Video Deep Dive: How Google determines authority of local entities
Mike Blumenthal


Deep Dive Into Local series from March 13, 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.


Mike: Welcome to the Deep Dive from Local U for March 13th, 2017. This week we are going to look at how Google determines authority of local entities by looking at authority of these entities on third-party sites and their own websites, plus other ranking factors as we understand them in the world of Google Local today.

Do you want to kick off Mary with your summary of Darren’s talk at Local U and Moz, and about what he found in this year’s search ranking factors?

Mary: The main thing that the latest local search ranking factor that everybody’s talking about is that proximity is now the number one local search ranking factor. I personally feel like it’s a little bit over-weighted. I think a lot of other people feel like it’s a little bit over-weighted, but I know you had some comments about how Google thinks about proximity.

Mike: I think Google’s view of the local algorithm in general hasn’t changed. They have always said, and I believe that they firmly believe, that local ranking is about proximity, relevance and prominence. I don’t think those have changed. I think that from time to time, they dial one or the other up, based on underlying architectural changes, or understanding of user behaviors as we’ve moved into a mobile world, we saw this with Possum. We saw it with Pigeon. We saw it with Hummingbird, where Google was reducing the radius that they were showing to users in their search.

Obviously, proximity has always been a factor. As they understand location of browsers better and as they understand locations of users better, and they understand search queries better, they have reduced the radius that is typically seen on these searches, but it’s just one of a number of factors.

In an experiment I ran a couple years ago, and I would encourage people to do this experiment, I drove through a geography from rural to urban, and I did the same search query. I think it was for storage units. And, as we moved out from rural to urban, about every five miles the results would change, but the result that would be number one might drop to number two, and then if you move another five miles out it would change again, etc. etc. This was immediately after Pigeon.

That will always be the case. I compare ranking to a loaf of bread. Any given location is like one slice, and over space it changes. So, it speaks to the issue of why a business needs to be located in a relevant location in town that serves their customers. Location has always been important and Google tries to reflect that in their algorithm which has gotten more sophisticated.

But, like you said, it’s not the only factor. Relevance and prominence are two critical factors. I’ve done a fair bit of research in Google Maps, and I talked about this at Moz. If you go into Google Maps and you do a portal of the United States, and type in a word like “school shooting,” or news events like “mall shootings,” “lion killers,” strange words like “murdered lawyers,” and look at the results. You will see the results that surface, and the only places these have been written about, these are not link-based results, these have been written about in major news media.

So, if you remember back in 2014, there was a map scandal when someone would put in a racial slur, up would come The White House, similar to that. That wasn’t a mistake. That wasn’t a Google bomb, somebody hacking Google. That’s how Google works. If people write those racial slurs in relation to The White House, Google thinks that’s news and indexes the data. It’s called a triple, indexes that information, associates it with the entity, and brings that into Google Maps as relevant.

Google has expanded their definition of relevance to virtually every unstructured piece of data. Now, relevance though isn’t necessarily enough to, in and of itself, to get it into the front page of Google. From what I’ve seen relevance can come from a category at a third party. I’m certainly doing a bunch of research and analyzing categories at Yelp that aren’t at Google, to see if they generate packs, and 60% of them do. And when I looked across rural midsize and large cities, we saw that 60% of Yelp categories did generate packs at Google.

So Google is clearly looking at those categories from third parties, because Google doesn’t have the categories, right? So relevance can be generated by structured data and citations, or by free form data from news articles. But relevance, in and of itself, might not be enough to move a result from Google into Google search, you also need authority.

I always thought that it was your website that Google was measuring in local. But really, what they’re ranking is the entity in their database.

Mary: Right, and I think that’s what that we all thought, and now we need to totally switch that thinking.

Mike: Right. And you can see this in the API to their knowledge panel, where they actually divulge some ranking information about, not about local entities, but about other knowledge panel entities. Clearly rank is of the entities in their knowledge graph. What I think is happening to move from maps into the main search result, you need some authority.

Authority, in my understanding and based on the tests I’ve done, can come from a number of areas. Reviews at Yelp can generate relevancy, we know that. But, sometimes just a review at Yelp doesn’t seem to bring it onto the front page. What you need is Yelp to focus on that local page, because it’s getting those reviews, and elevate it with the internal link structure that they have. So that page then starts elevating in the best 10 of pages, the top businesses for that area. Ultimately that local page at Yelp, when that surfaces in a Google search result, that authority gets transferred to the local entity in the database, and that allows the entity then to move out from maps into search.

So Yelp is but one of many authority pages. For Google to have trust in an authority page, there has to be confidence that that authority page is about the entity in their database. Sometimes they verify that, like with your website. When you do Google my business, you verify that this home page or this local landing page is the verified page for the entity in the database, so they can have a verified relationship that they trust a great deal. Then the authority of your local page, as it gains links and relevance from other good content, that authority can then transfer to the local entity, just like the Yelp page does. And, we’ve seen the same thing with Google Plus.

Authority on Google Plus, which doesn’t come from just having content or having followers, it comes from having content that is shared amongst followers. That authority from Google Plus can also transfer to the entity that’s in Google’s database, and can help that entity show on a broader range of relevant terms, and be more authoritative.

So, these third party sites that Google can associate with the listing, third party sites including your site. In other words, everything is a third party site in relationship to the entity. And your site is one of the more trusted and authoritative, because you have some control over link building, etc.

So I think the takeaway for me is that links are just one part of the graph of authority. Certainly a current part, and a part that we can have influence over, but it’s only one of the parts. Reviews at every site can confer relevance, but only confer authority if the site that they’re on agrees. So it’s contextual authority at Yelp, and I think what businesses need to do then is find, in their industry, which sites are always showing up, that are likely conferring authority onto the local listing that matches and then over to the entity. So that’s important.

Mary: And that’s going to be your barnacle, you know, your standard barnacle SEO sites, and now they’re giving us clues why, in the knowledge graph itself, showing us two or three other sites that they trust the reviews from.

Mike: Yes, and so another untested theory I have relates to OneBoxes, I’ve looked at a lot of OneBoxes. OneBoxes are interesting, because they are very rare. OneBoxes typically occur when there’s a high degree of relevance between the query and the business, and that typically occurs when there’s an exact match on the business name, right?

Mary: So when you say one pack or One Box, are you talking about a result with one single knowledge graph showing up as the answer, not a local pack with one?

Mike: Yes, exactly. The historical name was one pack, but now it’s called the knowledge graph, knowledge panel. In the past there was a map pin in the map results, and the knowledge panel. Now there’s just the knowledge panel, and the one pack has essentially become that. So it’s just historical naming, but you’re right, it’s the singular knowledge panel. But a singular knowledge panel is essentially one local result, right?

I started studying OneBoxes that showed up on broader queries that didn’t have name matches, and what I found there was that this relevance can be conferred to the entity on things beyond name. You could have a partial name match that would confer relevance. You could have categories that confer relevance, third-party categories, news articles, etc. etc. could confer relevance, and if the relevance of that entity was significantly more than the relevance of the next entity in the line, then you’d get the OneBox.

So in other words, it’s the same algorithm that favors brands, because it so favors business name, and this came out of Hummingbird. But there is a broader algorithm that favors any differential of relevance between two listings, and thus will show the OneBox. And the OneBox is illustrative.

I was looking at the things that triggered OneBoxes. High ranking Yelp page was certainly one of those things. But I’ve seen two other things that have a strong correlation. One is a Wikipedia article on a topic. If you do a strange search, “Nevada Middle School’s shooting,” up comes a one pack for a given local school, and that’s driven by the article relevance. You can see all these newspaper articles, all by the authority of the Wikipedia article. The school doesn’t have a Wikipedia article itself. Only the shooting has an article.

So the knowledge panel’s merged. The local knowledge panel merges with the Wikipedia knowledge graph, and you get this local result, on “Nevada Middle School Shooting,”!

Mary: Wow.

Mike: So now you have the relevance in the articles, and then you have the authority, driven by the Wikipedia page. I’ve also seen it when you get sometimes two, maybe three pages ranking for a business on a broad search query. That can sometimes correlate with the one pack. So, clearly, a high-ranking internal page, which is not an authority page for a business, it is just associated with that business through, say, the knowledge graph, or associated with that business. So it’s not as trusted as the page you verified, but a number of high ranking pages also seem to associate with OneBoxes.

I need somebody to test this, I haven’t tested either one of these things, and it’s hard to get a Wikipedia article for sure. You’re going to have to come back later for conclusions.

There are other authoritative understandings at third-party sites that can confer authority enough to trigger local results. And you see this a lot in the one pack. If anybody has examples of one packs that show on branded searches, send them along to me. I will add them to my bizarre collection of unusual searches, it helps me understand this, and hopefully I’ll be able to write a little bit more about it.

Mike: Now, what I have also determined is that sometimes if you do a query, and you get a OneBox, and then you look into maps for that query, you only see one result. Sometimes when you get a OneBox or a knowledge panel, you’ll see two or three results. I tested a few of those, and I found that I couldn’t make number two appear and convert a knowledge panel, a OneBox, into a TwoBox, or a two pack, right?

So, by increasing the relevance of the second listing when that shows up, you can get enough relevance so that there isn’t a singular result. Again, that demonstrates the idea that relevance is cumulative, authority aggregates and comes from a high ranking page someplace else.

Mary: I know a lot of the old school things we used to just do because we thought they might help are actually proving to have probably had some benefit, like linking to your profiles on Yelp, building links to that.

Mike: Yes, I’ve never tried it. I know other people have, but it’s a bit spammy. Although there are reasons why you might do it legitimately, like from your website to help users see your Yelp reviews. Certainly Yelp encourages you to do that, with badges and stuff. So, the question becomes, “Does it work on every site, or just certain sites?” And does it need more from an internal point of view at a site, like, let’s say, which we are now seeing show up in the reviews from the web more, or Foursquare, which, you know, we hadn’t seen in awhile, showing up in reviews from the web. You know, what could you do to help those? And does it need more than just a link from your site? And how can you do it without spamming? You have the same issues here, and if you abuse it, Google’s going to view abuse the same, whether it’s to your site or somebody else’s site, and, like happened last week, discount low grade links, right?

Mary: Right. And also I think that we need to be thinking again about schema as a way to link everything together that has to do with an entity.

Mike: Correct.

Mary: So that Google can see it all.

Mike: Right, and that’s so Google can understand that every page on your site, for example, or every page on a micro-site, is associated with that entity.

Mary: Right.

Mike: Right, I agree. I think that’s true. And the other thing we need to realize is something David talked about in his article about difference-making local ranking factors for 2020. Google has the ability to understand relevance and authority with no other third party inputs by virtue of the mobile phone revolution. The fact that people search for you, they come visit you, they spend x amount of time with you, they have a transaction with you. Google’s tracking all of that.

They’ve always tracked how many times your business gets searched in maps, for example. We know they’ve tracked that at least since 2008. They used to give you the entity popularity, like base images and maps. This is a function of how often these brands are searched for, that kind of thing.

But now, Google’s ability to understand that is dramatically increased, and I think their ability to understand direct user signals has increased. Which means that in that new world David describes of Google’s understanding of user behaviors, you need to be thinking about traditional advertising perhaps, or other ways of driving people to your brand, digitally and physically, so that Google can see that.

In fact, I even had the idea, why not five businesses, say in the wedding industry, you know, wedding dresses, wedding photography, wedding jewelry etc. get together and do a Android-based treasure hunt, using Foursquare and Google Maps, and you send people around to 15 businesses, and they have to spend and have to check in at those businesses at Foursquare, and save it as a favorite on Google Maps, and you give away some grand prize at the end, then you have a promotion that familiarizes potential wedding couples with all these businesses, by actually visiting the businesses, good idea, right?

I mean, why isn’t that kind of promotion possible? It is, and it could drive these kind of user interactions in a healthy sort of real marketing way that crosses the line between real world and digital, in a way that Google can see that information. Someone needs to do that promotion. Let me know how it goes! It’s a good one, isn’t it? I think it’s a creative idea. Somebody out there can have the idea, and let me know how it works.

Mary: Yes, it is a good one.

Mike: All righty. So, just to reference these articles, Moz, “Proximity to Searcher is the New #1 Local Ranking Factor.” “How Does Google Determine the Authority of Local Entities?” at Street Fight, David Mihm and Mike Blumenthal. My Moz Local 2017 presentation, and then David’s article at Tidings, “Difference-Making Local Ranking Factors of 2020.” We’ll link those in this article as well.

Good talking. We will talk to you later.

Mary: All right. Thanks, Mike. Bye-bye.

Mike: Take care. Bye-bye.

Posted in