Short answer: on a small Google Ads account, start with phrase match and exact match on your five to fifteen most valuable keywords, add broad match only once you have enough conversions for Smart Bidding to learn from, and keep a negative keyword list running from day one. Broad match is not the villain it was in 2018, but it is also not a free lunch on a €1,000 monthly budget.
Match types decide which searches can trigger your ads. Pick the wrong one and you either pay for irrelevant clicks or you starve the account of the volume it needs to learn. Here is how we think about the three options in 2026, what has changed, and a simple order of operations for small accounts.
What do the three match types actually do in 2026?
The names are the same as they were years ago, but the behaviour has drifted toward meaning rather than literal text.
- Exact match shows your ad for searches with the same meaning or the same intent as your keyword. It includes close variants such as plurals, misspellings and reordered words, so it is no longer “only this exact string”.
- Phrase match shows your ad for searches that include the meaning of your keyword, with extra words allowed before, after or in some cases around it. It is wider than exact but still anchored to the keyword’s meaning.
- Broad match is the widest. Google can match searches that relate to your keyword even when they do not contain its words, using signals such as your landing page, your other keywords in the ad group and what the searcher did recently.
The practical consequence is that every match type now has some flexibility. The question is how much freedom you hand to the system, and whether your account has the data to make that freedom pay.
Which match type should a small account start with?
For a new account spending €500 to €1,500 a month, we start with phrase and exact. The reasoning is simple: you do not yet know which searches convert, and every irrelevant click costs a meaningful share of a small budget.
A typical starting structure looks like this:
- One campaign per core service or product line, so budget and reporting stay readable.
- Within each campaign, one or two tightly themed ad groups.
- Exact match for the three to five terms that describe exactly what you sell and that you would be happy to pay for any day of the week.
- Phrase match for the next layer of related terms, including the typical “service + city”, “service + price” and “best service for X” patterns.
- A starter negative keyword list, covering jobs, free, DIY, course, template and any adjacent products you do not sell. Our guide to building a negative keyword starter list has the full set of categories.
This setup gives you control and a clean first month of data. It will not give you maximum volume, and that is fine. Volume is what you are trying to buy once you know what converts.
When does broad match make sense?
Broad match earns its place under two conditions: the campaign uses Smart Bidding, and it has enough conversion data for that bidding to learn. Google’s own guidance has long paired the two, and our experience agrees. Broad match without a bidding strategy that optimises toward conversions is just a wider net with no filter.
As a rule of thumb we use, not a platform rule: if a campaign is not producing roughly 30 conversions in a month, we keep broad match out of it or test it in a separate, capped experiment. Below that volume, the bidding system has too few examples of what a good click looks like, and broad match gives it a very large space to guess in. Our post on choosing between Maximize Conversions and Target CPA explains how to think about that data threshold, and how long to wait before changing anything covers the learning period that follows.
Good candidates for broad match are:
- Campaigns where phrase and exact are already converting steadily and you want to find adjacent queries you did not think of.
- Businesses with many ways of describing the same need, where customers phrase the problem rather than name the product.
- Accounts with solid conversion tracking. If tracking counts junk events as conversions, broad match will happily find you more junk.
Poor candidates are brand-new accounts, campaigns with fewer than a handful of conversions a month, and businesses whose product names are also common words.
What does the difference look like in money?
Here is an illustrative calculation, not a client result. Imagine a €1,200 monthly budget and an average cost per click of €1.50, which gives about 800 clicks. Suppose a landing page converts 5% of clicks into leads when the search is tightly relevant, and 2% when it is loosely relevant.
| Scenario | Relevant clicks | Loose clicks | Leads | Cost per lead |
|---|---|---|---|---|
| Phrase + exact, 85% relevant | 680 | 120 | 34 + 2.4 = about 36 | about €33 |
| Broad, 55% relevant | 440 | 360 | 22 + 7.2 = about 29 | about €41 |
The numbers are assumptions chosen to show the mechanism. The point is that broad match can dilute the share of clicks that behave like your best traffic, and on a small budget that dilution shows up directly in cost per lead. With enough data and a good bidding strategy, the system can learn to bid down on the loose clicks. Without it, you pay full price for them.
How do you keep each match type honest?
Match types are not a set-and-forget choice. The search terms report is where you see what you actually bought. We check it weekly on small accounts, and the routine takes about fifteen minutes:
- Sort the search terms by cost, highest first.
- Mark anything clearly irrelevant as a negative keyword, choosing exact or phrase negative depending on how widely you want to block it.
- Look for converting search terms that are not in your keyword list yet. Add them as exact or phrase keywords so they get their own bid and message.
- Check the share of spend going to terms with no clear connection to your offer. If it is above roughly a fifth of the spend, tighten match types or add negatives before raising budget.
Remember that negative keywords behave more literally than positive ones. They do not expand to synonyms or close variants the way positive match types do, so you may need to add several forms of a word you want to block.
Common mistakes we see
Switching everything to broad because a rep suggested it. Platform recommendations are written for the average advertiser with a large account. Test them on one campaign with a clear before-and-after window.
Mixing match types of the same keyword in one ad group with no thought. Google will usually serve the most specific match, but your reporting becomes harder to read. Keep exact in its own ad group, or at least label it clearly, so you can compare performance.
Judging match types after three days. Click-level noise is huge on small accounts. Give each change at least two weeks and a few hundred clicks before concluding anything.
Ignoring the landing page. Broad match leans on your landing page content to decide what you are relevant for. A vague page invites vague traffic. If you are also working on ad quality, the piece on Quality Score and the three fixes that matter shows how page relevance feeds into cost.
A simple order of operations
If you only remember one sequence, use this one. Launch with exact and phrase plus negatives. Collect 30 to 60 days of search term data and conversions. Fix tracking if the numbers look odd. Then test broad match in one campaign, with Smart Bidding on, a capped budget and a written success rule: for example, “cost per lead no more than 20% above the phrase and exact campaign after 30 days”. If it meets the rule, expand. If not, switch it off and you have learned something cheaply.
Match types are a dial between control and reach. Small accounts need control first. Reach is something you earn once the account has proved it can turn clicks into customers.