How to Read a Google Ads Search Term Report (And What to Actually Do With It)

Set a keyword. Watch it convert. That’s the story most Google Ads dashboards tell you — and it’s an incomplete one. The keywords you bid on and the words people actually typed before your ad showed up are two different lists, and the gap between them is where most wasted spend hides.

The tool that closes that gap is the search term report. It’s not new, and it’s not exciting, but it’s still the single most useful screen in Google Ads for anyone trying to understand where a budget is actually going — and in 2026, with match types behaving more loosely than ever and Smart Bidding expanding auctions on its own, it matters more than it used to, not less.

What the search term report actually shows you

Every keyword you add to a campaign — broad, phrase, or exact — is a target, not a guarantee. Google still decides which real-world searches count as a match, and with broad match in particular, that decision can be surprisingly loose. The search term report is the record of that decision: the literal queries people typed that caused your ad to show, sitting right next to the keyword that supposedly triggered it.

You’ll find it inside a Search, Shopping, or Performance Max campaign under Insights and reports → Search terms. The columns look familiar — clicks, cost, conversions — but the row is the actual query, not your keyword. That distinction is the whole point.

One thing worth knowing going in: Google now withholds a meaningful share of search terms from this report for privacy reasons, especially low-volume queries. You won’t see everything. That’s not a reason to skip the report — it’s a reason to review it more often, since you’re already working with a partial picture.

Why your keywords and your search terms don’t match

Two things are usually responsible. The first is match type behavior — broad match, and to a lesser extent phrase match, are built to find “related” searches, and Google’s definition of related keeps getting broader. A keyword aimed at people ready to buy can end up matching people who are researching, comparing, or job-hunting.

The second is automated bidding. Smart Bidding strategies are explicitly designed to explore new auctions in pursuit of your target — that’s a feature, not a bug, but it means the system will keep testing search terms slightly outside your intended range as long as the algorithm thinks they might convert. Left unchecked, that exploration is exactly where budget quietly leaks.

How to actually read the report without breaking anything

The fast way to find waste is to sort the report by cost, descending, and look at the rows with zero conversions first. That’s genuinely the right instinct — those are the queries spending money and returning nothing.

Where people get it wrong is turning that sort into a blanket rule: exclude every zero-conversion row, full stop. That approach cuts real waste, but it also does three things you don’t want:

• It can shrink a Smart Bidding campaign’s exploration space enough to distort how the algorithm learns, sometimes worsening performance on the keywords you kept.
• It can mask a conversion tracking problem — a query with zero recorded conversions might be a query with a broken tracking pixel, not a bad query.
• It can quietly narrow your reach until impressions collapse and you’re wondering why traffic dried up.

A more reliable read: look for a pattern, not a single bad day. A search term that’s spent money with zero conversions over several weeks, especially one that’s semantically off from what you sell, is a strong negative-keyword candidate. A single expensive click with no conversion is often just noise.

Turning findings into negative keywords, without overcorrecting

Once you’ve identified a genuinely irrelevant query, the next decision is match type — and this is where a small default trips people up. When you add a negative keyword directly from the search term report, Google Ads defaults it to exact match. For most cases you actually want phrase match instead, so the negative also blocks close variations of that same irrelevant query, not just that one exact string.

Negative match types work in the opposite direction from regular keyword match types: a negative exact match blocks only that precise search, a negative phrase match blocks any search containing that phrase in order, and a negative broad match blocks any search containing all the words in any order. Broad negatives are the ones to use carefully — they’re powerful enough to cut relevant traffic if you’re not specific.

The other habit worth building: add negatives before a campaign launches, not just after. A seed list based on obviously irrelevant categories for your business — “free,” “jobs,” “how to,” competitor names you don’t want to compete on — means the first week of a new campaign isn’t spent funding traffic you’d have blocked anyway.

Stop adding negatives one ad group at a time

A negative added at the ad group level protects that ad group and nothing else. If the same irrelevant query is bleeding budget across five campaigns, adding it five separate times isn’t just tedious, it’s a maintenance problem waiting to happen the moment you launch a sixth.

Shared negative keyword lists solve this: build a list once, attach it to every relevant campaign, and any addition to that list takes effect everywhere it’s attached, immediately. Most accounts do well with a small number of purpose-built lists rather than one giant catch-all — something like a “not relevant to us” list for obviously off-topic terms, a competitor-exclusion list if you don’t want to bid on competitor brand names, and a geographic list if you serve some locations and not others. Splitting them this way makes it obvious where a new negative belongs and keeps the lists easy to audit later.

How often you should actually be checking this

There’s no single right cadence, but spend level is a reasonable guide. A new campaign, or one spending several thousand dollars a month or more, accumulates waste quickly enough that a weekly look is worth the time — new query patterns show up fast, and letting them run for a month before catching them is expensive. A stable, established campaign that’s already had its obvious junk filtered out can usually move to every two weeks. A low-spend or genuinely evergreen campaign, where little changes month to month, can be checked monthly without much lost.

The pattern underneath all three: review frequency should track how fast new queries appear, not a calendar habit you set once and never revisit. An account that was fine on a monthly check-in in January can start needing weekly attention the moment you launch a new campaign or Smart Bidding starts exploring more aggressively.

Performance Max used to hide this — that’s changing

For years, one of the biggest complaints about Performance Max was that it gave you almost none of this. You’d see spend and conversions at the campaign level, but the actual search terms behind Search-network activity inside a Pmax campaign were effectively invisible, which made it impossible to add negatives the way you would on a standard Search campaign.

That’s genuinely changed in 2026. Search term visibility for Performance Max has been extended to smaller accounts — spending as little as $50 a day can now see the search terms report for a Pmax campaign, not just large advertisers. Alongside it, Google has rolled out proper negative keyword support at both the campaign and account level for Performance Max, so you can now exclude an irrelevant or brand-damaging query without dismantling the automated campaign structure to do it.

If you’ve been running Performance Max and treating it as a black box for search term purposes, it’s worth checking again — the same review habits from the rest of this article now apply there too, just with a shorter history of data available since the visibility is newer.

A quick example of how to categorize what you find

Say a weekly review turns up three search terms that spent money but didn’t convert. How you’d typically treat each one:

• A near-exact match to your product, misspelled or pluralized differently. This is usually fine to leave alone — it’s the kind of variation broad and phrase match are supposed to catch, and a single non-converting week doesn’t mean it never will.
• A query containing “free,” “jobs,” “salary,” or “how to become a,” where your business sells a paid product or service. This is a strong, low-risk negative — these patterns rarely represent buying intent regardless of industry, which is exactly why they’re common seed-list entries before a campaign even launches.
• A query that’s topically adjacent but not what you offer — for example, a query for a broader category than the specific service you provide. This is the one worth sitting on for a second week before deciding. It might be genuinely irrelevant, or it might be early-stage research from someone who converts later through a different, more specific search. Cutting it too fast is how relevant reach quietly disappears.

The pattern across all three: obviously off-topic and low-intent terms get cut quickly, near-misses on your actual offering get left alone, and anything ambiguous gets another week of data before a decision. That middle category is where most of the judgment calls in search term review actually live.

Common questions

Does adding negative keywords hurt Smart Bidding?
Not when it’s targeted. Smart Bidding adapts to whatever auction space is available to it; removing genuinely irrelevant queries just narrows that space to more relevant ones. What hurts Smart Bidding is aggressive, blanket exclusion that removes so much volume the algorithm no longer has enough signal to optimize against.

Should I use the same negative list across every campaign?
Only for terms that are irrelevant to your entire business, not just one campaign — things like job-seeker terms or clearly unrelated categories. Campaign-specific irrelevance (a term that’s off-topic for one product line but fine for another) belongs in a campaign-level or ad-group-level negative, not a shared list that would block it everywhere.

What if my search terms report looks unusually empty?
That’s worth investigating rather than treating as good news. An empty-looking report can mean your negatives are working — or it can mean you’ve over-restricted the campaign and impressions have collapsed with it. Check impression volume and reach alongside the search term list, not the search term list in isolation.

Where this fits with automation

None of the above requires anything beyond Google Ads’ own reporting screen — it’s a manual process, and for a lot of accounts, a manual weekly or biweekly review is genuinely sufficient. The limitation isn’t the report, it’s time: a thorough search term review, done properly with pattern-checking rather than a blind cost sort, takes real attention, and it’s the first thing that gets skipped when a week gets busy.

That gap — the review that should happen weekly but happens monthly, or not at all — is exactly what continuous account monitoring is built to close. Growera’s Google Ads Manager reads search term activity daily rather than waiting for someone to open the report, and proposes negative keyword additions with the reasoning attached, so the decision — and the final approval — still sits with you. It doesn’t replace understanding your own account; it just means the review happens every day instead of whenever there’s time for it.

The short version

• Your keywords and your search terms are different lists. The report is where you compare them.
• Sort by cost and zero conversions to find candidates fast, but look for a pattern across weeks before cutting — a single bad click usually isn’t a signal.
• Default new negatives to phrase match, not the exact match Google suggests, unless you specifically want the narrower block.
• Use shared negative keyword lists so one addition protects every relevant campaign, not just one ad group.
• Review weekly for new or high-spend campaigns, every two weeks for stable ones, monthly for low-spend evergreen accounts — and more often than that default the moment something changes.

The search term report isn’t going away, and no amount of automation changes what it’s telling you. What changes is how often someone actually looks.