Most B2B search accounts carry three copies of the same keyword and pay for the privilege of splitting their own data.
If your account produces ten to forty qualified conversions a month, running a term as exact, phrase, and broad in three parallel ad groups does not buy you control. It buys one bid, three thin rows of data, and a reporting job nobody wants. Google's own documentation is blunt about it. Broader match types capture every query the narrower ones do, so carrying all three of the same keyword adds nothing you did not already have.
The decision worth making is not which match types to keep. It is how many genuinely different buyer intents your conversion volume can afford to hold apart. In most mid-market B2B accounts that number is small, and the keyword list is built as if it were large.
This one is for whoever inherited a search account with four hundred keywords, a dozen ad groups named SKAG, and a monthly qualified lead count you could write on one hand twice.
Google says the mirrored ad group is pointless, in writing
Start with the source rather than the folklore. The About keyword matching options help page states that broader match types capture all the queries of narrower match types plus more, and then says directly that you get the same benefits of multiple match types in one broad match type without needing to repeat keywords.
The same page explains what happens when you do repeat them. Near-duplicate keywords are recognized as duplicates, the one with the higher Ad Rank is used, and you have only one bid in the auction. You are not stacking coverage. You are choosing which of your own rows gets the impression, and Google is choosing for you.
That single fact kills the standard defense of the mirrored structure. People keep exact, phrase, and broad versions of a term because they believe each one catches searches the others miss and each one gets its own bid. Neither is true anymore.
What the mirrored structure does reliably produce is fragmentation. Twelve conversions a month spread over nine keyword rows is one conversion per row per month. No bid strategy, no statistical test, and no human reading a report can do anything useful with that. It is the same arithmetic problem that wrecks over-split campaign structures, showing up one level further down in the account.
Ready for paid ads that pay off?
Book your free auditPhrase match is the right default for B2B core terms
For the terms that describe what you actually sell, phrase match is where most B2B accounts should sit.
Exact match now covers searches with the same meaning or same intent as the keyword, not the literal string. Phrase match covers searches that include the meaning of the keyword, where the meaning can be implied and the search can be a more specific form of it. Both of those definitions are broader than what the brackets and quotation marks used to promise, and the practical gap between them in a low-volume account is small enough that maintaining both costs more than it returns.
Phrase wins the default slot for one reason. B2B buyers search in long, qualified strings that you will never predict. Someone looking for your category writes six words, four of which are about their own situation. Phrase match catches the specific version of the intent you wrote down. Exact match frequently does not, and the searches it drops are often the good ones.
The practical setup for a core commercial term looks like this.
→ One phrase keyword per intent, not per phrasing variation.
→ One ad group per intent, where intent means a different problem the searcher is trying to solve, not a different word order.
→ Exact match reserved for the specific cases below, and branded terms kept in their own campaign so the branded and non-branded numbers never mix.
If you cannot write one sentence explaining what a searcher wants that is different from the ad group next door, the two ad groups are one ad group.
Exact match has become a Performance Max defense, more than a precision tool
Here is the take people push back on. In 2026, the strongest argument for exact match in a B2B account is not precision. It is priority.
The same Google help page describes how Performance Max interacts with your Search keywords. If a search query is identical to an eligible exact match keyword in your Search campaign, the Search campaign is prioritized over any broad or phrase keyword and over any Performance Max campaign. Phrase and broad keywords only share priority with an identical Performance Max search theme.
So exact match is the one thing that reliably stops Google's more automated campaign type from taking a query you specifically want your controlled Search campaign to answer. That makes it worth using, and worth using narrowly.
Where exact match earns its row.
→ Your brand name and your main brand plus category combinations.
→ The five to fifteen head terms you are unwilling to hand to an automated campaign.
→ Competitor names, where the literal query matters and the surrounding intent does not.
→ Anything with a legal, compliance, or pricing sensitivity where you want one specific ad against one specific query.
Outside that list, an exact keyword sitting next to its own phrase version is a maintenance cost. If you run Performance Max alongside Search at all, this priority rule is the part of the Performance Max versus controlled Search decision that most teams never read.
Broad match is something you earn, not a setting you flip
Broad match is the default assigned to every keyword, and it is also the match type most likely to burn a B2B budget in a fortnight.
Google is explicit that broad match needs Smart Bidding, because every query is different and the bid should reflect the auction-time signals. It also says broad match may take the user's recent search activity, the content of your landing pages and assets, and the other keywords in the ad group into account when it decides what counts as related.
Read that list again from a B2B seat. Your landing pages, your asset copy, and the rest of your ad group are all being used as evidence about what you meant. If your pages are generic and your ad group is a mixed bag, the evidence is bad and the matching follows.
Three conditions have to be true before broad match is a reasonable idea in a B2B account.
→ Your conversion action is a qualified outcome, not a newsletter signup. Broad match plus a junk conversion signal is an efficient machine for buying the wrong traffic, which is the whole argument in what you feed Smart Bidding.
→ Somebody reads search terms weekly and adds negatives. Not monthly. Weekly, as a standing discipline.
→ The ad group is tight and the landing page is specific, because both are inputs to the match, not just the destination.
Miss any of those and broad match is not an expansion strategy. It is a subscription to search terms about jobs, free templates, and university courses.
Negative match types do not behave like positive ones, and that gap leaks money
This is the part that catches experienced people, because it breaks the symmetry everyone assumes.
Negative keywords do not match close variants or other expansions. Google's About negative keywords page gives the example directly. Exclude the negative broad keyword flowers and your ads are blocked from red flowers, but they can still serve on red flower. Singular and plural are two different exclusions. Synonyms are two different exclusions. Casing and misspellings are handled for you, and nothing else is.
Positive keywords have spent a decade getting looser while negative keywords stayed literal. The result is a predictable failure mode. A broad match campaign expands into a whole family of unwanted searches, the team adds one negative per bad term as it appears, and the account keeps paying for the singular, the plural, and the three synonyms nobody thought to add.
How to build a negative list that accounts for the asymmetry.
→ Add singular and plural for every exclusion. Every time.
→ Add the obvious synonyms in the same sitting, while you are still thinking about the term. Free, template, tutorial, salary, jobs, course, and definition each have four or five cousins.
→ Use negative phrase for a concept you want gone in any context, and negative exact only when you want to block one precise query and keep the longer ones. Negative exact blocks running shoes and allows blue running shoes, which is almost never what people intend when they pick it.
→ Keep the account-level list for the universal exclusions, since it applies across eligible search and shopping inventory, and keep campaign-level lists for the ones that are only wrong in one context. A term that is junk for your enterprise campaign can be fine for your self-serve one.
→ Know that a negative sitting past the sixteenth word of a long query will not block it. Rare, but it explains the occasional weird search term that survives a negative you are certain you added.
Consolidating a bloated keyword list in one sitting
Five steps, in this order, and the order matters because each one shrinks the work for the next.
1.) Pull ninety days of keyword data and mark every row with zero impressions. Those are not keywords. They are inventory. Delete them, or pause them if your reporting needs the history.
2.) Group the surviving keywords by intent, on paper, before you touch the account. Write the intent as a sentence describing what the searcher wants. Rows that share a sentence are one keyword.
3.) For each intent group, keep the phrase version and remove the duplicated exact and broad copies of the same term. You are not losing coverage. You are consolidating three rows that were already collapsing into one bid.
4.) Rebuild your exact match list deliberately, from zero. Branded terms, the head terms you are protecting from Performance Max, competitor names. If a term does not have a reason, it does not get an exact row.
5.) Re-add negatives with the singular, the plural, and the synonyms in one pass. Then put the weekly search terms review on somebody's calendar with a name attached, because an unowned routine is not a routine.
Expect the keyword count to drop by half or more, and expect the reports to get less flattering and more honest. Fewer rows with real data in them beats many rows with one conversion each.
When you should not consolidate
Two situations where keeping a term split is correct.
The first is when two match types genuinely point at different landing pages and different offers, and the difference is real enough that a searcher would notice. That is not a match-type decision. It is two products.
The second is when you are running a deliberate, time-boxed test and you need the rows separated to read it. Fine. Put an end date on it, and accept that at twelve conversions a month the test will take a quarter to say anything.
Outside those two, a split match type in a low-volume B2B account is usually an artifact of whoever built the account in 2018, preserved because nobody wanted to be the person who deleted keywords.
Do this today
Open your search keyword report, filter to the last ninety days, and count how many distinct terms appear more than once under different match types. That count is your consolidation backlog. Then check one thing in your negative list. Pick your three most common exclusions and see whether the singular and the plural are both there.
If one of them is missing, you have been paying for that query all year.
We do this cleanup as part of the free audit that starts every engagement, and it is usually the first thing that changes the numbers in our B2B Google Ads work. No new budget, no new campaign type, just a keyword list that matches how matching actually works now.
