Most Google Ads accounts have a leak nobody has looked at. It is not in the bidding and it is not in the creative. It sits in the gap between the keywords you chose and the searches you actually paid for. At The Growth Bully, a Malta performance marketing agency, the search terms report is the first thing we open on an inherited account, and it has usually never been read.
It is a five minute check, and most accounts fail it.
What is the search terms report in Google Ads?
It is the list of real searches people typed before your ad appeared, as opposed to the keywords you added. The two lists are never identical. The keyword is your instruction to the auction, and the search term is what the auction decided your instruction meant, which is the only one of the two that cost you money.
You will find it under the search terms view on any search campaign. Set the date range to the last ninety days, sort by cost descending, and read down. The first screen is usually enough to find money. What you are looking for is not a bad word, it is a bad intent: searches from people who were never going to buy what you sell.
How do negative keywords actually work?
A negative keyword blocks your ad from showing on searches containing that term. It does not lower a bid or deprioritise anything, it removes you from the auction entirely. That makes it the one lever in the account that reduces spend directly instead of moving it somewhere else, which is why it is the cheapest optimisation available.
Negatives carry match types of their own, and this is where people get caught. A negative broad blocks any search containing all of your words in any order. A negative phrase blocks the words in that order. A negative exact blocks only that search exactly. Negative broad is the strong tool and the dangerous one: one careless addition can silently kill a profitable term containing the same word.
Why does my exact match keyword show for the wrong searches?
Because exact match is no longer exact. Google expands every match type to include close variants: plurals, misspellings, reordered words, abbreviations and searches it judges to have the same meaning. There is no setting to switch that off. Your keyword is now a signal of intent rather than a literal instruction.
Same meaning is doing a lot of work there, and the judgement is made by a system that does not know your business. It cannot tell that a term describing your service also describes a job title, a free tool or a DIY project. That is the mechanism behind almost every wasted euro in a search campaign.
What should you look for first?
Spend, not words. Sort the report by cost and read down, because one expensive irrelevant search costs more than fifty cheap ones. Work through the list below in order: it puts the biggest recoverable spend first and leaves the judgement calls until last, when you have already banked the easy savings.
- Job and career intent. Anything with jobs, salary, vacancy, apprenticeship or course in it. Pure cost, zero chance of revenue, and almost always present.
- Free, DIY and how-to intent. People looking to do it themselves are not shopping for a supplier. Some of these are worth content, none are worth a click on a commercial campaign.
- Wrong service, wrong product. The adjacent thing you do not sell. Common where one word in your service name has a second meaning.
- Wrong place. Searches naming a country or town you do not serve. Location targeting does not catch these, because the search text and the searcher's location are different things.
- Research intent on a high value term. Definitions, comparisons, meaning-of searches. Expensive, early, and better served by an article than an ad.
- Everything with spend and no conversion over ninety days. Judgement needed here, because a term with low volume may simply not have had its chance yet.
Add each one at the right level, which is where most accounts go wrong.
Should negative keyword lists sit at campaign or account level?
Use both, deliberately. One shared account level list holds the permanent exclusions that will never be relevant to anything you sell: careers, free, DIY, the countries you do not serve. Apply it to every search campaign and never think about it again. Campaign level negatives hold the terms that are wrong for that campaign but right for another.
The second half matters more than it sounds. If you run separate campaigns for separate services, each should exclude the other services by name. Without that they compete for the same searches and you can no longer read either one, because the spend and the conversions have mixed.
Should you bid on your own brand name?
Usually yes, and the argument against is better than people admit. The case for is cheap clicks, the highest intent there is, control of the message, and the risk that somebody else bids on your name. The case against is paying for a click the organic result below would have given you free.
The honest answer is that it depends on whether anyone is bidding against you, which your own search terms report and a plain search will tell you in a minute. What is not defensible is leaving brand traffic mixed into a generic campaign, where its very high conversion rate flatters every other keyword in the account and hides the performance of the terms that are actually finding you new customers.
Why does Performance Max hide most of this?
Because it does not report at the keyword level and never did. You get a search themes view that groups terms into categories rather than listing them, so the specific search that took the money is often not visible at all. The control you have is the exclusion list, not the keyword list.
That changes the job rather than removing it. Apply your account level negative list to Performance Max, add brand exclusions where you want brand traffic separate, and watch the asset group reports for drift instead. We cover what that campaign type lets you steer in our breakdown of Performance Max.
How often should you read the report?
Weekly on an account spending enough to produce new terms, monthly below that. The discipline matters more than the frequency, because the leak reopens continuously: close variants shift, new searches appear, and a term that converted in spring stops in autumn. A list built once and never revisited is out of date within a quarter.
Fifteen minutes a week is the whole commitment, and it is the best return per minute of any routine task in a search account. It does depend on your conversion data being trustworthy, which a tracking setup that survives has to solve first. On a small budget it matters more, because a small budget cannot afford the wrong searches at all.
What does a clean account look like?
Spend concentrated on searches that describe what you sell, a shared negative list you no longer think about, brand traffic reported separately from everything else, and no term above your allowable cost per lead without a conversion behind it. Nothing clever, just an account where the money follows intent.
That is the baseline before the more interesting work is worth doing. Bid strategies, creative testing and budget increases all amplify whatever the account is buying, so they amplify the leak too. Our Google Ads management starts here, and the same logic runs through how we build lead generation. To have the leaks in your own account found and ranked by what they cost you, book a pipeline review.