend is in "Other search terms." What can I still control?

Last verified October 2026

When most of a Search campaign's spend sits in the search terms report's "Other search terms" row, no route will hand you the text of those searches: not the API, not a third-party tool, not search terms insights. Google lists a search only after enough people have typed it across all of Google, and Google's Ads Liaison has said that threshold applies to every query the same way.

The hidden spend still has an owner and a limit wherever one of your keywords bought it. Google's reports include hidden clicks in their totals, so a keyword's totals should include the hidden searches it bought, and the steps below include a check that they do. Google also documents what each match type may match. Put those together and "mostly hidden" becomes a dollar figure for each keyword, with a known limit on what its unseen searches could have been. None of this depends on conversion counts, so it works in campaigns too small to judge by cost per conversion.

The method covers a whole campaign when every click comes through exact or phrase keywords. On Manual CPC, AI Max's search term matching stays out of that picture: Google says it "will not work with manual CPC bidding." On other bid strategies, AI Max can match searches outside your keywords, and Google turns it on by default in new Search campaigns. Dynamic Search Ads targets use no keywords at all. Spend from either source has no keyword owner, and the steps below show how to find it and set it aside. Broad keywords do have an owner, but their limit is much looser; that case is covered at the end.

How unusual 90% is, and why budget won't buy visibility

Google publishes no benchmark. PPC Chief has collected published analyses, and they run from about 27% of spend hidden to a vendor-reported 51% average, with individual accounts higher. The highest single-campaign figure in that collection is 80% of spend, in an all-exact-match campaign that Search Engine Journal reported in 2024. Ninety percent is high against all of it, and that 80% case shows tight match types don't prevent high shares.

Niche accounts run high because of where the line falls. Google's 2021 announcement says its thresholds report only on terms "that have seen sufficient search volume across all Google searches." What counts is how many people typed a string anywhere on Google, not how often it reached your ads. A specialist service whose customers describe their problem in their own words can generate many rarely typed strings, and those stay below the line at any spend. More budget buys more of the same searches, not a view of them. The share can still move when spend shifts between keywords, and the section on recent jumps covers that.

Measure the share in cost. Cost is the money you're accounting for, and it's the column the method works in.

What can show you more, and what can't

The API and third-party tools. When Google expanded the report in 2021, its API team said the change applied to the API's search terms resource too. Tools built on the API see what the API sees. If a tool claims to list your hidden searches, ask it where it got them.

Search terms insights. Cost isn't among its metrics. So the clicks and conversions it reports, with visible and hidden searches mixed together, come without a price. Google says its categories take all search terms into account, and a category's details list its subcategories and search terms, but the omitted queries are grouped "without exposing the queries themselves." Google also places terms that "have lower spend" in insights' own "other search terms" group, so in a small account much of the traffic can end up unlabeled there too.

A longer date range. Google doesn't say whether a longer window lets more searches qualify. You can test it yourself: compare the hidden share over 30 days with a window several times longer, counting only the days the campaign ran.

None of these gives you the strings. What you can still total, keyword by keyword, is the money.

Find which keywords bought the hidden searches

The search terms report's rows are only the visible searches, but its totals include the hidden ones. When Google expanded the report in 2021, its API team said "metric totals from search terms reports will now be consistent with other reports, such as campaign, ad group, and ad reports." The keyword report isn't named in that statement, so the steps below end with a check that its rows account for the whole campaign. Where they do, each keyword's total minus its visible rows is the hidden part that keyword bought.

1. In the search terms report, set the campaign and date range, segment by Network, turn on the Keyword column (it's off by default) and download the report. Google's advice for downloads applies to both reports: make sure the table "contains the date range, columns, filters, and segments you'd like to include in your report."
2. For each keyword, total the cost, clicks and conversions of its visible rows. The per-keyword figures come from keyword-matched rows only. On Manual CPC, Google says AI Max's search term matching "will not work with manual CPC bidding," so that matching isn't a source of rows outside your keywords there. Leave out the report's total rows and any Display Network rows. If the same keyword text sits in more than one ad group or match type, keep those apart.
3. Download the keyword report with the same date range, filters and Network segment. Subtract each keyword's visible totals from its own totals. A keyword with no visible rows has nothing to subtract, so all of its traffic is hidden.
4. Add up your keyword rows and compare the sum with the campaign's total on the same networks and dates. If they match, every dollar has a keyword owner. If they don't, the gap can come from filters that differ between the two reports, from keyword totals that leave out some hidden searches, or from traffic none of your keywords owns. With AI Max, unowned traffic shows in the keyword report's "Total: AI Max expanded matches" and "Total: AI Max landing page matches" rows; Google describes the second as traffic "outside of your keywords." With Dynamic Search Ads, it's terms that Google says "won't return a keyword in the keywords field." Keep that spend out of the per-keyword figures and treat it as having no keyword limit. Whatever its cause, a gap that remains means the per-keyword figures don't account for all of the campaign's spend.

The Network segment matters if Display Expansion is on. It can put a Search campaign's ads "on relevant website pages and apps on the Google Display Network." Those clicks come from pages and apps, not searches, and Google's pages don't say whether they land in keyword rows, so compare search traffic with search traffic.

Here is a constructed example: a business that repairs and services commercial espresso machines in one city. It runs one Search campaign on Manual CPC with four keywords, no Dynamic Search Ads and no Display Expansion, and its keyword rows add up to the campaign. Over 30 days it spent $1,200 on 150 clicks and recorded 5 conversions. Its visible search-term rows account for $120, so 90% of the cost is hidden.

If columns extend beyond the page, scroll sideways. With a keyboard, focus the table and use the arrow keys.

Constructed example, 30 days, search networks only. Clicks, cost and conversions are the keyword report's, across all of each keyword's searches. Hidden cost is the keyword's cost minus the cost of its visible search-term rows over the same dates. The last column paraphrases Google's match type definitions.

The $48 of visible rows under "espresso machine service" is three searches: espresso machine service near me ($30), espresso machine service cost ($10) and espresso machine service manual ($8).

The 90% means something different in each row. Read across, and two rows that both say "hidden" call for opposite responses.

What a keyword's unseen searches can be

Exact keywords limit the meaning. Google says an exact keyword can match searches with the same meaning or intent. Its close variants include misspellings, plurals, reordered words, implied words, synonyms and paraphrases, and no match type can opt out of them. So the unseen $170 behind [commercial espresso machine repair] can be repair commercial espresso machines, commercial espresso machine repairs or a synonym for espresso machine. Same-intent matching can stretch: Google's own example lets [images royalty free] match free copyright images. But every one of those searches has to share the keyword's meaning or intent. Here, "hidden" mostly means "phrased in a way few people phrase it."

Phrase keywords limit only the core. Google says phrase match shows ads on searches that include the keyword's meaning. It adds that "the meaning of the keyword can be implied, and user searches can be a more specific form of the meaning." espresso machine technician jobs, espresso machine technician training and espresso machine technician near me all qualify for "espresso machine technician." The match rule doesn't distinguish between them, and the report can't tell you which ones bought the $210.

The visible rows aren't a fair sample. Reading the visible 10% as a preview of the rest is tempting. By Google's definition, though, the visible rows are strings many people type, and the hidden ones are strings few people type. Google says nothing more about them. Still, it's a fair inference that under a phrase keyword, unusual additions sit mostly in the hidden part, since each unusual word makes a string rarer. A clean visible list therefore doesn't clear the hidden part. And one junk visible row tells you the phrase admits that kind of addition, not how much of it there is.

Five conversions can't settle any row. The campaign row shows 5 conversions from 150 clicks, or 3.3% of clicks. At that rate, the 30 clicks behind "espresso machine technician" would bring about one conversion on average, and a month with none happens roughly one time in three. That zero isn't a verdict. Two conversions from 18 clicks on the commercial keyword isn't a triumph either: at 3.3%, those clicks average 0.6 conversions, and two or more happens about one month in eight. At this volume, per-keyword cost per conversion mostly measures luck. What the table does support is a judgment about exposure: how much money each keyword spends on searches whose meaning you can't pin down.

Act on the keyword, not the search

[commercial espresso machine repair] is 100% hidden, and that's fine. Its unseen searches have to share its meaning or intent, and that meaning is what you sell. Leave it alone and keep reading the leads it brings.

[espresso machine repair] is mostly hidden. Its limit is tight, but the meaning itself is wider than the business. Someone with an espresso machine at home types the same words. Visibility wouldn't fix that, because it's a property of the keyword. The evidence is in the leads: what did the people who contacted you from this keyword actually ask for? If home-machine inquiries keep arriving, the keyword is the problem. Lean on [commercial espresso machine repair], and lower this keyword's bid or pause it, rather than hunt for searches you'll never see.

"espresso machine technician" is 100% hidden, and the phrase admits searches the business would refuse. Nothing in the report will show whether job seekers or would-be trainees are using it. Refusing them doesn't need data, though. If you don't hire through ads and don't train technicians, those searches are outside the offer today. To reach hidden searches, the negatives should be broad or phrase negatives: apart from the misspellings Google accounts for, an exact negative doesn't reach searches with extra words or other variants, and the last paragraph of this section explains why. Write them with every form you mean: job and jobs, career and careers, salary and salaries, training, course and courses. Google says negatives don't match close variants, so "you'll need to add synonyms and singular or plural versions if you want to exclude them." Google also says negatives "account for changes in casing and misspellings for which Google changes the search results," and its own example warns that with "Google" excluded, users may search for "gogle," which the negative doesn't exclude. Check each negative against searches you want before adding it. Then ask whether the phrase keyword earns its place at all, or whether its few customer forms would be better bought as exact keywords.

"espresso machine service" is the biggest spender, and 90% of it is hidden. Its visible rows are where to look for searches to refuse, and espresso machine service manual is a do-it-yourself search. The single word manual would reach furthest, but it would also block manual espresso machine service, a search about lever machines this business may well want. The phrase negatives "service manual" and "service manuals" refuse the do-it-yourself search without that risk. Then reshape the keyword. Move the forms you clearly want, such as [espresso machine service] and [espresso machine service near me], into exact keywords, and lower the phrase keyword's bid so it pays less for whatever remains.

On Manual CPC, that bid applies to every search the keyword enters, seen or unseen. So do the campaign's device, location and ad-schedule bid adjustments, which Google says let you show ads "more or less frequently based on where, when, and how people search." Google's chart lists those adjustments as supported on Maximize Clicks as well. Smart Bidding strategies such as Maximize Conversions or Target CPA handle bid adjustments under their own rules, which Google sets out on its bid adjustments page. On those strategies too, the levers that change what a keyword may match still apply: negatives, wanted forms moved into exact keywords, and pausing or replacing the phrase keyword.

Two conditions come with that reshaping. First, exact keywords won't necessarily make the searches visible; the all-exact campaign above still had 80% of its spend hidden. The point is not a better-looking percentage. It's that more of the money runs through keywords whose unseen searches mean what you sell. Second, because the conversion counts are tiny, make one change at a time. Judge it over equal periods long enough for leads to arrive, and give the leads' content more weight than the conversion column.

The reason for those particular negatives is how negatives reach hidden searches. An exact negative copied from a visible row blocks that string, along with casing changes and the "misspellings for which Google changes the search results," and nothing with extra words. Every hidden search is, by definition, some other string; unless it is one of those misspellings, the exact negative can't touch it. Hidden searches with extra words or other variants are beyond an exact negative; what reaches them is a broad or phrase negative whose terms they contain, and Google's matching rules "apply whether the negative keyword is one word or multiple words." A single word reaches furthest and risks the most wanted searches. A phrase like "service manual" reaches less and risks less. That trade-off is the choice behind each negative above.

If the share jumped recently

Start with the denominator. In a campaign spending $1,200 a month at about $8 a click, each click moves the cost share by about two-thirds of a point. A quieter month swings further. Then compare equal stretches of running time before and after the change, keyword by keyword, in cost:

• If a keyword's own hidden share rose, what that keyword matches has changed, or the month was noisy. Look at its visible rows for new kinds of addition.
• If each keyword's share held but the campaign's rose, spend moved toward keywords that were always mostly hidden. That's a change in mix, not in Google's reporting.
• If you added keywords or loosened a match type in the window, compare like with like before drawing anything from the total.

On Manual CPC you can set aside one explanation, because Google says AI Max's search term matching doesn't work with manual CPC bidding. On any other bid strategy, check the campaign's AI Max setting. Google says search term matching and related features "require conversion-based Smart Bidding strategies in order to function effectively," so the setting matters most on those strategies. Google turns AI Max on by default in new Search campaigns. It has also said that, starting in September 2026, campaigns using text customization (formerly automatically created assets) will be upgraded to AI Max. A rise in the AI Max total rows from step 4 shows how much spend moved to AI Max's own matches. One trap: filtering the search terms report's Match type column to "AI Max" "may show lower numbers for AI Max," because, Google says, "this filter doesn't consider 'Other search terms'."

When this answer changes

With real conversion volume, the hidden aggregate becomes worth judging on results. PPC Chief's framework splits brand from non-brand first, then compares the hidden aggregate with your real target. Its verdict on the hidden spend is two-sided: its "performance may be worse, sometimes dramatically, but it is not universally worse and the reporting threshold has not been shown to cause the gap." At 5 conversions, one more conversion lowers cost per conversion by about 17%, and one fewer raises it by 25%. At 38, the number in PPC Chief's own example, either moves it by under 3%.

With broad match, the limit in the table's last column loosens. Google says broad can match "searches that are related to your keyword, which can include searches that don't contain the direct meaning of your keywords." The subtraction still tells you which keyword bought the hidden spend, but much less about what that spend could have been. Treat a broad keyword's hidden spend as exposure with a loose limit, and use the same levers harder: move the forms you want into exact or phrase keywords.

In Performance Max, none of this arithmetic applies. Google's API notes that its ad-group search terms view excludes Performance Max, which has its own search terms reporting, so keep those numbers separate from your Search campaigns'.

In an exact-and-phrase campaign, which searches get listed is Google's decision. What each keyword is allowed to buy, and what you pay for it, is still yours.