Google’s match types have become a moving target. We asked The Secret Digital Veteran whether they could make sense of the latest changes - and whether paid search is on the right path towards bidding on meaning, rather than simply text.

This week’s update to how Google match types work reminded me of a Personality Test that I find much more reliable than the strangely popular Myers-Briggs test (Sagittarius, in case you were wondering).

First, you need a group of people who know very little about how Google Ads works in practice. If customer support at Google aren’t answering the phone / chat window you could try it out on a few new trainees.

It goes something like this: Take a keyword. Any keyword.

But let’s say it’s google q4 revenue and we add it to our Account in both [Exact] and “Phrase” match types. A user types in google q4 revenue. Only one keyword can fire an ad. Which is it – [Exact] or “Phrase” match?

“The Exact keyword because it’s, you know, exactly the same as the search term!” reply the optimists, optimistically.

“Both keywords are eligible for the search term so it’s the one with the highest bid!” say the pessimists (correctly, at least in the past).

The way to resolve this was to use Negative Keywords. You separate Exact and Phrase keywords into their own Ad Groups and apply Negative Exacts to the Phrase Ad Groups. Similarly, you’d use Negative Phrase to make sure traffic for “don’t always trust smart bidding” (Phrase) didn’t go to “always trust smart bidding” (Phrase) instead.

Not many agencies could be relied on to do this properly.

But in any case, recent changes mean I probably can’t use the match type personality test any more.

Earlier this year, Google said an Exact match keyword that is identical to a query is now always “preferred”. And now, both Phrase and Broad match have this behaviour – and on top of that, when the search term isn’t the same as any keyword, the system will select the nearest keyword based on the meaning of the search term.

Definitely a good thing for those (many) accounts where Negative Keywords weren’t applied as they should have been – and another step in the march towards biddable elements being blocks of meaning rather than blocks of text, and match types describing the closeness of meaning rather than the closeness of textual relationship.

Which is great news if it works – if only because it reduces our reliance on the (always error-prone) negative keyword system to ensure efficiency.

But: reality check! If Google can really understand the meaning of every search query – where the hell is my robot butler?

Our experience with new-and-improved Broad match shows that often traffic relevance is not as-advertised. As long as we still have Search Query Reports (SQRs) showing the actual terms which triggered ads, then we can see this and question it. We need this visibility. We shouldn’t have to just trust that Google is sending relevant traffic even if (perhaps especially if) automation is improving performance overall.

Sometimes I think “Maybe I am cynical. Maybe I should trust Google more.” but then I remember back to a few Christmases ago…

We noticed Impressions were spiking in some of our Shopping accounts. Strange, we hadn’t made any changes to account for it. We looked at the SQRs and we were showing for all kinds of irrelevant queries. Nothing “broad match on Bing”-level irrelevant but still pretty awful. As if someone at Google had pulled a lever to reduce the relevance required for search terms to match keywords across the whole world.

But that lever can’t exist can it?

So that’s when I searched google q4 revenue performance vs target and got some clues about what might be happening.

Anyway, a final thought – another answer to the “personality quiz” question might just be another question: “Why can’t Google design a system that warns you when you have overlapping keywords so you don’t waste money?”

I think we all know the answer to that one, don’t we?