Broad Match Works. But Is That Really Enough?

I’ve been sceptical of Broad Match for a long time.

That’s not because I believe it can never work. I don’t know everything, I haven’t looked inside every Google Ads account and I’m certainly not going to tell someone getting exceptional results from Broad Match to stop using it because my experience has been different.

But in the accounts I have worked on, I’ve yet to look through the search terms generated by Broad Match and come away thinking there was nothing we could improve. Sometimes that means excluding searches that simply aren’t relevant. More often, the bigger opportunity comes from looking beyond whether a search converted and understanding which searches are actually producing the best results for the business.

That’s why I think asking whether Broad Match “works” sets the bar too low.

If you ask Google to generate leads at a £50 CPA and Broad Match delivers them at £48, you can call that successful. Google has achieved the target you gave it and the business may be perfectly happy with the result.

But that £48 doesn’t tell you whether you could have generated those leads for £40. It doesn’t tell you whether some of the searches you paid for were irrelevant, whether others produced much better quality leads or whether some of that budget could have generated more value somewhere else.

What does “works” actually mean?

A lot of the discussion around Broad Match seems to end up in one of two camps. Broad Match works or Broad Match doesn’t work.

I don’t think either is particularly useful.

A campaign can spend £10,000, generate 200 conversions at £50 each against a £55 Target CPA and objectively be doing what you’ve asked it to do. But if a meaningful proportion of that £10,000 was spent on searches that were irrelevant, poor quality or commercially less valuable, I’d struggle to call that exceptional performance.

And that distinction becomes even more important when budget is limited. Every pound spent on a search that adds little value is a pound that isn’t available for a better search, another campaign or potentially another channel entirely.

That’s why I’m much more interested in what’s happening underneath the headline CPA or ROAS.

Broad Match doesn’t need to be “crap” for there to be an opportunity to improve it. In fact, it may be doing exactly what it was designed to do: using the information and signals available to find enough conversions at a cost that satisfies the target you gave it.

The question for the advertiser is whether that’s good enough.

Look at what you’re actually paying for

One of the simplest things you can do is spend some time in the search terms report. I’ve yet to work on an account where I couldn’t fairly quickly find opportunities there. Sometimes it’s the obvious stuff: searches that simply aren’t relevant to the business and can be excluded. But that’s really only the first layer.

The more interesting work starts when you look at the searches that are relevant and ask whether they should really be treated equally. A search doesn’t automatically become good because it generated a conversion and two searches relating to exactly the same product can represent very different levels of intent.

None of the searches on the left are irrelevant. They’re all perfectly reasonable searches for businesses offering those products to appear for. But compare them with the searches on the right and the likely difference in commercial intent is clear.

Broad Match and Smart Bidding will, of course, consider far more than the words in the search. Google can use other signals, previous search activity and the likelihood of a conversion when deciding whether to enter an auction and how much to bid. But ultimately, if the system can include searches from both sides of that table and still achieve the Target CPA or Target ROAS you’ve given it, it may be doing exactly what you asked it to do.

That doesn’t necessarily mean it’s the most efficient mix of traffic for the business.

Google may be perfectly happy generating enough conversions or conversion value across those searches to achieve the overall target. Our job is to go further and understand whether particular types of searches consistently convert better, produce more qualified leads or ultimately create more valuable customers.

That also means we shouldn’t simply assume that the searches on the right will always perform better. The data might tell us something completely different. The important point is that hitting the bidding target doesn’t remove our responsibility to look for opportunities beyond it.

And this is where search term analysis becomes much more interesting than simply finding irrelevant searches to exclude. Once you’ve cleared out the obvious waste, you can start understanding which types of relevant searches actually produce the best results and use that information to improve the account further.

If we stop our analysis at “it converted”, we potentially miss all of that. There’s a progression from relevance, to conversion, to lead quality and ultimately to commercial value. The further we can connect our paid media decisions along that journey, the better the decisions we can make.

Search intent is only one part of this. Even when the language looks remarkably similar, performance can change depending on the context in which that search appears. The same word or modifier can behave very differently across products, campaign themes or audiences, which is another reason why broad averages can hide useful information.

Small differences in searches can produce very different results

This isn’t only about separating research-led searches from those that appear more commercial. Sometimes the differences are much less obvious.

I saw a good example of this recently while analysing a financial services account. Searches containing the word “deal” appeared across campaigns targeting broader vehicle searches and campaigns built around specific vehicle types. On the surface, you could reasonably group those searches together and make a judgement about whether “deal” was a useful modifier.

Once we separated them, they behaved very differently.

In the broader vehicle campaigns, searches containing “deal” converted at close to 10%. In the specific vehicle campaigns, the conversion rate was closer to 4%. However, the clicks in those specific campaigns were substantially cheaper, meaning the resulting CPAs ended up relatively close.

If we’d stopped at CPA, there wasn’t much to see.

Digging further showed two very different patterns of behaviour. And the answer wasn’t simply to decide that “deal” was either a good or bad search modifier. It gave us something to investigate, structure and test differently.

There’s another important point here: neither example was using Broad Match. The campaigns were using Exact Match.

Exact Match certainly isn’t the Exact Match many of us managed years ago. Close variants and Google’s interpretation of intent mean even Exact Match keywords can trigger searches you might not have expected from looking at the keyword sitting in the account.

If we can find this much variation underneath Exact Match, there’s even more reason to interrogate Broad Match when Google has considerably more freedom over which searches we appear for.