I get it, I’m not a huge fan of the “killing something off” blogposts. I’ve joked about this before when I defended the survival of the marketing funnel.
That being said, I think the title is warranted in this case, because in my opinion (this is also the premise of this article):
Google has indeed killed match type segmentation in Search Campaigns as a worthwhile strategy moving forward for the PPCer.
Notice I didn’t have an opinion voiced in that statement. I am not saying I think it’s best for PPC in the long-run to act in this way, I’m at this point simply stating a fact I believe is based in objective reality. The way automation and close variants work in the Google Ads system in 2022 leaves little-to-no room for Match Type Segmentation.
There will likely be some hesitation in PPC-land, so let me lay out my case.
First, I have traditionally been a staunch defender of match type segmentation.
I literally wrote the article on it… back in 2015: Here's Why You Should Separate Match Types By Ad Group and another one with Google’s close variant match announcement: Match Type Segmentation Wins with Google's Exact March Matchness Update. I even made memes about it:
Match type segmentation has been a key way to minimize wasted spend and target only the keywords that would traditionally lead to higher performance in an account.
This is especially true of budgetary and bidding control. There was a lot of power in maintaining separation of more restrictive types like Exact Match in order to ensure budgets did not get capped on those core terms. All this while excluding those terms in your Broader match type ad groups or campaigns, thus maintaining lower bids in the prospecting terms campaigns and remaining aggressive in exactly the keywords you wanted to target. It was a method that worked well for a number of years, at least in our accounts.
So why the change of heart?
Because, frankly, this just isn’t the way Google Ads as a system works anymore. This is for two core reasons:
- Close Variants
- Increased Algorithmic Intelligence
This brings me to my next point…
Second, Close Variants and increased Algorithmic Intelligence have decreased the need for (or possibility of) match type segmentation
Let’s begin with close variants.
If you’re newer to PPC, then the short of it is, that Google has increasingly over time, changed what user search terms (what a person actually searches for) can actually match to which keywords (the word or phrase an advertiser bids on within Google Ads).
A helpful article on the timeline of events can be found at Karooya: Chronology of Google Ad Match Type Changes. Here is part of their visualized timeline below, to see the whole thing, visit the previous link.
We also cover this in some detail in our initial PPC Ponderings podcast episode on the Changing Landscape of the Search Keyword.
Over the years, Google has increased what it allows to match to more restrictive match types.
For example, in the past, targeting [Star Wars iPhone 13 Pro Max case] would limit you to only people searching for those exact words in that order.
Then, misspellings and word order crept in, so you might match to someone actually typing in: “iphne 13 promax starwars case”
The final nail in the coffin of close variants came with intent matching. In this era, you could theoretically match for someone Google has learned in the past searches for iphone 13 pro max accessories, who also likes Star Wars, and who then searches for “what’s a good phone case”.
The purpose of this article isn’t to argue whether this has been a good progression overall for Google or not. We would again refer you to our podcast episode on the Keyword for opinions around that evolution.
Regardless, what it does mean, is that you can no longer limit your search terms (easily) to the specific ad groups or campaigns in which you want them to show.
We recently did an analysis on an account who had things set up with specific segmentation by ad group. One of the things we showed them is that a core ad group including the word “appraisal” was matching for a lot of inquiries about “value”… without any mention of the word “appraisal”. However, the intent was clearly the same, so in this case Google was hitting a home run. They were finding people with the same intent, who never would have otherwise seen this accounts’ ads.
This is the other side to the way Google has killed match type segmentation, that is, the algorithms are just smarter than they used to be. I remember @siliconvallaeys talking about Moore's Law in regard to Google ML targeting the right terms prob around 2017. In accordance with Moore's Law, that means ML would have doubled TWICE in that time. There's a reason broad match is better than it was in 2017.
This certainly doesn’t mean it doesn’t have significant issues at some time… that’s why humans are still needed, and one of the reasons why it is crucial Google desists its current path of hiding not only process, but reported data as well. Performance Max campaigns do not reveal search terms, though they do share “categories” of search terms “insights”. This does not allow the advertiser to identify additional insights to help service the campaigns, or other marketing efforts such as SEO, that Google may not be aware could be helpful insights to see.
That being said, we at ZATO have absolutely seen the machine increase in its ability to target better according to user intent (like in the example shared above). Overall, our confidence in Google’s abilities to better target close variants to the correct match types, especially when grouped with smart bidding, has increased rather than decreased over time.
So why change anything? Does it matter if things are segmented by match type if this is the case?
Third, in this new era, one runs the risk of limiting exposure more than helping with overly segmented campaigns and ad groups.
Keep in mind that with Smart Bidding models like tCPA and tROAS, Google is processing the data at the ad group level first, so segmenting out exact match terms can remove those signals from the bid strategy… which would have actually improved broad match targeting if left together.
As things move increasingly toward Google Smart Bidding, it will require organizing the account in ways that plays to the strength and better feeds the algorithms. In other words, you can better target broad match keywords with smart bidding… the system will now literally target higher intent queries than it did a few years ago (Moore’s Law comes into play here in a big way).
But it’s helpful to the system to keep exact match keywords next to the broad match in the same ad groups collecting data (as a note on this point, this is what we have surmised from our working in and with the Google system from various documentation points. There is not a current Google policy doc website we can share that states things specifically as such).
Fourth, so how should we organize our Search campaign accounts?
Let me begin, by saying we are still figuring this out along with everyone. Google constantly changing core things in structure requires never-ending adaptation, and we do our best to remain on top of best practices, while being open to testing and evolution.
That being said, we tend to agree with those who have pronounced thematic ad group structures rather than match type structures, while our campaigns then mirror some form of high level navigational client structure targeting Exact and Phrase match keywords primarily.
Let’s say we take on a new DTC Toy Client as a client.. We may organize things in the following way (Campaign > Ad Group).
Search - General Inquiries - US
- Star Wars Toy searches
- Harry Potter Toy searches
- Best Toys in 2022
Search - Brand - Core - US
- Core brand
- Website brand
- Brand + reviews (landing them on a reviews/ratings page)
- Brand + new releases (landing them on a new releases page)
- Etc… (based on other highly searched for keywords with modifiers that would change the landing page we would want to direct the searcher towards).
- Brand + Set Type = BRAND Star Wars Sets
- BRAND Harry Potter Sets
- BRAND Halo Sets
Since this toy company has so many sets, we would target their top selling categories, and then likely utilize a series of DSA campaigns to fill in the gaps, especially by layering on TROAS smart bidding. It would likely focus on categories of products, in which we exclude our existing targeted search keywords so we ensure it’s just picking up new customers.
Search - DSA - Categories
- Star Wars
- Harry Potter
While there is still a bit of planning and organizing, there is no more concern about segmenting by match type in this new thematic way of doing things. Hopefully that gives you insight into how we’re thinking about Search campaigns in 2022.
What About with a Brand Heavy client?
Finally, it’s worth noting that a heavy brand account may want to consider utilizing more exact match, as we find broad match still confuses brand and non-brand too much for our (and our clients’) liking. In other words, if you built out brand campaigns targeting Lego categories (Star Wars legos, Harry Potter legos, etc), you may want to build out a number of Exact match campaigns to target your strong brand, and then utilize Broad and DSA campaigns (with your brand added as negative keywords) to target more into the prospecting terms such as Lego competitors and/or “best toys 2022 christmas”, etc.
EDIT: What About Phrase Match?
Thanks to James Svoboda for asking about clarification on this. He wondered if we only use Broad match at ZATO with Exact and ignore Phrase. We don't, we just utilize all three match types in the same thematic ad group. We see Phrase and Exact close variants as expanding on intent, but still fairly limited in their scope to the terms we thought to include in the account. In this way, Broad and DSA act similarly when used correctly, in that they are truly "prospecting" keyword types since they find audience and searches that are high intent, but we hadn't thought of adding into our account yet. BTW, the only thing I still ponder on this route, is whether it's ideal to actually separate broad into its own separate prospecting campaign in order to give it unique targets, budget, etc... my team has argued me down based on their knowledge of how machine learning uses the keywords within an ad group with smart bidding... but I can see there being a use case for some account with the scenario I described above. Always pondering...
EDIT: Does ZATO Implement Broad Match in every ad group?
Heck no. I realized I wasn't clear on this point as I was focused more on the philosophical side of things. See the previous point on "Phrase Match" for how we see match types interacting with one another. I think there is a strong argument for using only Exact and Phrase (though, grouped in thematic ad groups) for targeted keywords such as Brand terms, specific product SKU keywords, etc. In this case, you're using Exact and Phrase as a narrowed, directional target, whereas Broad and DSA have more leash (albeit, better in intent targeting than they used to be) in being true prospecting campaign types and should be considered accordingly. So in this case, our accounts may look more like:
Search - Top SKUs - Exact|Phrase - US
Search - Brand - Exact|Phrase - US
Search - General Toy Queries - US (all match types)
EDIT: Smaller accounts will need to be more selective in Match Types (this article isn't saying Match Types don't matter, just that the segmenting of them has less impact than previously on account performance).
The great thing about community, is that it helps guide us toward better clarification. There have been some great conversations on Twitter about this article now, and one of the key conversations I'd like to seek clarification over has been on account size (thanks to Amalia Fowler for the great summary of concerns). We definitely do not treat every account the same in how we think about Match Types. If it's a low budget account, we can easily spend that budget on some core Exact/Phrase match terms(though we don't segment by match types on those like I would have in the past) and we will ignore Broad until they're ready to start scaling with prospecting. Even then, we may begin with DSA + audiences + tROAS bidding (if Ecommerce) for smarter prospecting. So, to be clear, I'm not saying we completely ignore match types, or should always use broad in every account. It really does depend, which is why hiring someone who knows the Google Ads system well is so crucial.
I hope that better defines how ZATO thinks about utilizing specific match types... it's just more complex at this point than always splitting all MT out into unique ad groups (i.e., the point of my article).