How Google Enforces Category Diversity In Local Search Results

Google is currently enforcing local search result diversity through a category system, which utilises a mathematical function to scale initial SERPs.

This function resembles a logarithmic scale, where the more of a single category results there are, the less likely content of that type is to rank.

Search Results Diversity in 2013

Matt Cutts first introduced “host clustering,” where under each subdomain or domain, it was only possible to have 2 search results. Prior to this, pages would simply be scored according to the algorithm, which could result in the whole first and second SERP consisting of only a single domain.

Webmaster’s got around this enforcement in 2013 by creating many different subdomains in order to obtain more “SERP space.” As each subdomain was limited to two search results, this would still allow for domains to own entire SERPs.

After the webmaster diversity abuse in 2013, Matt Cutts explains:

A pages on a Search Engine Results Page, originating from a single host name, would be logarithmically scored.

This makes it harder and harder for additional pages to rank under a single domain.

What this means, is that it would still be possible to “abuse” the subdomain system to some extent, however each progressive page would find it more and more challenging to rank. This allows for a form of forced diversity.

However, this ranking adjustment only really applied to the first page. It was imperative to Google to ensure diversity on the first page, so that queries had a diverse, ample range of resources to read from. The second SERP, as Matt Cutts explains, is less impacted by this mathematical change, where it is still possible to have much less diversity.

I hadn’t seen anything quite like that merger between organic results and a local result happen again after that. It is impossible to tell if Google has been using that kind of merging since then. But that patent was all about providing more diverse search results to searchers

Bill Slawski

Search Results Diversity in 2019

The current Google patent “Enforcing Category Diversity” was approved in May, and likely has been commissioned by now (late September). This patent builds upon Google’s landmark image SEO system.

The key here is that the results diversity is dependant on the category of information being presented. Typically, I would assume this meant image-based SERPs, video-based SERPs, or traditional link based SERPs, but this actually goes beyond that.

Google has employed a method of diversifying the type of information presented, being that if you were to google “things to do in Madrid, Spain,” you could potentially receive only a list of bicycle routes. While you might very much enjoy cycling, it doesn’t cover all the bases in terms of search diversity.

I believe that this enforcement has not only come from link adjustments, but also the changes in the SERP Presentation. Google has recently introduced a “top things to do in ${city}.”

“When a searcher asks for points of interest information at a certain location, the local search system may generate a collection of candidate POIs and receives information relating to each candidate POI’s respective category and a score and rank within the category for each, and, for categories a searcher may select, promotes or demotes the score of each ranked candidate POI within its respective category through a scaling process.”

US 20170061025 A1

The 2019 System

The diagram to the left is the simplest explanation of the new patent system.

Effectively, upon query, Google identifies key POIs and ranks them as it usually would. It then multiplies each of the POIs according to a “category scale,” which influences the final ranking of the POI.

It then finalises the list, and presents the SERP to the client.

This is an extract directly from the patent description:

  1. Selecting, as the one or more categories, one or more categories that are each associated with more than a predetermined number of candidate POIs the predetermined number is two
  2. The method includes selecting, as the one or more categories, one or more categories that are each associated with one or more candidate POI
  3. Scaling, for each of the one or more categories that are associated with only one candidate POI, the non-scaled score of the ranked candidate POI associated with the category comprises multiplying the non-scaled score of the ranked candidate POI associated with the category by a factor of one
  4. Scaling the non-scaled scores of the ranked, candidate POIs includes increasing the respective non-scaled scores of the top n ranked candidate POIs
  5. Scaling the non-scaled scores of the ranked, candidate POIs includes leaving unchanged the non-scaled scores of one or more of the top n ranked candidate POIs
  6. Scaling the non-scaled scores of the ranked, candidate POIs includes decreasing the non-scaled scores of one or more of the top n ranked candidate POIs
  7. Dynamically determining a scaling factor to use to scale one or more non-scaled scores of the ranked, candidate POIs of a particular category based on a non-scaled score associated with a top ranked candidate POI of a different category; and/or the method includes dynamically determining a scaling factor to use to scale one or more non-scaled scores of the ranked, candidate POIs of a particular category based on a quantity of the candidate POIs of the particular category identified in the data.

This surmises to be a very complex system that deals with local search diversity.

Key Takeaways

Search diversity is putting emphasis on local search results. When searches are made by client machines, the first figure taken is the location of the search, and as is evident above, results in a weighting score when the initial SERP is generated.

It’s critical, therefore, for SEOs to stay on top of local search results and ensure that correct structured markup is being used. Developing Castle Site profiles, a local backlink profile, and relevant anchor text are also critical in improving local search results.

Furthermore, due to the mathematical implication of this algorithm, being in a highly-saturated market in terms of a brick-and-mortar business can result in you falling under the fold when considering the logarthmic-esque function. Bringing on category diversity changes the link driven diversity we’re used to seeing. As a result only the best of a specific category are displayed. If you are a restaurant, you are scaled against all other restaurants and down-weighted based on the volume of restaurants. This makes it far more challenging for an individual site to rank for local POI queries.

Leave a Reply