The essentials
- Three types in 2026: broad match, phrase match and exact match; the broad match modifier disappeared in 2021 and close variants now apply to every type.
- Practical rule: exact for queries that already convert, phrase to explore around a theme, broad only with an automated bidding strategy and reliable conversions.
- The real lever: negative keywords, which have no close variants and remain the only way to shut off delivery.
- Weekly check: the search terms report, even though it has hidden a growing share of queries since 2020.
The match type determines which searches trigger your ad. This setting has changed a great deal: the broad match modifier was removed, close variants were extended, and artificial intelligence now handles interpretation. What you learned in 2019 no longer describes what happens in 2026. Here is what each type actually does, how to choose between them depending on how mature the account is, and how to stay in control.
The three match types and how they really behave
| Type | Syntax | What it triggers in 2026 | Example for « plumber lyon » |
|---|---|---|---|
| Exact match | [plumber lyon] | Searches with the same meaning or the same intent, including rewordings and synonyms | « plumber in lyon », « plumbers lyon 69 », « plumbing tradesman lyon » |
| Phrase match | "plumber lyon" | Searches that include the meaning of the phrase, with words before or after | « cheap plumber lyon », « find a plumber in lyon », « emergency plumbing lyon » |
| Broad match | plumber lyon | Searches related to the topic, based on the keyword, the other keywords in the ad group, the landing page and audience signals | « water leak villeurbanne », « heating engineer lyon », « dripping tap » |
The key point: in 2026, even exact match is no longer literal. Google matches it to any search judged to carry the same intent. The difference between the three types is a degree of openness, not a clear boundary. On a French services account, an exact match keyword receives on average 15 to 30% of its impressions on searches that differ from the keyword itself.
Choosing the type according to the account's situation
The right type depends on two things: the number of conversions measured and how reliably they are measured. Broad match relies on Smart Bidding signals to decide which queries deserve a bid; without reliable conversions, it spends on anything.
| Situation | Recommended type | Why |
|---|---|---|
| New account, fewer than 15 conversions per month | Phrase match and exact match | The volume does not let Smart Bidding sort queries; manual control comes first |
| Established account, 30 conversions or more per month, tracking verified | Broad match with target CPA or target ROAS | The algorithm has enough data to bid low on distant queries |
| Brand keywords | Exact and phrase match, never broad | Stops the brand campaign absorbing generic queries and inflating its results |
| High-CPC sector (legal, insurance, B2B: €5 to €15 per click) | Exact first, phrase for discovery | Every off-target click costs too much to let the algorithm learn |
| E-commerce with Performance Max running | Exact and phrase on Search | Performance Max already covers the long tail; broad match on Search would create internal competition |
An account can combine types: the same keyword in exact and in phrase match in two separate ad groups, with different bids or different ads. Google gives priority to the exact match keyword identical to the query, then to the one with the best ranking; do not count on a strict hierarchy between groups. How to organise these groups is described in structuring a Google Ads account.
Negative keywords, the real control tool
Negative keywords have no close variants: an exact negative [free] only blocks the search « free » on its own, a phrase negative "free" blocks any search containing that word in that order, and a broad negative free blocks any search containing that word. Three practices make the difference:
- Shared lists at account level: jobs, training, free, reviews, definition, competitors; to be applied to every Search campaign and, since 2025, to Performance Max.
- Cross-campaign negatives: brand terms as negatives in the generic campaigns, and the reverse, so that each campaign only spends on what it is meant to.
- Weekly review of the search terms report: sort by cost, filter to zero conversions, and add as negatives everything that exceeds the cost of a conversion without producing one.
Since September 2020, Google has hidden low-volume search terms; on many accounts, 20 to 40% of spend goes to queries you never see. That is the main argument for keeping exact match keywords on the queries that carry the budget, and for monitoring overall cost per conversion rather than each query; this metric is detailed in cost per acquisition.
Broad match and Smart Bidding: the pairing Google recommends
Since 2021 Google has been pushing the combination of broad match and automated bidding, with a valid argument: the algorithm bids low on unpromising queries and high on those that resemble past conversions. On accounts with more than 50 conversions per month, this combination widens coverage by 20 to 40% at a comparable cost per conversion. On smaller accounts, it mostly produces off-target clicks. Before switching it on, check three conditions: at least 30 conversions over 30 days, conversions that correspond to real sales or enquiries, and a negative list that is already solid. How automated strategies work is detailed in Smart Bidding.
Checking that the setting produces the expected effect
Add the « Match type » column to the search terms report and compare, for each type, average CPC, conversion rate and cost per conversion. On a well-tuned account, broad match has a lower CPC but a lower conversion rate, and its cost per conversion stays within 20% of exact match. If the gap exceeds 50%, broad match is spending on queries that the negatives are not filtering out well enough. Tracking by match type is one of the regular checks described in what a paid search audit covers.
How GreenRed helps
Rather than juggling several tools, GreenRed's Google Ads module brings these metrics together in a single dashboard, compares them over time and tells you which actions come first. You can try it free, with no card, from the Pricing.
Frequently asked questions
Does the broad match modifier still exist?
No. Google merged it with phrase match in 2021: keywords with the + sign were converted and now behave like phrase match, with word order taken into account loosely. The old keywords are still visible in some accounts but can no longer be created.
Why does my exact match keyword trigger other searches?
Since 2018, exact match has included close variants: typos, plurals, rewordings, synonyms and searches with the same intent. Google alone decides how close they are. The only way to block an unwanted variant is to add it as a negative keyword, in exact or phrase match depending on the case.
Should you still use phrase match?
Yes, it is the most balanced type for an SME account: open enough to discover queries around the theme, closed enough to stay readable in the search terms report. It is the recommended default type for launching an ad group, before moving the best queries to exact match.
Do match types exist in Performance Max?
No. Performance Max does not use keywords but signals, search themes and asset content. Since 2025, it has been possible to add negative keywords at campaign level and brand exclusions, which gives partial control, but without the match types of Search campaigns.