Video Deep Dive: Categories and Relevance in Google Local Search

Mike Blumenthal

Mike Blumenthal & Joel Headley take a Deep Dive into categories and relevance in Google local search.

This is our Deep Dive Into Local from August 28, 2017. In our Deep Dive series, we take a closer look at one thing in local that caught our attention and deserves a longer discussion.

If you have a special topic you would like us to discuss for the Deep Dive in Local, please reach out to us. If you would like to be on one or the other of our segments, reach out and send us the topic and your availability.

If you are interested in sponsoring this weekly show also please let us know.

Our weekly discussions are also now available as a podcast as well. SUBSCRIBE TO THE PODCAST HERE.

Mike: Hi, welcome to Local U’s Deep Dive. This week we’re going to be covering categories and relevance in Google local search with my guest, Joel Headley, who previous to his current job as head of local search at PatientPop, worked in Google Local, created their support teams, was there for many years and is doing some interesting research at PatientPop, and recently wrote an interesting article about how categories work at Google. So why don’t we start by you just summarizing that article both specifically terms in of medical, but also generalize it.

Joel: The article is really kind of rudimentary in terms of an introduction to the topic because that blog is really for SMB directly providers and speaking to them. Not quite this audience. But in general, Google is trying to get all the relevant signals it can…relevancy signals it can, right? And categories is one of those signals. But it can also pick up information from other places whether it be attributes or other sites. In fact, I remember when Google Local first launched and Bret Taylor, who launched the product, was excited about explaining how you could find “best sushi restaurant” in San Francisco. And the key was being able to find out “best” and “sushi” when back in the day “restaurant” or “Japanese restaurant” were the only categories available in a lot of the yellow page directories online. And the way they did that is through crawling and picking up citations that had additional relevancy signals. So that’s kind of the whole game that Google is trying to play was how can we incorporate more content from other sources into a local listing to turn that result to give you the best stuff.

Mike: So in the age of entities, it’s sort of … and in categories it’s sort of a hierarchical view of … and I’m talking about the article is primary category, and then perhaps attributes or modifiers or secondary categories of information that might return a local search. In the article you talk about how Google understands hierarchy particularly in the medical field, doctor and then specialist, maybe you could just talk about that a little bit.

Joel: Yes. So categories are — for a long time Google didn’t do its own categories. They just use categories provided by third parties and then —

Mike: As I recall, it was Superpages was the initial category provider, if I recall it correctly — like, 2008.

Joel: Yes. I know…

Mike: I once did this experiment where somebody wasn’t showing up under “window repair” and all these other people were and there was no “window repair” in Google, right? This was in 2008. And when I removed their Google category, which is maybe “construction,” they then showed up back in “window repair,” which was their category at Superpages. And so it was clear that they were getting category information some place. And at this point their own category information was overriding —

Joel: Overriding. Yes.

Mike: — third-party category information which is like … well, that’s not very good. They sort of fixed that, right? But talk about the hierarchy of categories.

Joel: Yes. So, hierarchy. So for a long time like I said, Google didn’t really have this structured in the same way. Categories were being added from multiple sources. At one point they had a project and said, “We want like a canonical list of categories.” And in that canonical list they built in a hierarchy. So you have a root category that is usually something like “establishment,” right? And then it grows from there, it goes from “establishment” to “restaurant” to “Mexican restaurant.” And so if as you choose categories when you think about why Google is asking for the most specific category first, because if you say you’re a Mexican restaurant, Google knows that you’re also a restaurant and also an also, you know, a POI type of establishment.

Mike: So why did the user interface never reflect this hierarchical view? In other words, why not force users instead of them saying, “I’m a restaurant, I’m a Mexican restaurant,” all both, why not force users to pick Mexican restaurant in the interface?

Joel: Yes. I think because they are still restaurants, right? So it’s not…

Mike: Right. I’m not saying you shouldn’t be able to select the restaurant, I’m just saying that if you selected a sub category, you wouldn’t be able to still select the primary category, the other.

Joel: Oh, the primary category.

Mike: Right.

Joel: I think it’s just simplicity and I think…

Mike: Because the interface ends up confusing people, right? If you’re a lawyer, but you’re also a personal injury and a criminal lawyer, you’re going to pick lawyer, personal injury and criminal. Right? You’re not going to pick criminal or personal injury as your primary one even though what you’re saying is you should, because Google knows you’re a lawyer. You’re wasting the law category.

Joel: So when you think about a lawyer or personal injury lawyer, if you choose the most specific category that’s what Google is going to put the most weight on. So you want to do that particularly for the category that probably brings in you the most money, right? So if you want to be primarily getting most of your business from Google as a personal injury lawyer, you’ll want to choose that above, like, workplace accidents or whatever is your secondary.

Mike: Criminal defense, whatever. So.

Joel: Yes. Criminal defense.

Mike: So that raises the question what is the relative difference in terms of Google’s thinking about a primary versus the secondary category? In Bing, you can actually weight them, right? But in Google you’re saying, number one is primary and all others are secondary.

Joel: Yes, yes. And I think even the UI calls them additional categories.

Mike: Right. And upon display when you search, the primary category is the one that shows.

Joel: Yes. And it’s the only one that users can edit. So it’s not just how it’s resulting in search, but it’s also kind of like the meta description. It’s kind of a click through thing, too, right?

Mike: Right.

Joel: If you want to be… that’s how you’re going to be marketed online through Google.

Mike: So if you had to guess on a scale, on a logarithmic scale of 0 to 10, what value a primary category gives you versus a secondary category? On this scale, where would you put primary category versus secondary category?

Joel: Eight and…

Mike: And secondary category is two?

Joel: No, that’s high.

Mike: Another thing… they both contribute to relevance, right? But one is more important.

Joel: Yes. And I think as a relevance signal they’re actually very close. But as a branding and a click-through signal, they’re much further apart.

Mike: I see. So you look at it in two ways which is conversion optimization versus Google understanding the business.

Joel: But you see in more competitive categories where that difference does make a difference. So, if you’re selecting dentist first versus cosmetic dentist, and people are really competing at the cosmetic dentist that different — it is a big difference where you’re choosing to compete in that space as a general dentist versus a cosmetic dentist. So depending on the competition, I think it can make a big difference where you end up being returned for different searches.

Mike: So let me step back a little and explain my understanding of the algorithm as I sort of articulated it at Moz the past. It appears that category is a cumulative value or … you know, semantic understanding. It’s broader than category. I get that, but let’s use category as the reference point. It’s a cumulative value at Google, right? So like, I found that a third-party category will allow business to show on Google Maps, but not in Google Search in and of itself. But a Google category might be enough to show a Google search, or even if you had a third-party category plus review content. In other words, it was clear that Google was looking at review content for a signal of relevance, would it be enough? Obviously, the strongest category signal that we see is business name. When a business name matches the query it becomes what call a brand search. But my thinking has evolved to the idea that brand search is just the most relevant search versus all these other searches which are becoming longer tail and less relevant. That it’s a single relevant sort of view of the world. So first I ask you is that sort of general understanding broadly, you know — I’m not asking how the algorithm worked, I’m asking if the understanding is broadly correct and helping me understand how categories work at Google.

Joel: Yes. I think ultimately all these things do lead into relevancy, right, for a listing. So categories, titles, reviews, content on web pages that act as citations, all that information needs to be fed and …

Mike: But before we go on, we want to point out that we’ve had this discussion about descriptions on citations being an unlikely, untrusted source for relevancy, right?

Joel: Yes, yes.

Mike: That category information which is driven by the owner of this site is much more trusted. And which brings me to my other… well, let’s finish your point now before I bring to my current research.

Joel: No. I think what what you brought up there in terms of depending on where the content is sourced from, I remember when Google had its own version of a description field and everybody…

Mike: Or it’s own version of attributes. Remember before when you could do your own attributes?

Joel: Yes, yes. And everybody would ask me at Local U, “What should I be putting in there? How do I optimize it?” And without saying … Google doesn’t index it, so there’s no time worth spending optimizing it. Like, I would say things like, “You want it to be very readable so if people read it, if a human saw it, they would understand and care to click on your business.” And a lot of people didn’t necessarily take that advice, but what I was trying to say is like, Google doesn’t use it at all for relevancy signals. And I think that is played out, you know, Google kind of drop that certainly from Google My Business. It’s still kind of there in Plus, but…

Mike: Although it’s really hidden in Plus, it’s hard to find it.

Joel: Yes, yes. That’s right. I said “kind of.” But a lot of things are becoming structured data now. So, answer this question in a very specific way. I guess that’s different on Q&A and it’ll be interesting to see how that works.

Mike: Right. But Q&A is going to be much more unstructured than the current attribute query that you can either add attributes via or edit them. It doesn’t let you create your own attributes. I mean, they’ve obviously established a fixed number of predetermined attributes that they then query in any given industry which get added to your listing, you can either accept or reject them or whatever, right?

Joel: Yes, yes.

Mike: So attributes clearly contributing. I have found just in my research and this was last fall, that Google Plus, because of it’s verified and trusted relationship with GMB, is that if you got a fair bit of sharing in Google Plus posts, that that content — structured by the idea of collections, but unstructured by the idea of content — seemed to increase relevancy quite a bit. And then, you know, we had a conversation, gosh, it must be four years ago now, three years ago? We were sitting in Seattle and we were sitting around the table, and the question came up is “how does Google know that a particular bar has shuffleboard or not when it’s not a category at Google?” And you said, “We know, we know.” And you called up a Seattle search for local bar with shuffleboard and up came three to which everybody me…

Joel: Eight.

Mike: Eight?!

Joel: We found… Yes. Eight got returned and we called every one of them. And every single one of them except one — because they didn’t answer the phone — said they had shuffleboard.

Mike: Right. So it’s clear that Google was reaching beyond their limited category set. One other recent test that I did was I took, with Darren Shaw at Whitespark, we looked at 10 categories that were at Yelp that weren’t at Google and scraped across 50 markets, you know, did category plus city searches across 50 markets, 10 categories. So whatever that works out to… 500 searches and 85% of those searches showed pack results, right, from Yelp category. So clearly, they are looking elsewhere for categorical information. You know, to me the question is how do you… when we’re dealing with unstructured and we first saw Google using blogs and citation sources… a number of years ago. I think Carter Maslan was still there, right?

Joel: Yes.

Mike: This was 2009 or 2010. How do you…I mean, this is just really good algorithms, parsing free-form content and understanding… well, it seems in the case of Google Plus that they understand both the content, but also how much it’s being shared. Or in the case of Yelp, they’re looking at some sort of prominence signal to indicate how important this relevance signal is. Do you think that’s, in fact, the case?

Joel: Yes. I think always when you’re thinking about relevance, you have to understand a little bit about how prominent the signals are giving for that relevance, right? So Google knows for example, how many times a certain category has been mentioned across citations. So is the category of personal injury lawyer, is that happening… we have 300 citations for this single business and across…is it happening across five of them or is it happening across 270 of them? So Google does see that voting behavior, right? When you’re thinking about links being votes, that voting behavior for content as well. So it’s…

Mike: Which explains why citations have always been a critical component of a local search campaign. I mean basically table stakes at this point, but still some value in terms of degree of relevance, let’s put it that way.

Joel: Yes, and you can imagine influencing this as you create a blog on your site, right? How can you increase the amount of content that’s linked to a certain category or concept that your business provides so that you can continue to increase the number of signals, so that relevancy signal also becomes more prominent for your business.

Mike: Right. That’s interesting. Is there anything else to add or you think we should call it a wrap at that logical conclusion?

Joel: Yes. I think that’s great.

Mike: All righty. Well, with that I wanted to thank you and your beard both for joining us today. And I hope to see you next week on Deep Dive. Thanks again.

Joel: Thank you.