Google Ads match types in 2026: what broad, phrase and exact actually do now
Exact, phrase and broad match in 2026: close variants, what phrase absorbed, how broad reads intent, where each belongs in an account, reading the report.
By Daniel Stoychev, Webso Digital · 4 September 2026 · 8 min read
There are three match types left and none of them means what its name says. Exact match has included close variants since 2018, phrase match absorbed the old broad match modifier in 2021, and broad match now reads the landing page, the other keywords in the ad group and the searcher's recent activity to decide what counts as related. The practical rule for most UK accounts in 2026: exact and phrase for the searches you know, broad only with Smart Bidding, conversion data and a serious negative list.
The three, side by side
| Match type | What it matches in 2026 | Keyword | Can show for | Should not show for |
|---|---|---|---|---|
| Exact [ ] | The search, plus close variants: misspellings, plurals, word order, function words, and searches Google judges to have the same meaning | [boiler repair chester] | "boiler repairs in chester", "chester boiler repair" | "boiler service chester" (a different intent, though it sometimes creeps in) |
| Phrase " " | Searches that include the meaning of the keyword, with words before or after; the old modified broad behaviour | "boiler repair" | "emergency boiler repair near me", "cost of boiler repair" | "boiler installation" (a different meaning) |
| Broad (no punctuation) | Anything Google judges related, using the keyword, the landing page, the ad group and the user's recent searches | boiler repair | "heating engineer", "my radiators are cold", "combi boiler not firing" | "boiler suit", though without a negative it may |
Close variants: why exact is not exact
Exact match once meant the exact words. In 2018 Google extended it to close variants with the same meaning, which in practice means plurals, misspellings, reordered words, added or dropped function words, and searches its systems consider equivalent. The change made exact match more useful and less predictable. The keyword [boiler repair chester] will show for a dozen phrasings of the same intent, which is good, and occasionally for a phrasing that is not the same intent, which the Search Terms report catches on Monday. Exact is still the tightest control you have, and it is where the money terms belong.
Phrase: the old modifier with a new name
When Google retired the broad match modifier in 2021, phrase match took on its behaviour: the search has to contain the meaning of your keyword, and words can appear on either side. "boiler repair" shows for "emergency boiler repair near me" and "how much does boiler repair cost", and the second of those is the reason phrase match needs a negative list beside it. Phrase is the workhorse for a service business: broad enough to find the long searches you did not think of, tight enough that the Search Terms report stays readable.
Broad match and Smart Bidding: the deal Google offers
Broad match is now the default when you create a campaign, and every Google recommendation nudges towards it. The offer is this: let the algorithm decide which searches are related, and let Smart Bidding decide what each is worth, and it will find customers your keyword list would have missed. On an account with thirty or more conversions a month, values on those conversions and a negative list grown over a year, the offer is often good. On a new account, or one with poor tracking, it spends the budget on the algorithm's guesses and learns from clicks that never convert.
In 2025 Google added AI Max for Search campaigns, which layers keywordless matching and automatically adjusted ad text on top of your keywords. Treat it the way you treat broad match: a test in one campaign, with the negative lists applied, judged on the Search Terms report and cost per conversion after a month, not on the reach it reports.
Where each belongs
- Brand campaign: exact match only. You know exactly what your name looks like, and the point is to spend as little as possible on people who were coming anyway.
- Money terms, the five to fifteen searches that produce most customers: exact and phrase in the same ad group, so exact takes the search when it can and phrase catches the variants.
- Discovery, the long searches you have not seen yet: phrase match with a wide negative list, read weekly, so the winners graduate to exact.
- Broad match: one campaign, on a mature account, with Smart Bidding, its own budget, and the shared negative lists applied on day one. If its cost per conversion sits above phrase after two months, it goes.
- Shopping: no match types at all. The feed does the matching and the negative list does the excluding, which is why Shopping needs the weekly read more than Search does.
Reading the Search Terms report by match type
The report has a column for the match type that triggered each search, and it is the most useful segment in the account. Sort by cost and look at broad first: it is where the strange searches live. Then look at phrase for the searches that converted and are not yet keywords; each one is a candidate for exact. Then look at exact for close variants that are not the same intent, which is rarer but expensive when it happens. The report also shows the keyword text beside each search term, so you can see which keyword claimed it, and whether that was the one you meant.
What to expect next
The direction is clear: fewer manual controls, more automated matching, and more of the account's behaviour decided by the conversion data it is fed. That makes two things more important, not less: conversion tracking that counts customers rather than page views, and a negative list that tells the algorithm what a customer is not. An account with both can afford to test each new form of automation as it arrives. An account without either is handing the algorithm a budget and a guess.
A test for any account: segment the Search Terms report by match type for the last ninety days and compare cost per conversion across the three. If broad is spending most and converting least, the match types are doing the algorithm's job for it, and the negative list has some catching up to do.
Questions on this topic
Is exact match still worth using?
Yes, for the searches you know produce customers. It is the tightest control left, its close variants are mostly helpful, and it lets you bid on a money term without the algorithm's opinion of what else is related.
Google keeps recommending I switch everything to broad match. Should I?
Not everything, and not all at once. Test broad in one campaign with its own budget and the negative lists applied, and compare its cost per conversion with phrase after a month. Keep the brand campaign on exact whatever happens.
Why does my exact match keyword show for searches I did not want?
Close variants. Google judges some searches to have the same meaning as your keyword, and sometimes it is wrong. The Search Terms report shows which searches each keyword matched, and a negative at exact match removes the wrong ones without touching the rest.
Do match types matter with Smart Bidding?
Less than they did, because the bidding sets a different price for each search, but they still decide which searches the campaign enters at all. A negative keyword still blocks a search whatever the bidding strategy, and a tight match type still keeps the Search Terms report readable.
What is the difference between a broad match keyword and a search theme in Performance Max?
A broad match keyword enters auctions on its own account. A search theme guides Performance Max towards searches without restricting it, and the campaign can still show for searches outside the themes. Neither is a hard filter; the negative list is.