Christmas break is happening at the end of this week for me, so what the heck, let's stir up the pot a little!
Over the past few years I've been pondering the idea of Paid Keywords and their role in marketing, as well as specifically how their targeting has changed within the Google Ads ecosystem.
Based on this, I'd like to push back on something I hear a lot, the argument goes something like this:
"Well, keywords are just topics anyway."
This argument typically is given after some sort of discussion on close variants, the usage of broad match keywords, the expansion of audiences within Google Ads, or even the utilization of PMax in an account.
Look, there is much I agree with in this statement, and it's helpful to begin with that.
I believe what is meant here, is that we can't be so rigidly obsessive over specific phrases, that we forget that people overall mean different things by what they say, and overall people can say different words to mean the same thing.
In other words, people asking Google for the: "best treadmill desk" and "most durable treadmill desk" and "a treadmill desk that won't break on me" are all, arguably, wanting to see the same ad. Let's say, an ad for a higher end treadmill made with more commercial grade materials that just works for them. In this case, the searchers have different starting places, but really are speaking about the same topic.
Here's where things get muddy, and why I think the above statement is an unhelpful oversimplification of the Paid Search world of targeting:
Keyword targeting isn't JUST about topics, it's also about intent. Someone may be interested in the same topic, but have very different intent. This is where things like close variants can really fall apart, and why I've argued before for a return to making "exact match be exact". By this, I simply mean allowing for expansion by Google as budget allows in the phrase and broad matches, but allowing for a purposefully restrictive match type like Exact to remain for more advanced marketers. Yes, even for misspellings and plurals, as Brad Geddes has previously made a case for. In other words, treating Exact match like Negative Keywords are treated.
Actually, pause to ponder that point a bit, if Google truly believed that close variant behavior was the best way to manage targeting, wouldn't they also apply close variants to your negative keywords as well? If their highest objective (as they note for a keyword close variants defense) is to allow for helpful targeting shortcuts based on semantical connection between various words.. wouldn't they also apply that same logic to negative keywords that would "help" an advertiser be more restrictive to their specific topic/intent? Clearly there is more at play here for Google than simply finding the best semantical matching, and we would be wise to keep that in mind with close variants as well.
That being said, words are topical, but we can also identify actual intent-based trends based on historical performance. In other words, the exact match keywords: [contractor license] and [licensed contractor] appear to be the same word with the order switched around, but they have very different meanings (see this old but still insightful article by Brad Geddes) and Google doesn't always get this correct (to which any of us with Search Terms Reports can attest). There are even times when misspellings and plurals show dramatically different purchase intent by users. Why? I don't know why. I don't know if we can know why, and that's exactly the point. People don't always make sense, but we can still set unique bidding targets and create unique ads for semantical phrases with matched purchase intent. In other words, if an exact match term converts better than another, even if they are very close in supposed intent and they clearly are within the same "topical realm", it still behooves the wise advertiser to give those terms dedicated bids/budgets/ad text/landing pages to eke out all the difference in performance they can.
You see, in that instance, the specific arrangement of words IS the audience. We have to stop pitting audiences against keywords as if they're two separate entities. Sometimes the keyword needs to be seen as the audience based on historical performance, and then we can go about optimizing. That's old school Paid Search, but that's because it matches the beauty of marketing keywords. Keywords aren't just semantical phrases. We can't assume we know what a keyword will be simply because it appears to be high intent, and that's why keywords still matter as unique targeting entities.
They are far more complex than simple topics. They are living, breathing marketing targets that have topic and demographic and intent based activity within them, and sometimes the very best way to manage them is to look at performance, and then pull them out into separate entities and optimize the snot out of those individual terms. This is especially true with limited budget Google Ads accounts. I still manage my wife's local photography business ad account with $250/mo successfully by focusing almost solely on Exact (and some phrase) match keywords (while keeping up on negatives!) because exact match keywords are fantastic for limited budgets!
"But you have to trust the machine!!" We've been trying, our search terms reports continue to get in the way of this trust. Just yesterday I was in an account for a prospect preparing a proposal for them, and pointed out that Google was close variant matching a NON-brand ultra generic term to their decidedly BRAND (not generic brand either) exact match keyword. Well, perhaps they perform at a similar level so Google is just finding high value customers typing in general terms? Nope, not in this account. The non-brand keywords were dramatically underperforming historically in the account while the Brand keywords were performing at a high ROAS level. There was enough data in the account. Machine Learning failed. Watch social media long enough and you'll see case after case of this. Here's the problem, it's not that Google always fails... it's that it fails enough, in enough accounts, to really mess up the targets and budgets of a lot of accounts. In many cases like this, the business owner doesn't even notice because they don't really understand what's going on, so they could be in essence FAR overbidding for those "generic" terms that are close variants for the Exact Brand keywords since those terms tend to be higher bid than generic NB keywords. That's why it's an actual issue, with close variants in those cases, the advertiser ends up spending more than they would have otherwise while advertisng THOSE SAME GENERIC terms elsewhere in the account at lower bids... and that's why it's difficult to trust Machine Learning. We've seen too much stuff like this.
So, rambling soapbox down, all that to say, I think it's helpful to view keywords as having a topical element. I would agree with many of the conclusions we drew in our PPC Ponderings Podcast episode on the keyword:
But with that in mind, there is certainly a purchase intent that needs to be taken into account, even when topical behavior is shared across two search terms... and close variants can struggle with this. I think there will always be a place for exact match keywords if Google will allow for them.
Them's my $0.02.