Press enter to see results or esc to cancel.

Understanding Google My Business & Local Search

What is Location Prominence?

click for a slideshow on Location Prominence

In the local space a number of us have reported on and studied factors that affect ranking in Google Maps. One factor that has been difficult for us to qualify, much less quantify, is that of “Location Prominence.”
 
In the organic world all are familiar with the concept of PageRank, a reiterative view of the web that ranks a website based on the strength of the links coming into a site. Google wanted to fundamentally change how local had been done in the 2004-5 timeframe and looked to create a similar system to evaluate and rank business listings in the Maps index. Whereas PageRank is a heuristic number used to rank websites, in the system that Google engineers came up with, they termed the analogous collection of Local data points the “Location Prominence Score” to rank local business listings.
 
In some ways it is perhaps a more robust deployment of the PageRank concept into the geo spatial world of business listings. It seems to embrace all that PageRank was intended to be, and added a significant component of geo references to the mix. PageRank scores commonly available today represent a general indication of the strength of website. The Location Prominence Score, if we could see its value, of a business listing would probably provide a similar insight as it relates to that listing.

For those of you that have not read the patent here is a salient summary of the concept from the conclusion of Google’s patent filing on Scoring local search results based on location prominence. Here is a relevant description of Location Prominence from the patent filing:

Further, it has been described that a location prominence score may be generated based on a set of factors that includes one or more of the following factors: a score associated with an authoritative document, the total number of documents referring to a business associated with the document, the highest score of documents referring to the business, the number of documents with reviews of the business, and the number of information documents that mention the business. In another implementation, the set of factors may include additional or different factors.

[0073] For example, one factor may relate to the numeric scores of the reviews (e.g., how many stars or thumbs up/down). Another factor might relate to some function (e.g., an average) of all the scores of the reviews. Yet another factor might relate to the type of document containing the review (e.g., a restaurant blog, Zagat.com, Citysearch, or Michelin). A further factor might relate to the types of language used in the reviews (e.g., noisy, friendly, dirty, best). Another factor might be derived from user logs, such as what businesses users frequently click on to get detailed information and/or for what businesses they obtain driving directions. Yet another factor might relate to financial data about the businesses, such as the annual revenue associated with the business and/or how many employees the business has. Another factor might relate to the number of years the business has been around or how long the business has been in the various listings. Yet other factors will be apparent to  one skilled in the art. 

[0074] It may also be possible to use the factors to train a model using machine learning techniques. The model may be used, for example, to determine the probability that a user might select a particular document in the search results.

This patent filing is from 2004 and to the best of my knowledge has not yet been approved. Bill Slawski has discussed this patent in a fair bit of detail. What follows is a specific look at the ranking factors that Google considered in the patent. Much could have changed in the way that Google currently calculates rank but in this summary you can find many of the attributes that were identified in our research last summer and in David Mihm’s Local Search Ranking survey.

Here is a breakout of the set of factors that are prioritzed in the Patent:

Google’s Patent Language My Comments
a score associated with an authoritative document, This refers to the document that Google determines is the authoritative website for the business
the total number of documents referring to a business associated with the document, References as noted here are likely to be links, possibly containing geo references.
the highest score of documents referring to the business, The reiterative thinking of PR applied to the the looser content of the geo web
the number of documents with reviews of the business, and The total number of reviews have been shown to correlate with ranking
the number of information documents that mention the business These are the “citations” that are do not include links but enough geo information to be associated with the business listing in Maps.

You can see how Google has expanded its reiterative PageRank thinking to include non-website-based factors from across the geographic web. Local Search indices evolved when very few small businesses had websites, making it essential to come up with an alternative to PageRank that could use the information that was available at the time.

Google, however, didn’t limit themselves to these five above factors. The patent assumes that geo information across the web will change over time but not in a lock-step fashion across all industries. There is the implicit assumption that available information about any given business will become richer over time. Low-signal industries might rely on the simpler geographic web information while high-signal industries that are very competitive might move closer to the traditional link-based web ranking model, with important geo enhancements.

Google Patent My Comments
one factor may relate to the numeric scores of the reviews of all the scores of the reviews This has not shown up in any research to date,
the type of document containing the review (e.g., a restaurant blog, Zagat.com, Citysearch, or Michelin) We know that a lot of citations come from industry-specific sites
A further factor might relate to the types of language used in the reviews (e.g., noisy, friendly, dirty, best). I think this may be acutally being used in determing relevance.
Another factor might be derived from user logs, such as what businesses users frequently click on to get detailed information and/or for what businesses they obtain driving directions. Clearly whether Google continues to show an Authoritative One Pack, Three Pack and 10-pack seems to be predicated on this concept
Yet another factor might relate to financial data about the businesses, such as the annual revenue associated with the business and/or how many employees the business has.  
Another factor might relate to the number of years the business has been around or how long the business has been in the various listings.  
It may also be possible to use the factors to train a model using machine learning techniques. The model may be used, for example, to determine the probability that a user might select a particular document in the search results.  

There was general agreement in David Mihm’s Local Search Ranking Factors survey as to the need to be located in the geographic area searched and to have appropriate categories related to the search query. Those elements allow the business listing to be considered relevant to a particular local query. It appears that this concept of Location Prominence Score when combined with the relevance factors form the core principle behind Google’s ranking.