Google announced they will be sunsetting broad match modifier and changing phrase match to work like the old broad match modifier. These changes should be rolled out by July 2021, and then, you will no longer be able to create broad match modifier terms.
That means your old phrase match terms will expand to show for more search terms than they have previously. If you have used a lot of phrase match, you’ll run into the same issues that people who have used a lot of broad match modifier and exact match have run into:
– Duplicate search terms
– Irrelevant search terms
Let’s take a look at what these issues are and why you want to pay attention to them.
If you have used modified broad match and haven’t examined these items, you should read on to see why this analysis should be an important part of your PPC workflow.
Not enough people talk about duplicate search terms, and they can severely hamper your account performance. A duplicate search term is when Google is showing the exact same search term from multiple ad groups. This can hurt your ad testing and bidding as you don’t have a consolidated view of how this search term is performing in your account.
As an example, in this account, they have two search terms that have been shown from multiple ad groups very consistently over time. In the second expanded example, when the search term is shown from ad group 1, the conversion rate is 18%. When the search term is shown from ad group 2, the conversion rate is 32%. When the search term is displayed from ad group 3, the conversion rate is 20%.
You can click on any image to see it in more detail.
By simply adding an exact match negative keyword to the 2 ad groups with lower conversion rates, this account will see a nice increase in total conversions. As much as Google wants to automate your ad serving, if you aren’t finding these poor ad serving areas and fixing them, you’ll lose quite a few conversions.
With this change, we can expect to see more accounts run to duplicate search term issues as more keywords can match to the same search terms across various ad groups.
If you haven’t used N-grams, this primer will walk you through what they are. Essentially, they examine common 1, 2, and 3 word patterns across search terms.
As Google will expand how your phrase match terms will show, n-grams will help you get insight into that data. This change hasn’t happened yet, so as an example, we’ll show how much Google is expanding exact match at the moment.
This exact match term is a 3-word term, which is a product name plus the word quote. There are over 100 search terms that are 5+ words and 50 that are 8+ words that are matching to this exact match 3 word term. Nowhere in this ad group is a question word (who, what, how, etc). When we look at the search terms, most are question terms that have added several words to an exact match term.
We might initially look at this and think Google is matching based upon us talking about quotes, and therefore the question terms are OK to show for if they involve asking about cost since we mentioned a quote. That’s why we want to use n-grams to see if that’s a true statement or not.
The account overall has a CPA in the $450 range. If the word ‘how’ is in the query, our CPA is over double. If someone searches for ‘quote,’ our CPA is under $300. If they ask a question about cost, the CPA is over $1000. Obviously, looking for quotes versus cost data is very different in the eyes of a user, but not in the matching search terms of Google.
This line of thought can lead us to check other question words, such as search terms that contain the term ‘what,’ and see how our data looks:
We still see highly elevated CPAs.
As a point of reference, here’s what good n-grams look like in the account:
That tells us we need to either:
Add all question terms as negative keywords. This will reduce cost and conversions as these terms do lead to conversions, just overly expensive ones. If we are trying to cut costs, adding these as negatives is a good course of action.
Add all question terms as keywords with lower bids. If we want to hit our target CPA, then we wouldn’t want to stop showing for these terms; we’d want to control the cost of the terms. This is easy if you are using manual bidding.
These campaigns are using Google’s automated CPA bidding. CPA bidding does not seem to take into account these obvious trends, and it looks like we need to add these terms to different ad groups with lower CPAs and try to manually work around Google’s CPA bidding since Google’s CPA bidding is not accounting for the difference in user conversion rates of quotes versus cost.
The ideas behind some of Google’s match type changes are good ones. If the search term is the exact same intent as your keyword, then it would make sense for Google to match those terms to your keywords. Unfortunately, in practice, Google can decipher when meanings are similar, but they can’t figure out when the intent is similar or different.
Cost is different than quotes. Deals are different than limited sales. Trademarks are different from copyrights. 2 fo1 tickets are different than 2 for 1. Wholesale is different from bulk sales.
Unfortunately, Google hasn’t figured out many of these subtleties. Therefore, that is still your job – ensure that your search terms match the intent of your keywords, and when they do, ensure the proper ad and landing page are used with that search term.
The newest changes will make that a bit more difficult.
These types of analysis can be much easier with software that just automates it all for you. If you’re looking for software that will do all of this analysis (and a lot more) for you, then please take a look at what Adalysis can do for your Google Ads account.
Find out, in minutes, what your account score is with our FREE audit report.
Great article as ever, Brad!
With matching search terms, a few questions:
1) Wouldn’t Google automatically favour the ad group with the higher conversion rate in almost every auction?
2) Do you think this matching search terms issue means that keyword lists in campaign builds should be much smaller and that each keyword should be completely distinct to the other?