Match type

The names are historical and they mislead. All three types have loosened over the past decade, so it is better to think of them as three widths of net than as three degrees of literalness.

Broad match in particular no longer behaves like a keyword setting at all. It uses the keyword as one signal among many, alongside the landing page, the other keywords in the ad group, and what the platform knows about the person searching.

Close variants arrived on exact match in 2018 and are the single biggest source of surprise in a search account. Google extended exact to searches with the same meaning, not merely the same words, so an exact match keyword now matches synonyms, paraphrases and implied intent.

The reasoning Google gives is that a large share of searches every day have never been seen before, so a strictly literal account misses demand it would have wanted. That is true, and it is also true that looser matching serves more ads.

For an agency the consequence is practical. An exact match campaign left alone will accumulate traffic nobody chose, and the only place that shows up is the search terms report. An account with tight match types and no negative keywords is not controlled, it is unexamined.

This is also the answer to a client who asks why they appeared for a competitor name or an unrelated phrase when the keyword list clearly does not contain it. The keyword list stopped being the whole story in 2018.

Broad match today is closer to a targeting recommendation than a keyword. Alongside the term itself it uses the content of the landing page, the other keywords in the ad group, previous searches in the session, and the platform signals about the person.

This makes it strongly dependent on the quality of the conversion data underneath it. Given a clean signal and a smart bidding strategy, broad match can find profitable searches that no keyword researcher would have written down, and it will keep finding them as language shifts.

Given a poor signal it does the same thing in the wrong direction and at speed. If the account counts page views as conversions, or fires a conversion on a form load rather than a submission, broad match will efficiently buy more of whatever produces that event.

That dependency is why the sequence matters more than the setting. Verify conversion tracking, let the bidding strategy accumulate data on a tighter structure, then widen. Reversing those steps is how accounts spend a month teaching the system the wrong lesson.

There is no correct default, and the confident answers in either direction usually come from someone generalising one account. The right question is not which match type is better but whether this account has a conversion signal good enough to steer a loose one.

Things worth checking before widening: is every conversion action a real business outcome, is the volume enough for the bidding strategy to learn from, is the tracking verified end to end rather than assumed, and does the landing page content actually describe what is being sold.

Small budgets change the calculation. An account with enough spend to gather data quickly can afford to let broad match explore; one spending a few hundred a month will burn most of it on exploration and never reach the part that pays.

Whatever is chosen, the search terms report settles it. Reviewing it weekly for the first month after any match type change is the difference between a decision and a hope.