Search engines like Google or Bing or Yahoo often respond to searches with lists of web pages or links to videos or pictures from websites. When Google responds to searcher queries, it may give results that are lists of URLs based upon how relevant web pages it might point to might be to a query submitted by a searcher, and those results may be ranked based upon certain measures of quality of those resources, and the search results may provide links to those resources as well.
Those “measures of quality” often involve things such as PageRank, a metric that has been calculated based upon the number of links pointed to a page and the quality of the pages that are doing that linking. Google is likely looking at other measures of quality as well, and a patent granted to Google this week gives us a look at another measure of quality – how location might play a role in whether or not some pages are returned in response to a query.
That patent is:
Distance based search ranking demotion
Invented by: Neha Arora
Assigned to Google
US Patent 9,262,541
Granted February 16, 2016
Filed: October 18, 2013
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for processing local search results. In one aspect, a method determines a first local document in a set of local documents for search results is eligible for a demotion operation.
The determination is based at least in part on a proximity measure based on the location of the user device and a location specified for the first local document.
In response to the determination, for each other local documents in the set of local documents, a proximity measure location of the user device and the location specified for the other local document is determined. The first local document is demoted in its ranking when at least one of the proximity measures indicates a respective other local document has a specified location that is within a threshold distance of the location of the user device.
When it comes to local searches, the major search engines do provide “maps” results in response to queries that have an intent based upon location associated with them. Search for “Pizza” around lunch time, and chances are that you are looking for a place to buy a pie or at least a couple of slices. A search engine showing you nearby pizzerias makes a good deal of sense. In addition to organic web results, Google might also show you Google Maps results that are definitely based upon location, and aimed at addressing a situational need that you likely have.
But in addition to those Maps results that have been inserted into the organic Web results, location may influence the other results that are shown. When I type “Amusement Parks” into a search box, the results I see seem to start off with geographically closer amusement parks, based in San Diego, and then move north to sites that start mentioning places in Orange County. I later see the website for Knott’s Berry Farm, but not for Disneyland. I see the page for Universal Studios, too.
But then, I see Universal Orlando, in Florida, and Disneyworld, also in Florida. I wonder if those are pages that just rank well for the term “Amusement Parks”. They could, based upon relevance, or information retrieval scores, and high quality scores involving metrics such as PageRank. Why no Disneyland in the organic search results? I’ll have to see if I can gauge how well the SEO appears to be done on that site. I’m wondering if it was demoted from these search results pursuant to something like this patent. Why would the Florida Amusement parks show up highly in my search results (top 20)?
Local Significance and Boosted Results
The patent tells us that some websites have a “local significance” to them. Based upon the search engine understanding my search device’s (computer or phone) location, and the location of those websites, if they are within a certain threshold distance apart, those sites might be considered locally significant, and the patent tells us that they may be boosted in search results.
So some sites show up highly in search results based upon relevance and quality and whether or not they are boosted because of location.
Demoting Unboosted High Ranking sites
On a search for a query such as “coffee shop,” we may see local search results for web pages for coffee shops that are nearby. Someone searching for a coffee shop likely is looking for a close place to get coffee and possibly some food to go with it.
But, the patent tells us that some local results, may have very high scores independent of a local scoring boost, and this is referred to as a “location independent score.” and could be high because the website may be of interest to people outside of the region that includes the location of the local result. I mentioned not seeing Disneyland on my search for “Amusement Parks”, and it didn’t show up. I expected it to, because it is unique and really quite famous. It is possible that a search algorithm that may take into account search traffic and query log data may rank a document for the amusement park (e.g., a web page) very high. It’s quite possible that Disneyland has a high Location independent score, because of that popularity.
The patent provides another example that illustrates how it may work to limit rankings for sites that tend to rank well for location independent scores for searchers likely looking for queries that have high location intents behind them:
By way of another example, a power company named after Thomas Edison may have very heavy local traffic in a large state. Thus, the search algorithm that takes into account search traffic and query log data will also rank a document for the power company very high due to the sheer volume or traffic to the webpage by its many customers. However, the power company webpage may be of very little interest to a user several states away. For example, a user may desire to learn about Thomas Edison, and not the power company that bears his name. However, due to the large volume of traffic to the power company web page, that web page may be ranked higher than a webpage about the man, the latter of which is more likely to be of interest to a user that inputs the query “Edison.”
The patent tells us that the searcher multiple states away, who might be served pages about the power company named after Edison may have the power company demoted in rankings of the pages received by him or her because of the distance. That makes sense – there likely wasn’t a local intent to see a page for the power company on the part of that searcher.
Is this approach described in the patent the reason why I wasn’t seeing search results while looking for “Amusement Parks” that included Disneyland?
I do have to state that a carousel shows up at the start of these results that include Disneyland within them. The patent is silent on carousels.
Some queries have local intents behind them, and some sites fulfill those local intents and may be boosted in rankings in search results to meet the situational needs of searchers potentially interested in visiting them. Some sites that may fulfill local intents may also rank highly based upon location independent factors, such as being well known and visited by lots of people. Those pages ranking well because of the location independent factors may be demoted in search results (and the patent includes a formula that could determine how much they might be demoted in rankings).