What a search-terms review finds: a worked example
By Abhishek Pal, performance marketer (apships.com). Last updated 2026-11-01.
A Google Ads search terms CSV analysis should show which searches deserve attention, why they were flagged and which should stay open. This worked example uses a fictional lamp store and a made-up CSV from a test fixture. The reported $162 in flagged spend is a synthetic result, not a customer's saving.
Start with what the file covers
The sample covers June 1 to June 30, 2026, a 30-day period. It contains 47 audited terms and $669 in spend. That is the review's boundary. I would read every conclusion against that file, rather than assume it describes the whole business.
The free summary reports $162 in wasted spend across 5 suggested negatives. It also shows projected monthly savings of $162, labelled as an estimate. The sample covers 30 days, and those displayed amounts are the same. Neither is proof of what will happen after exclusions are added.
The decisions are 32 keep, 5 negative, 9 watchlist and 1 noise. The sample also records 1 term already excluded in the account. Do not treat that status as another new negative to add. The useful next step is to read the reasons behind the suggested changes.
Read the most expensive suggestions first
The free browser check shows the summary and the three most expensive suggested negatives. Here is how I would review those entries in this sample.
"solar light bulk supplier": $90.00
This term appears in Weak-Search. The rule is R1 cluster: b2b_wholesale. The engine's reason is "matched 'bulk'".
The flag points to wholesale intent. Before accepting it, ask whether the store sells bulk orders. A retail-only shop may reject this traffic; a shop serving business buyers may want it. The matching word explains the flag. It does not settle the business decision.
"cheap lamp wholesale market": $48.00
This term appears in Generic-Search under R1 cluster: b2b_wholesale. The engine's reason is "matched 'wholesale'".
Again, the question is whether the offer fits. The reason does not say that anyone looking for a cheap lamp is a poor customer. I would review the full query and the store's sales model before approving the suggested exclusion.
"lamp png transparent": $12.00
This term appears in Generic-Search under R5x CTR kill. The engine's reason is "CTR 0.20% is under 30% of branch median 5.00% at 2000 impressions".
CTR means the share of impressions that became clicks. The flag compares this term with the sample's branch benchmark. The wording also suggests someone seeking an image file, which is my reading of the query. The quoted reason is the engine's numerical check. These are different pieces of evidence, and neither should be silently substituted for the other.
Check the remaining negatives and the warning
Generic-Search also contains "how to make a lamp at home" at $6.00. Its rule is R1 cluster: diy_content, with the reason "matched 'how to'". "lamp repair shop jobs" costs $5.50 and is flagged under R1 cluster: services_rental, with the reason "matched 'repair'". Although the latter contains "jobs", that is not the trigger recorded here.
The account-health warning reads: "'weak' converts at 0.00% - statistical kills are HELD as watchlist here; this is an account-level problem, not a term problem".
That matters when reading Weak-Search. A weak overall result should prompt a wider diagnosis before individual terms are rejected on performance alone. The "solar light bulk supplier" suggestion has an intent-based reason, "matched 'bulk'". It was not presented as a statistical rejection. I would keep that distinction visible during review.
Read what was held back
The watchlist is useful because a report should explain uncertainty as well as rejection. These examples remain open for different reasons:
- "brightlamp coupon code": "brand term - human review, never auto-negated". Check the brand context before deciding whether the search belongs in this campaign.
- "glowmax lamp store": "'glowmax' (a competitor name) below economic threshold - monitor". The sample does not supply the threshold, so this example cannot justify a new cutoff.
- "designer floor lamp sale": "tier1 match but 0 conv (~1.9 expected, cost $38 vs CPA $14)". CPA means cost per conversion. Despite the spend and absent conversions, the supplied decision is watchlist.
I would leave a note explaining what evidence would change each decision. A watchlist without a next check becomes a place to forget terms. It should preserve the reason for waiting.
Inspect the paste-ready file before using it
These are the four supplied preview lines. They are not the entire set of suggested negatives.
| Campaign | Ad group | Keyword | Match type |
|---|---|---|---|
| Weak-Search | All | solar light bulk supplier | Phrase match |
| Generic-Search | All | cheap lamp wholesale market | Phrase match |
| Generic-Search | All | lamp png transparent | Phrase match |
| Generic-Search | All | how to make a lamp at home | Phrase match |
"All" is the supplied ad group value. Do not assume that label selects every ad group in your account. Check the destination and the import preview. If entering negatives in a text box, preserve the match type through the required symbols. See Google's instructions for adding negatives.
Review, paste, then return next month
Reject suggestions that conflict with the offer. Choose exact or phrase using the match-types guide. Paste only accepted entries into the intended negative-keyword destination, check the saved changes and record the date. Next month, use the export guide to repeat the review and revisit the watchlist.
Run the summary and top suggestions locally in your browser with the free browser check; nothing is uploaded. For the full review and paste-ready file, use the search-term review. Every suggestion still needs your decision.
Questions
- Is this a real customer account?
- No. The lamp store, CSV and results are synthetic. They come from a test fixture and show how to read a report. They are not evidence of a customer's savings.
- Does the $162 estimate mean I will save $162?
- No. It is the sample's projected monthly savings estimate, based on flagged spend. Future searches, spend and your accepted exclusions can differ. It is not a measured result after changes.
- Why does the report keep a watchlist?
- Some terms need more evidence or a business decision. The sample holds a brand query, a competitor query below its economic threshold and a relevant product query with no recorded conversions.
- Does the report change Google Ads for me?
- No. You review the suggestions and add the accepted negatives yourself. The tool does not access your Google Ads account. Check the match type and destination before saving.
Sources
Have an export ready? The search-term review finds the wasted spend in it for free. The full report is $39. Run it on your file.
Guides
- Negative keywords not working? The checks, in order Check negative match types, campaign scope, list attachments, variants and reporting dates before deciding a Google Ads exclusion has failed.
- N-gram analysis for Google Ads search terms, in plain words Find repeated words and phrases across your Google Ads searches. Learn to read n-gram costs, check conversions and review a phrase before excluding it.
- Search terms insights versus the search terms report: which one to use for negatives Use search terms for negative keyword decisions and insights for themes. Learn what hidden-query coverage means and why a category label is not a query.
- Search terms missing from your Google Ads report: what is hidden and why Understand missing search terms, privacy limits and reporting gaps. Learn what a CSV review can recover, plus Performance Max and AI Max reporting checks.
- How to export the Google Ads search terms report (2026) Export Google Ads search terms as CSV. Set the last 30 days, keep review columns, separate Performance Max, and check spend before excluding terms.
- How many negative keywords per month: a review routine, not a quota Build a monthly search-terms review without a keyword quota. Protect converting, brand and live keyword traffic, and keep uncertain terms on a watchlist.
- How this review decides which search terms become negatives See the five rule layers behind the review: row checks, protected terms, intent patterns, statistical evidence and benchmarks from your own export.
- Negative keyword match types explained, with examples Learn negative broad, phrase, and exact match with examples. Review close variants, shared lists, Performance Max, and mistakes that can block searches.
- Negative keyword tools compared: free generators, suites, and a reviewed report Compare free negative keyword generators, account tools and a $39 report by price, input, output and review workflow. Choose what your account needs.
- Starter negative keywords by industry, with the exceptions Review starter negatives for plumbers, dentists and ecommerce. Each intent group includes exceptions so you can avoid blocking customers you serve.
- Performance Max negative keywords in 2026: what you can do and what you cannot What Performance Max negatives, lists and brand exclusions control, how to read its search terms report, and why this tool sets PMax rows aside.