There is one report in Google Ads that tells you what you actually bought, and in most accounts we audit nobody has opened it.
Everything else shows you what you asked for. This one shows you what turned up.
It is not hidden. It is two clicks from the campaign view, it costs nothing, and it needs no setup. It is simply the report that requires you to look at something uncomfortable, which is a fairly reliable predictor of whether a report gets opened.
You Bid on Keywords. You Pay for Search Terms.
The distinction sounds pedantic and it is the whole thing.
A keyword is the instruction you set. A search term is the real query somebody typed that Google decided your keyword was close enough to cover. Those two lists are never identical, and the gap between them is where budget disappears.
A rehabilitation equipment manufacturer we audited sells clinical devices to physical therapy practices. Their search terms report was full of consumer queries: carpal tunnel exercises, stroke recovery, neck strengthening. Real people with real problems, none of whom will ever buy a piece of clinical equipment.
Nobody added those keywords. Loose match types matched them, and the report was the only place it was visible.
Death by a Thousand Cuts
The reason this waste survives is that it never looks dramatic.
That manufacturer's junk clicks were costing around $1.74 each. Nothing about that number sets off an alarm. But one of those terms had been clicked 29 times, which is roughly $50 on one irrelevant query, and there were pages of them.
No single line is worth investigating. Together they were a meaningful share of the budget.
This is why sorting matters more than scrolling. Which brings us to how to actually read the thing.
Sort by Cost, Then Read Down
Open the report, set a sensible date range, and sort by cost, descending.
Not by impressions, which shows you things that never cost anything. Not alphabetically, which shows you nothing. Cost, descending, so the first rows are literally where your money went.
Then read down until the numbers stop mattering. In most accounts that is the first fifty rows, and it takes about fifteen minutes.
Two columns matter alongside cost. Clicks, because a term with many clicks and no conversions is a decision waiting to be made. And conversions, because a term you would have sworn was irrelevant occasionally converts, and the report is how you find out before you block it.
That last point is worth taking seriously. The instinct when reading this report is to block anything that looks wrong, and the report will happily let you block something that has quietly been paying for itself. Check the conversion column before you act, not after.
One practical note on the date range. Too short and there is not enough to see; too long and you are looking at terms you already dealt with. Thirty days is a reasonable default, shortened to a week during a new account's first month when things move quickly.
Four Decisions, Not Two
Most people treat this as a binary: block it or leave it. There are really four, and separating them is what turns a chore into a system.
We work through the export with a number next to each term.
1. Hard negative. Irrelevant, will never convert, block it. Consumer searches on a B2B account, job seekers, homework, the wrong product entirely.
2. Competitor. Somebody searched a competitor's name and your ad appeared. This is not junk, it is a decision: do you want to show there? If yes it belongs in its own campaign with its own budget, not leaking into a general one. If no, negative it.
3. Keep. On target. Leave it, and if it converts repeatedly, consider promoting it to a keyword of its own so you can bid it deliberately.
4. Exact match negative. The subtle one, and it gets its own section.
That third category is the one people forget exists. The report is not only a waste-finding tool. It is where your next good keywords come from, written in your customers' own words rather than yours.
Phrase Negatives and Exact Negatives Are Not the Same
Getting this wrong either blocks too much or too little, and it is the most common mechanical mistake.
A phrase negative blocks any search containing that phrase. Add "coffee pot" as a phrase negative and every search containing those words together is blocked. That is what you want for a term that is wrong in every context.
An exact negative blocks only that precise search. This is what you need for a single broad word that is sometimes right and sometimes useless. Take "coffee" on its own. Somebody searching just that word could want anything, and you do not want to show. But "commercial coffee machine" is exactly your customer, and a phrase negative on "coffee" would kill it.
So: phrase negative when the words are wrong in every combination. Exact negative when the bare term is useless but its longer forms are valuable.
Getting this backwards is how accounts end up accidentally blocking their best traffic while trying to tidy up.
Daily at First, Then Weekly
The cadence matters as much as the method.
For a new account we are in the search terms report every day for the first month. Not because we enjoy it, but because a new account has no negative list and a loose keyword can spend a lot on the wrong thing before anyone notices. The first thirty days is when the waste is highest and the learning is fastest.
After that it drops to weekly, then to a regular monthly review once the list has matured. The volume of new junk falls sharply as the negative list does its job.
The failure mode is treating it as a project rather than a habit. Somebody cleans up the account once, adds two hundred negatives, and never returns. Six months later the report is full again, because the queries people type keep changing and a negative list only blocks what has already happened.
That last clause is the part worth sitting with. A negative keyword is retrospective by nature. It can only block a search somebody has already made and you have already paid for. You are always one step behind, which is fine as long as you are taking the step regularly, and expensive if you are not.
It is also why the account gets cheaper to maintain over time rather than more expensive. The first month is genuinely laborious. By month six the same review takes a fraction of the time, because most of what would have arrived is already blocked.
Gardening, Not Firefighting
The right way to think about this is closer to gardening than repair.
You are weeding what you do not want and feeding what you do. Early on it is mostly weeding, because the bed is full of things nobody planted. Over time the balance shifts, and you spend more of your attention on the terms that are working.
On a coffee equipment account we manage, the early exports were dominated by hard negatives and competitor terms. Months later, running the same process, roughly 80 to 90 percent of incoming searches were on target. The report had not become less useful. The account had become cleaner, and the job changed from blocking to promoting.
That number is a reasonable thing to aim at. If most of what arrives is relevant, your match types and your negative list are doing their jobs. If it is nowhere close, the problem is upstream in what you agreed to buy, and no amount of negatives will fix a net that is too wide.
The Twenty Minute Version
- Open it. Campaigns, then Insights and reports, then Search terms. Set the date range to the last 30 days.
- Sort by cost, descending. Read the top fifty rows.
- Mark each one 1, 2, 3 or 4. Hard negative, competitor decision, keep, or exact negative.
- Add the negatives at the right level. Account-level shared lists for things that are wrong everywhere; campaign or ad group level for things wrong only in that context.
- Promote the winners. Any term converting repeatedly should become a keyword you bid on deliberately.
- Check your existing negatives first. Before adding, look at what is already there. We have seen accounts block their own best terms years earlier and never find out.
- Put it in the calendar. Daily for a new account's first month, weekly after that.
This is the least glamorous recurring job in Google Ads and the one with the most reliable return, because you are not guessing at improvements. You are reading exactly what happened and responding to it.
We do this on every account we run, every week. If you would rather hand it over, we can. If you would rather do it yourself, open the report and sort by cost. Whatever is sitting in the top five rows will tell you within a minute whether this is worth your Friday afternoons.
