Google recently changed how phrase and modified broad match types work. Here are the new definitions:
Phrase match: Matches of the phrase (or close variations of the phrase) with additional words before or after. Close variations include terms with the same meaning.
Modified broad: All the terms designated with a + sign (or close variations of those terms) in any order. Close variations include terms with the same meaning. Additional words may appear before, after, or between the terms.
There are two main issues we see with this change: new queries and the same queries matching to multiple keywords (loss of control).
Showing for new queries isn’t necessarily a bad thing unless the new queries are unrelated to what you sell.
For instance, we see quite a few matches where the words might be somewhat similar, but the user intent is totally different.
Keyword (modified broad matched) | Query | Result |
copyright logo | Copyright symbol | -100% conversion rate. Google is changing logo for symbol across huge amounts of keywords. |
register trademark | Registration symbol | -100% conversion rate |
Patagonia tours (or trips, vacations, etc) | Patagonia Chili | -100% conversion rate. Google is showing many vacation keywords for location queries with no intent to travel. |
Keyword research | Keyword analysis | Research and analysis are not the same things. -100% conversion rate. |
Discount Branson show tickets | reserve Branson | -100% conversion rate. |
44030001 | Weber 2 burner grill | Google ignored the keyword in the account |
We can go on and on here, but what we wanted to know is how many more queries are keywords showing for than the same time last year.
We were approached by a client who was seeing a considerable downturn in conversions, and they wanted some help. Since their decline happened just after the match type changes, we wanted to look at how their phrase and modified broad match keywords were doing compared to the same time last year. So we looked at how many unique queries their top keywords showed for in August of 2018 vs. August of 2019.
What we saw was that some keywords showed for fewer queries; however, some were showing for a significantly higher amount. For instance, they had one keyword that matched to 1314 queries in 2018 and an astonishing 2745 search terms in 2019.
Across the account, some keywords had shown for quite a few fewer queries, but their issues were stemming from a few keywords that expanded what they were matching to, and most of the new matches were not converting.
The second problem they had is that their total unique query growth had not changed that much. What was happening is that Google was triggering new ad groups for the same queries than they had in the past, and this was one of the significant problems with their account.
Their issue is related to most account’s second problem, Google matching the wrong term.
Let’s say I have these keywords, each one in its own ad group:
You might expect a search term like register trade mark to show from the ad group that includes the exact match [register trademark] as it’s the closest match. You would probably expect the search term trademark registration cost to match to the ad group with the keyword “trademark registration” since this match would meet the old definition of phrase match, which is more restrictive than the new definition.
After the change, the keyword that received the most traffic for those search terms was the ad group with the modified broad match, +trademarking. This ad group uses a different landing page since the term is not as specific and a completely different ad.
Where it gets more frustrating is there were times Google would not show the ad group that has the exact match keyword in it and instead would show a lower performing ad group.
While Google laid out some rules for how the closely matching keywords should show, there are so many exceptions to this rule that it might as well not exist.
When the same query is showing from multiple ad groups, this causes what we call Duplicate Queries. Essentially, the same query is showing from multiple ad groups. When this happens, here are some of the account issues that arise:
This has become a huge issue. We saw a travel account where 1,078 of their total 4,035 unique queries were coming from multiple ad groups. In fact, 28% of their impressions have come from queries that have shown for 2 or more ad groups.
In Adalysis, we look at how many ad groups a query comes from as well as the cannibalization rate of other ad groups to the primary ad group for that query. When duplicates get this high, you often see many fewer conversions. I’m pretty impressed that with a total of 1591 impressions, Google managed to find 15 different ad groups to show an ad from within a single campaign.
There’s a nice workflow you can use to determine what search terms become keywords:
The main problem is this isn’t the workflow most people use. Most people look at queries that are converting and add that query to that specific ad group. If you do that, while Google is showing the same query from multiple ad groups, then you end up with duplicate keywords. That means you have added the same keyword to various ad groups, and now Google can legitimately serve either one of those ad groups for the same set of queries.
This leads to the same bidding issues, ad testing issues, and so forth that duplicate queries have.
For instance, in this account, there are over 1,100 duplicate keywords. For this keyword, when it is displayed from ad group 1, it has an 8.6% conversion rate. From ad group 2, it doesn’t have a conversion.
There are a few steps you can take to correct these issues.
Step 1: Monitor Duplicate Search Terms
You can use a script or a system like Adalysis to monitor these issues automatically.
When this issue arises, you want to add the exact match negative keyword to the group. If you are using scripts, you’ll have to do this by hand. If you are using a system like Adalysis, you have many options to add these in bulk to fix the ad serving issues.
Step 2: Monitor Duplicate Keywords
You can use a script or a system like Adalysis to monitor these issues automatically.
When these issues arise, you want to pause the worst preforming duplicates. You can do this by hand in Google’s interface or use bulk options in 3rd party systems, like Adalysis.
Bonus Step: Monitor Keyword Conflicts
The last issue that arises as you add queries as keywords and exact match negatives is that you end up with keyword conflicts. Those are keywords that are not showing because of your negative keywords. While negative keywords are essential to any account, Google doesn’t inform you about negatives that you added due to organization (making one ad group show instead of another one) or any conflicts in negative keyword lists.
Therefore, use scripts or systems like Adalysis to monitor these conflicts (we’ve never seen an account with at least two negative keyword lists that didn’t have a conflict) so that you are showing ads for search terms that do properly match your keywords.
Odds are with the match type changes; you are showing for brand new queries. Also, you probably have some search terms showing from multiple ad groups. It’s essential to monitor and analyze these changes to make sure you don’t see significant downturns in performance or end up with accounts where your bidding, ad testing, and query management isn’t working properly due to how search terms are now matching to your keywords.
As always, great article Brad.
Part of the problem is that Google does not share a lot of information in the search terms report.
I have never seen more than 30% of the impressions on a search terms report.
To make matters worst, search terms with zero CTR are not reported, limiting our ability to add negative keywords and improve the CTR.
Thanks for the excellent breakdown, but i’d bet the majority of agencies are not capable of carrying this level of technical sophistication into PPC matching analysis, in response to these changes which are almost exclusively in Google’s favor. The problem is that Google feels little competitive pressure, so they simply turning the dial in the direction that takes those pesky humans out of the ad management decision making, while increasing profits to themselves. If Google were more responsive to advertisers and agencies, they would not continue to push the matching control in this direction (looser). If there were some actual competition in the PPC ad market, Google would once again need to care what advertisers actually want, and the PPC market would flourish. Likely Google, as well.
Hi – My account has mostly phrase match keywords. Since the roll-out of the 2019 close variant change Google is regularly showing my ads for a single word picked out of a phrase match. For example, if I advertise “rs232 network converter” the algorithm decides to show my ad for “rs232”. Googles statement to me is “the algorithm is making an assumption about intent and deciding that each searcher implies the additional words whether they include them or not.” The cost implications are huge. I am now adding exact match negatives to my account but I am curious – are other people seeing this?
Really good explanations and examples. I am trying to find a way to explain this to my AMs .