There is a report inside every Google Ads account that shows exactly what people typed before they clicked. Almost nobody opens it. The Google Ads search terms report is unglamorous, takes about half an hour a month, and is still the fastest way to find money leaking out of a search account.
I read it by hand on every account I run. Not because I enjoy it, but because nothing else tells you what your ads are actually being shown for.
This is for you if your search campaigns are running and you have not looked at what they matched.
What the Google Ads search terms report shows
Two different lists exist in your account and people confuse them constantly.
Your keywords are what you told Google you wanted. Your search terms are what people actually typed. Google matches one to the other, and the looser your match types, the more creative that matching gets.
So you might bid on “emergency plumber” and appear for “how to fix a leaking tap yourself”. Both contain plumbing. Only one of them is a customer.
The report is where that gap becomes visible. Everything else in the interface shows you your side of the arrangement. This shows you Google’s.
Why the Google Ads search terms report still matters
The common objection is that automation has made this obsolete. The opposite is true.
Automated bidding decides how much to pay for a search. It does not decide whether that search deserves your money at all. So a smart bidding strategy pointed at a loose keyword will efficiently, tirelessly buy you clicks from people who were never going to become customers.
Broad match makes it sharper still. It reaches further than it used to, which is genuinely useful when it works, and expensive when nobody is watching the edges.
The search terms report gets read by hand, every account, every time. Nobody does this any more and it is still where the money is.
How to read the Google Ads search terms report
Half an hour, once a month, in this order.
Set the date range to the last 30 days. The default is usually shorter and hides the occasional expensive oddity.
Sort by cost, highest first. You are not trying to review everything. You are looking for the terms that took real money, because a term that cost eleven pounds and converted nothing matters more than forty terms that cost twenty pence each.
Read down the list and put each one in a bucket. Irrelevant, which becomes a negative. Relevant and converting, which may deserve its own ad group and its own ad. Or relevant but oddly phrased, which is a note for your copy.
That third bucket is the one people skip and it is often the most valuable. The words customers use are rarely the words the business uses.
What to do with what you find
Negatives first, because they stop the bleeding.
Add the terms that cost money and cannot convert. Do it at the right level: a negative that applies to the whole account belongs at account level, not buried in one ad group where it will be forgotten.
Look for themes rather than strings. If four variations of “free”, “jobs” or “DIY” turn up, block the theme rather than the four exact phrases you happened to see this month. Next month brings four more.
Then take the converting terms and check they have a home. A phrase producing enquiries while sitting inside a vague ad group is being served by an ad that does not quite match it, and matching the ad to the search is one of the cheapest improvements available.
What this looks like in an account
The best search result on my books is 6,070 leads at £11.40 each for a UK staffing and childcare business, on Google Search alone.
Performance Max was tested on that account and dropped. It kept delivering people who wanted visa sponsorship rather than candidates who could take the work, and the search terms report is where that pattern showed up first. In a campaign type that shows you much less of what people actually typed, the same problem would have taken months longer to see.
There is nothing clever behind that figure. Negative lists reviewed monthly, one landing page built for one job, and conversion data flowing back so bidding had something real to work with.
The monthly review was the part that kept it there. Costs did not creep, because the drift was caught while it was small. An account left alone for a year does not fail dramatically. It gets slowly, quietly more expensive, and by the time anyone notices the cause is a hundred small things rather than one.
For wider context, WordStream’s 2026 study of 13,474 US search campaigns put the all-industry average at $66.69, which makes £11.40 look unusual. The market average includes a great many accounts nobody is reading this report for.
The judgement a script cannot make
Rules and scripts can flag terms above a spend threshold with no conversions. That is worth setting up and it saves time.
What automation cannot do is tell you what a phrase means. “Cheap solicitor” and “affordable solicitor” look almost identical to a machine and describe two different people. One is price-shopping, one is budget-conscious and serious. Only a person who knows the business can call that.
That is the actual argument for doing this by hand. Not nostalgia, but that relevance is a judgement about meaning, and meaning is the one thing the tools still cannot read.
There is a second reason worth mentioning. Reading the terms tells you things about your market that no dashboard summarises. You find out what people are worried about, what they think your service is called, and which competitor they were comparing you against. That is research you would otherwise pay for, sitting in a report you already have.
What to change this week
Four steps, about half an hour.
Open the report, set the range to 30 days, and sort by cost descending.
Work down the top twenty terms and bucket each one: negative, promote, or copy note.
Add your negatives at the level they belong, and block themes rather than single strings.
Then put a recurring half hour in your calendar for the same day next month, because the value is in the repetition rather than in any single session. One good session finds the obvious waste. Twelve of them stop it ever building up again.
The wider piece on judging lead cost covers what all this feeds into, why search costs fell this year covers the market backdrop, and the order I use to reduce lead cost shows where this sits among the other levers. If you would rather someone else did the monthly pass, Google Ads management includes exactly this, and you can book a teardown to see what your report is hiding.