Most explanations of keyword match types are still teaching a version of Google Ads that stopped existing some time ago. Three neat boxes, increasing precision, pick the one that suits your budget. At The Growth Bully, a Malta performance marketing agency, we open accounts every month that were structured on that advice and are quietly paying for searches nobody would have chosen to buy.
The useful question is not which match type is tightest. It is how much interpretation you are willing to let the system do on your behalf, and what you have in place to catch it when the interpretation is wrong.
What do keyword match types actually control now?
They control how far the system may travel from the words you typed, not whether it may travel at all. Every match type now includes close variants: misspellings, plurals, stemmings, abbreviations and, at the looser end, searches with the same meaning and none of the same words. Match type sets the length of the lead, not the presence of one.
That matters because the old mental model treated exact as a contract and broad as a gamble. In practice all three are interpretations of intent at different confidence thresholds, and the system is making the judgement in every case. You are choosing how much benefit of the doubt to extend.
Is exact match still exact?
No, and it has not been since close variants were extended to same meaning searches. An exact match keyword can now serve a query that shares no words with it, provided the system reads the intent as equivalent. Plurals, word order, function words and synonyms are all in scope, and there is no setting to switch the behaviour off.
There is one practical consequence worth internalising. If exact match can drift, then the only place you can see where it drifted is the search terms report, and the only place you can stop it is the negative keyword list. We cover both in where Google Ads budget leaks.
When does broad match genuinely work?
Broad match works when three things are true at once: the account has enough conversion data for smart bidding to steer on, the negative list is already mature, and there is enough search volume for the extra reach to find something you were missing. Remove any one of those and broad match becomes an expensive discovery exercise.
When it does work, it works well. Broad match plus a reliable conversion signal is now the most efficient way to find the phrasings you would never have thought to add, because the system is testing hundreds of them a week and you would be adding four. The condition is that it has to be told what a good outcome looks like first.
Why can tight match types starve a campaign in a small market?
Because precision costs volume, and in a market of a few hundred thousand people there may not be enough left to lose. A tightly matched campaign in a narrow Maltese niche can sit on a few impressions a day, never gather enough conversions for bidding to learn, and look like a targeting failure when it is a sample size problem.
This is where the standard advice inverts. The usual guidance to small budget advertisers is to go narrow and protect every euro. In a small market that often produces a campaign so restricted it never exits learning, which is why we weigh reach against control differently here. More on the budget side of that in advertising on a small budget.
Should you run the same keyword in several match types?
Rarely, and almost never for the reason people do it. The old practice of mirroring a keyword across all three types was a way to control which version got the click, and that control has largely gone: the system now prefers whichever keyword it judges most relevant, so duplicating mostly fragments your conversion data across near identical entries.
There is a narrow case for it. If a term is commercially critical and you want separate budgets for its tight and loose interpretations, splitting by campaign rather than by keyword inside one ad group keeps the reporting readable. Otherwise one keyword, one match type, and let the negatives do the shaping.
How do negative keywords change what match type you can afford?
They decide it, which is the part the textbook explanation leaves out. A mature negative list is what makes loose matching safe, because every interpretation you have already rejected stops being repurchased. Without one, broad match is simply a faster way to spend budget on searches you would recognise as wrong in a second.
Negative match types follow their own rules, and the one that catches people is negative broad: it blocks any query containing all of your words in any order, which is powerful and easy to over apply. Build the list in this order:
- Job and career terms first, because they appear in almost every service account and convert at zero.
- Free, DIY and cheap modifiers, unless you genuinely sell at the entry end.
- Adjacent services you do not offer, which is where most of the waste in a broad campaign actually sits.
- Wrong geography, which location targeting alone does not fully solve because intent and location are different signals.
- Research phrasings such as how to, meaning, example and template, kept at campaign level so informational campaigns are unaffected.
Keep the first four at account level and the last at campaign level. A single account wide list applied without thought is the most common way a campaign loses traffic it wanted.
What should a new account start with?
Phrase match on a small set of terms you are confident about, with exact match reserved for the two or three phrasings you know convert, and broad held back until the conversion signal is real. That sequence buys data at a manageable risk rather than paying for discovery before anything is known.
Then read the search terms report weekly for the first month and move what you learn in both directions: new negatives out, proven phrasings in as their own keywords. Campaigns that never get read drift regardless of how carefully they were built, and Performance Max hides more of this than search campaigns do, as covered in Performance Max explained.
How do you tell a match type problem from an offer problem?
By looking at what the traffic did rather than what it was. If the search terms are clearly right for your business and still nobody converts, the match type is not the fault and tightening it will only reduce the number of people declining your offer. That is a landing page or a proposition problem wearing a targeting costume.
If the search terms are visibly wrong, drifting into other services, other countries or research intent, then matching and negatives are the fix and they are cheap. The diagnosis takes fifteen minutes and saves months of pulling the wrong lever, which is the single most expensive habit in paid search.
If you want the structural version of this done once and kept clean, our Google Ads work starts with a search terms audit, and you can also see what a realistic cost per enquiry looks like in what a good cost per lead looks like. When you want the leaks found and closed rather than listed, ask us what your account is buying.