The search terms report tells you what searches actually triggered your ads. Not the keywords you bid on. The actual user queries Google matched against your keywords and decided were close enough.
That gap between "what you bid" and "what triggered" is where budget goes when you are not paying attention.
Most accounts check the report reactively. Spend climbs, CPA goes soft, someone opens the report and finds $900 of clicks on queries that have nothing to do with the business. Then they add negatives, feel good for a week, and go back to not checking it.
The fix is a short, consistent weekly workflow. Thirty minutes, the same day every week, the same filter sequence each time. Here is how to run it.
Why Reactive Is Not Good Enough
Match types in Google Ads have loosened every year. Broad match now pulls in queries that share "similar intent" with your keyword, which Google interprets generously. Phrase match has expanded its definition of phrase since 2021. Even exact match allows close variants.
In practice, this means every broad and phrase match keyword in your account is a small search engine in itself, generating queries you did not explicitly approve. Some of those queries are good. Many are not. A few are actively bad (competitor brand terms, job-seeker queries, research queries from students). None of them are under your control unless you run the report and make decisions about them.
Waiting until you notice a problem means the budget has already burned. Waiting until you notice AND investigate means you are probably weeks behind on negatives that should have been filed two billing cycles ago.
The Weekly Setup: Filters and Sort Order
Open the search terms report in Google Ads (Campaigns > Insights and reports > Search terms).
Set the date range to the last 7 days. Do not use a longer window for weekly review: anything older was either already reviewed or is too stale to act on with confidence.
Apply these filters in sequence:
- Impressions > 5. Filters out one-off queries that are not worth a decision yet. You will catch them if they recur.
- Match type: Broad match. Review broad match first. It generates the most unexpected queries and the most waste.
- Sort by: Cost, descending. Highest spend at the top. You want your decisions to matter, so start where the money is.
Run the same sequence a second time with Match type: Phrase match after you finish broad. Exact match rarely needs weekly review unless your account runs in a volatile category.
The Three-Bucket Triage
Every query in the report belongs in one of three buckets. Work through the list and make a call on each row.
Bucket 1: Add as negative. The query has no business being in your account. It is pulling spend on irrelevant intent, the wrong audience, or terms you actively do not want to show for.
Common patterns: job-seeker modifiers ("how to become," "jobs," "salary," "career"), research intent ("what is," "vs.," "review"), competitor brand names you do not bid on (add to a negative list, not campaign-level), and any query where the searcher is clearly not your ICP.
File these at the ad group level if the pollution is specific to one ad group. File at the campaign level if the pattern is broader. File at the account level (shared negative list) for terms you never want, ever, in any campaign.
Bucket 2: Watch and decide next week. The query is borderline. It could be your audience or it could be adjacent noise. It does not have enough data yet to justify a decision.
Do not add these as negatives prematurely. A query with 6 clicks and zero conversions is not yet evidence, especially on keywords where your conversion window is 30 days. Mark it mentally, let it accumulate one more week of data, then make the call.
Bucket 3: Promote to keyword. The query performed. It has clicks, and it has conversions or engagement, and it is not already in your account as an explicit keyword.
This is the category most accounts ignore. They use the report only to find negatives. But queries that performed well under broad or phrase match are pre-validated demand. Adding them as exact match keywords gives you direct bidding control, dedicated ad copy, and tighter Quality Score calculation. A query that converted three times under broad match at $24 CPA becomes an explicit exact match keyword you can bid on deliberately.
A Before/After Example
An account running broad match on the keyword project management software in a campaign targeting construction firms.
The search terms report for week ending October 3 shows:
| Query | Clicks | Cost | Conv. |
|---|---|---|---|
| construction project management software | 12 | $186 | 2 |
| project management software free | 8 | $124 | 0 |
| construction scheduling software | 6 | $93 | 1 |
| best project management tools 2026 | 5 | $77 | 0 |
| project management jobs near me | 3 | $46 | 0 |
Applying the three-bucket triage:
- "construction project management software" (Bucket 3): exact match candidate, 2 conversions, already aligned with the ICP. Add as exact match with its own ad group and construction-specific copy.
- "project management software free" (Bucket 1): free-seeker intent, zero conversions, $124 spent. Add "free" as a negative keyword at the campaign level.
- "construction scheduling software" (Bucket 3): close variant with a conversion, worth watching for one more week before promoting. Move to Bucket 2 if you want to be conservative; Bucket 3 if one conversion is enough signal in your conversion window.
- "best project management tools 2026" (Bucket 2): research intent, no conversions, but could be consideration-stage traffic worth capturing with the right landing page. Watch for one more week.
- "project management jobs near me" (Bucket 1): job-seeker query, immediate negative. Add "jobs" to the campaign-level negative list.
Five rows. Two negatives, one promotion candidate, one to watch, one borderline. That is what the workflow looks like in practice. Fifty rows takes about 25 minutes, most of which is skimming the tail.
Match Type Patterns to Watch
Certain patterns show up repeatedly across accounts and are worth knowing before you sit down with the report.
Broad match + generic modifier = high noise. Keywords like software, tool, platform, and solution attract the widest queries under broad match. If you are running broad match on a keyword ending in any of these, expect the search terms report to show queries from adjacent categories, informational searchers, and competitor comparisons. These keywords usually need either a tighter match type or a long list of negatives.
Phrase match erosion since 2021. Google's phrase match expansion means "keyword" can trigger for queries that Google considers semantically equivalent but a human would not. Watch especially for phrase match keywords where the expanded queries drop core modifiers (industry-specific terms, geographic terms, product-specific terms). The phrase match expansion tends to strip the qualifiers that make your keyword precise.
Exact match with close variants. Exact match now allows plurals, misspellings, and reordered words. Usually harmless. Occasionally, reordering changes intent entirely, particularly with two-word keywords where order matters.
How Often to Run This
Weekly for campaigns with meaningful daily spend (call it $50/day per campaign as a rough floor). Below that, every two weeks is fine since there will not be enough data to act on.
Do not skip more than two consecutive weeks, even in low-activity periods. Queries compound. A query that generated 5 clicks week one and was not caught generates 20 by week three. By the time you see it, you have burned budget you will not get back.
The accounts that run clean long-term are the ones where the report review is a standing calendar appointment, not a crisis response.
Frequently Asked Questions
The Google Ads search terms report shows which actual user queries triggered your ads. To mine it effectively, run a weekly review using this sequence: open the report in Campaigns > Insights and reports > Search terms, set the date range to the last 7 days, filter for impressions above 5, and sort by cost descending so the highest-spend queries appear first. Review broad match terms first, then phrase match.
Categorize every query into one of three buckets. First, queries to add as negatives: irrelevant intent, job-seeker modifiers, research phrases, and competitor brand terms. Second, queries to watch for another week: borderline terms without enough conversion data yet. Third, queries to promote to explicit keywords: terms that converted under broad or phrase match but are not yet in your account as exact match keywords. Promoted queries get dedicated ad groups and tighter bidding control.
Run this weekly for any campaign spending $50 or more per day. Letting more than two weeks pass between reviews allows waste to compound. The accounts with consistently low CPAs are the ones where the search terms report is a scheduled maintenance task, not a reactive audit.