Google Local Search Patent Application on Ranking Businesses at a Location

I was showing someone Google Local one box results (see H.) for a specific search, and they asked me why Google was showing distances in those results, and where the search engine got the distances from. A good question; my answer was that it was from the centerpoint of a defined geographic region that Google assigned to the businesses listed. And no, I couldn’t tell her where that point was located at.

My answer was based upon a patent application from Google that was published earlier this summer – Indexing documents according to geographical relevance. The process in that document depends upon finding a geographic region related to a local search query, and then returning results within a certain range of the centerpoint of that region.

One Box Local Results vs. Google’s Local Search Results

Interestingly, when I click on the link for Local results for pizza near New York, NY, found on the page that the screenshot above is from, and look at the list on the Google Local page, I no longer see distances in miles, and one of the results has changed.

Why are the results different on the Google Local page than on the Google one box results? Is Google using something else to rank those businesses?

That’s a good question, too. I’m not sure that I would have asked myself that until I came across the following new patent application from Google, which discusses providing results based upon additional factors other than just distance from some geographic center point.

Google’s New Local Patent Application

Scoring local search results based on location prominence
Invented by Brian O’Clair, Daniel Egnor, Lawrence E. Greenfield
US Patent Application 20060271531
Published November 30, 2006
Filed May 27, 2005

Abstract

A system may identify a first document associated with a geographic location within a geographical area and identify a second document associated with a geographic location outside the geographical area. The system may also assign a first score to the first document based on a first scoring function and assign a second score to the second document based on a second scoring function.

The first patent filing I mentioned was submitted to the US Patent and Trademark office in December, 2004. This newer one was filed May, 2005. So it could indicate a change in the way that Google thought about the order of sites listed on the local search page. We’re told in an early section of the patent filing, the Description of Related Art (a section that usually explains what issues a patent filing is intended to address), the following:

When scoring the results, a local search engine may identify a location within the geographical area. This identified location may be associated with the location of city hall, downtown, or a geographic center of the area. The local search engine identifies all business listings and/or web pages within a predetermined radius of the identified location. The local search engine may then identify those business listings and/or web pages that match the search query. The identified business listings and/or web pages are assigned scores according to their distance from the identified location and ranked based on their scores. If the user does not live near the identified location or is not interested in business listings and/or web pages near the identified location, the search results are not meaningful to the user.

So the idea behind this new patent application looks to be to provide results in an order that may more meaningfully reflect a relationship with a specific area instead of a distance from an arbitrary center point.

The Use of Location Prominence

The “location prominence” noted in the title to the document refers to “factors that are intended to convey the “best” documents for the geographical area rather than documents based solely on their distance from a particular location within the geographical area.”

What kind of factors would those be?

The patent describes how it might associate a query with a certain set of geographical locations based upon such things as postal codes, or alternatively it might take the latitude and longitude of the map window when the search takes place while the searcher is looking at a map of the area in question. After that, it might look at factors such as:

  1. A score associated with an authoritative document,
  2. The total number of documents referring to a business associated with the document,
  3. The highest score of documents referring to the business,
  4. The number of documents with reviews of the business, and;
  5. The number of information documents that mention the business.

The patent describes each of those in more detail, including what it might consider an authoritative document associated with a business. It’s possible that may be described in another patent application which came out this summer looking at some ways of determining which web site is the most authoritative for a business. See: Authority Documents for Google’s Local Search.

References and Mentions as Factors

While the first listed factor (location associated with an authoritative document) may involve links, a review, reference or mention of a business with some geographical location information attached to it doesn’t mean that “links” are being looked at:

The number of information documents that mention a business associated with a document may be used as a factor in determining the location prominence score for the document. An information document may refer to a document that provides important information about a business, such as the address, telephone number, and/or hours of operation of the business, reviews and/or atmosphere of the business, whether the business accepts credit cards, etc. Examples of information documents may include Dine.com, Citysearch, and Zagat.com. In one implementation, the total number of information documents mentioning a business may be used as a factor in determining the location prominence score of a document associated with the business.

Distance as a (Continued) Factor?

The document does mention that these factors may be considered in conjunction with a distance score, to “provide a better user experience by presenting the user with documents associated with businesses that are closer together rather than scattered apart.”

The distance score is defined here as being the centerpoint of the postal codes associated with the query, or the midpoint of the latitude and longitude showing in a map area when the search is conducted while the searcher is looking at a map of the region.

Conclusion

Distance, links, and mentions all could contribute to having a business listed first. But what if all of the links pointing to a site are warnings, the reviews are bad ones, and the mentions less than kind? Like pagerank, this method seems unable to distinquish between popularity and infamy.

So a bad restaurant in the center of town that people talk about may be listed higher than an excellent eatery on the outskirts of town that less sites link or refer to. Guess that’s why they let you see those reviews.

Added (12/2/2006): Another Google Local Patent application This Week and Another Local Search Blog

This additional patent application is closely related enough to the one I originally wrote about in this post, that I felt it was more appropriately added to this post instead of a new post. I mention above how a distance score could be defined as “a midpoint of the latitude and longitude showing in a map area when the search is conducted while the searcher is looking at a map of the region.” This patent filing explores using those map boundaries as an area to focus upon when someone is conducting a search with a map in front of them:

Using boundaries associated with a map view for business location searching
Invented by Brian O’Clair
US Patent Application 20060271280
Published November 30, 2006
Filed May 27, 2005

Abstract

A system aggregates entity location information from multiple documents distributed among multiple locations in a network. The system searches the entity location information to identify a first set of entities located within the entirety of a first geographic region selected by a user. The system provides a first digital map to the user via a network, the first digital map including the first geographic region and further including visual representations of the first set of identified entities and their associated geographic locations.

I also wanted to point out the blog of Frank Fuchs, who is EU Product Manager for local search at Yahoo!, and covers issues related to maps and local search in locally type(d) thoughts a local search & maps blog. His latest post provides some thoughtful feedback on this one.

Share

26 thoughts on “Google Local Search Patent Application on Ranking Businesses at a Location”

  1. Interesting Bill: A few months ago I tested this phenomena a bit with different listings w/in G Maps looking at the total number of references to see if that had some bearing on what would be ranked first. At that time I didn’t detect total number of references skewing the G Maps #1 rank toward a particular site. In some cases I saw examples w/ as many as a factor of 8 times the total # of references favoring one site over a second yet the rankings were totally defined by distance.

    Of course the references in G maps seem to be of 2 types, one being local and the second being topical or vertical. Maybe this aspect of the patent relies totally on local type references…and it might rely on a more overwhelming factor than a factor of 8 or it may or may not have been implemented when I was testing this.

    Now here is a wierd situation I found…and with so many different discussions on G Maps I wasn’t sure where to place it…so I thought I’d place it here.

    I pulled a query to my business and site. The business is located in Arlington Virginia. The query was for a “neighborhood” or local description of a part of Arlington–where we are NOT located. The query was …..business service in Crystal City, VA.

    Even as the business is not located in that part of the county, the query brought up our business as #1 in the organic G search results …WITH A MAP from G maps. That was wierd. Just another strange aspect of G Maps applications of serps.

    Dave

  2. Now here is a wierd situation I found…and with so many different discussions on G Maps I wasn’t sure where to place it…so I thought I’d place it here.

    I think I have at least one more post on local in the pipeline for the very near future. I did create a new category (local search and maps) to try to make it easier to find posts and discussions related to local search here. It’s also possible now to subscribe to individual thread comment feeds, with the new design.

    The new patent application does refer to two different types of potential results like you do – business listings and topical/regional results, but doesn’t do much with the distinction.

    I have a followup post, on another new local patent application dealing with business locations and maps, that might address some of the things you raise, and might bring up some new issues. I haven’t drilled down through the patent application involved yet, in a signficant manner. But a quick look is interesting. I’m going to try to write and post that today.

  3. Bill: I just checked and from one search it appears the algo has been updated and reflects the comments from the patent.

    This wasn’t the case from a few months ago. I used a geo term for my business that reflects a town not only closer to a competitor but the competitor is between that town and my location. My site has 9 times the # of references as does the competitor.

    Once entered in G Maps (or before) I’d get tons of references on the web for a site.

  4. Bill:

    In another case I saw G maps ranking my site higher than a site that is significantly closer to the starting location while my business/site has a little less than 1.5 times the # of references than the competing site.

    It appears you don’t have to go hog wild with regard to # of references. But it seems having the most # of references is critical.

    Dave

  5. Dave,

    Thanks for the results of your explorations regarding this patent application. I haven’t done much yet regarding its usage, so it’s great to get some feedback from you.

  6. Bill: Here it is Dec 6th and I’ve been looking at long tail searches for my biz off the logs and found 2 interesting phenomena.

    1. 5 days later and the G Maps rankings are not based on # of references but on distance from a starting point as was being used some months ago. Hmmm. Doesn’t look to me like they experimented very long on this one. LOL

    2. On checking the logs for long tail and specific phrases for my biz — certain organic search results for specific sites w/in the serps were accompanied by a reference to a map with a direction aspect. I assume they pulled the map from G maps. This didn’t affect serps order for my industry.

    It is an interesting plus for businesses/sites with G maps entries (assuming the map is pulled from that source)

    This appears to be further application of G integrating G Maps into G organic search. It provides the specific retailer/business with a little advantage within the serps versus other listings in the top 10 or 20. (Incidentally, the site/biz need not be #1 to have a reference to a map in this instance).

    In early 2005 G instituted changes in its organic serps for local. These changes were outlined in the patent on local serps that G released in summer 2005 and you reported on at that time.

    Those changes, IMHO, created greater logic for local type searches and this little enhancement is an example of further embellishing and emphasizing businesses w/in local search and that probably reflects the importance of G Maps.

    That leads me to a different point that you and I discussed w/in a different venue than this blog.

    While there are many significant advertising and media sources competing for the local advertising dollar I would currently put my bet on the strength of Google and the predominance of usage of google long tail search versus all other sources for local type searches.

    The reference to a map highlights local businesses over other less relevant or less local sites in the serps. Ultimately, I’d bet Google will monetize G Maps. And possibly they should. If it is the primary source for customers it should attract the most advertising dollars.

    Dave

  7. Rather than a total change in algorithms, there is an alternative possibly happening.

    It’s possible that Google may be trying different types of algorithms for different types of services, or testing the algorithms against each other by offering different versions and measuring clickthroughs and other user actions.

    Some nice info on log file analysis results.

    Long tail searches may show the best results for businesses that know how to optimize well for them, but not for businesses that don’t know how. There’ have been some interesting ideas in past patents involving ways to monetize maps – such as possibly paying to be used as a landmark along a route. I guess we’ll see what develops.

  8. Interesting idea on a place to provide observations on the patent applications.

    In my post on Human Friendly Driving Directions, I wrote about Visually-oriented driving directions in digital mapping system. Here’s a snippet from that which talks about advertising through the use of waypoints:

    [0066] In one particular embodiment, businesses are allowed to bid or otherwise pay to be included as a waypoint. The “cost-per-use” of a waypoint could then be an additional factor that would be taken into account when scoring each waypoint. Furthermore, incorporating one or more of the scored waypoints into requested driving directions could include selecting waypoints related to a destination of the requestor. For instance, if the user is asking for driving directions to a national park, then waypoints such as camping equipment stores could be selected for integration into the served driving directions. This choice of waypoints has several benefits:

    (1) the waypoints might be of interest in their own right;

    (2) the waypoints are more likely to be familiar to the user and thus better waypoints; and

    (3) advertisers would be likely to pay more for such targeted waypoint usage.

  9. It would be smart for G to continue to test alternatives. They have the largest data base of actual searches and the largest data base of clickthroughs. On top of that G analytics gives them the largest data base of information on what searchers do inside of websites.

    I suppose we’ll have to watch what they are doing over time.

    It would be great if there was a place where observations on these patent applications was reported. It would help us to narrow in on what G is thinking and how they are proceeding.

    And with your last comment, Bill. Geez, I better be reading even more about patents. I missed that one. LOL

  10. Pingback: Alberto Saavedra 2.0 » Análisis de los factores que influyen en el ranking en buscadores
  11. Pingback: 10 SEO Questions

Comments are closed.