Broad vs phrase vs exact match: which wastes money
Broad match, phrase match and exact match decide how far Google may stretch a keyword. Broad match vs phrase match is the choice most accounts get wrong, but exact match is not exact either: in the founder's account, keywords set to exact matched 600 different searches in six months, and only 28 of them were the keyword itself. Here is what each type really matches, and how to tell from your own account which is costing you.
The three match types in one table
| Match type | Typed as | Google may show your ad for |
|---|---|---|
| Broad | store locator | Searches related to the keyword's meaning, judged with other signals such as your landing page and the searcher's recent searches. "map of shops near me" can qualify. |
| Phrase | "store locator" | Searches that include the meaning of the keyword, with words before or after: "best store locator for shopify", and often "shop finder app". |
| Exact | [store locator] | Searches with the same meaning or intent as the keyword: plurals, misspellings, reordered words and what Google calls close variants, such as "store finder". |
Each step down the table narrows what you pay for. None of them is literal. Since 2021 phrase match has absorbed the old broad match modifier, and exact match has matched close variants for longer than that. The match type sets how far the stretch goes, not whether there is one.
What six months of real searches show
The founder's account ran 104 exact-match and 304 phrase-match keywords between May and October 2026. Every search that triggered an ad is in the search terms report, with the match type that caught it. Grouped by that column:
| How the search matched | Different searches | Cost | Sign-ups | Cost per sign-up |
|---|---|---|---|---|
| Exact match, the keyword itself | 28 | $1,265 | 2 | $633 |
| Exact match (close variant) | 572 | $3,914 | 10 | $391 |
| Phrase match, the keyword itself | 25 | $131 | 0 | none yet |
| Phrase match (close variant) | 227 | $1,825 | 1 | $1,825 |
Three things stand out, and none of them is the usual advice.
- Exact is mostly not exact. 572 of the 600 searches caught by exact-match keywords were close variants. That is where most of the spend went, and also most of the sign-ups.
- Close variants are not automatically waste. Here, exact-match close variants won sign-ups more cheaply than the exact keywords themselves. Cutting them all would have cut the best traffic.
- The waste hides inside the variants. Phrase-match close variants spent $1,825 for one sign-up. Inside the exact variants sat searches like "location tracker" and "geolocation tools", from people who wanted to track a phone, not add a store map to a website.
The lesson is not "use exact match" or "avoid broad match". It is that the match type column tells you where to look, and the searches themselves tell you what to cut.
Broad match vs phrase match
Broad match is Google's recommended default, especially with automated bidding. It reaches the most searches, including ones you would never have thought to bid on, and it relies on the bidding to pay less for the poor ones. That works when the campaign has enough conversions for the bidding to learn from: Google's own guidance is about 30 a month. Below that, broad match buys a wide spread of searches with very little to steer it.
Phrase match keeps the keyword's meaning at the centre and lets words around it vary. For a small account with a clear product, phrase match usually finds most of the searches broad would, with far fewer surprises. A campaign with AI Max switched on matches even more widely than broad match, so treat it like broad.
Which match type to use when
- A search that has already won customers below your target cost: add it as an exact-match keyword, so it gets its own bid and you see its results on their own line.
- A new campaign with a clear product and fewer than 30 conversions a month: start with phrase match and a short negative list, and read the search terms report weekly.
- A campaign with steady conversions on automated bidding: broad match is worth testing against phrase, in an experiment or a separate campaign, not by switching everything at once.
- Any match type: pair it with negatives. The match type decides how wide the net is. Negatives take out what the net catches by mistake.
How to check your own account
- Open Insights and reports, then Search terms, for the last 90 days or more.
- Add the Match type column if it is not showing. It says which match type caught each search, and whether it was a close variant.
- Download the report, or use the segment options, to total cost and conversions by match type, as in the table above.
- In the group with the worst cost per conversion, sort by cost and read the top 20 searches. Add the ones that could never buy as exact negatives.
- Repeat for the close variants of your exact keywords. That is where a winning keyword quietly lends its bid to searches that are not its own.
AdFabric reads every search your keywords matched, splits them by match type, and prices the ones that spent without a sale. Free scan, no card.
Run a free Waste ScanView a sample reportHow AdFabric reads match types
AdFabric reads the search terms report for every campaign, keeps the match type of each search, and prices the ones that spent without a sale against what a customer is worth to you. A broad match keyword whose searches have cost more than a customer is worth, with nothing to show for it, is its own finding, with the change to make: tighten it to phrase, or add the negatives that would have stopped the waste.
Questions people ask
What is the difference between broad match and phrase match?
Broad match can show your ad for any search related to the keyword's meaning, using other signals such as your landing page. Phrase match requires the search to include the meaning of the keyword, with other words allowed before or after it. Phrase is narrower and more predictable.
Does exact match only show for the exact keyword?
No. Exact match also covers close variants: plurals, misspellings, reordered words and searches with the same meaning. In the account in this guide, 572 of 600 searches caught by exact-match keywords were close variants.
Is broad match a waste of money?
Not always. Broad match works best with automated bidding and enough conversions for it to learn from, about 30 a month. In a small account with few conversions it tends to buy many searches that never sell, so start with phrase match and test broad later.
What happened to broad match modifier?
Google retired broad match modifier in 2021 and folded its behaviour into phrase match. Old modified broad keywords now behave like phrase match.
Which match type should I use for negative keywords?
Usually exact, for a search you have seen and want to block. Negative keywords do not match close variants, so each match type behaves more literally as a negative than as a keyword.