Last week, we found an interesting bit of research that determined it possible to project a couple’s chances at building a successful romance by looking at the Facebook friend profiles of the two involved in the relationship. The studyby Lars Backstrom of Facebook and Jon Kleinberg of Cornell University concluded that the more diverse the two sets of friend groups was from one another, the better the chance of the relationship lasting beyond a two-month period.
This conclusion runs against the commonly held misunderstanding that the more friends two people have in common, the better their chances at romantic “success.” A two-month time frame might not qualify as a successful long term relationship, but it is very significant that the relationship among friend groups may be a signal about the couple’s chances of going into the next change of seasons with the same romantic interest.
The roots of this study are also significant. The practitioners used billions of records and huge sets of data to reach their conclusions. It is in looking at how these conclusions were arrived at that provides some insight into search engine algorithms – particularly in the latest “Hummingbird” update.
The recent Google algorithm update, commonly referred to as “Hummingbird” arrived with little fanfare outside of SEO circles late last summer. Although the immediate impact of the change was minor – impacting a fairly low percentage of existing queries, the future impact could be much more significant.
Good Quality Results
The one thing all search engines compete for is of course, users. The only real competitive point then is which search engine has the best quality results for each query. Google dominates search owing to its ability to understand and deliver the most relevant results for each user search. The more relevant the results, the more likely the user is to utilize that search engine for subsequent search activity.
How do Search Engines Tell if they Got it Right?
This is where it gets interesting. The search engines understand which links and results they present to each search request. They can also monitor user behavior once the reader leaves the search engine results page and opens a page referred by the search engine. If the reader opens a referred page and then hits the return button to try a different result, the search engines capture this bit of behavior and make the judgment that their referral may not have been what the reader actually wanted. If on the other hand the reader stays on the page, consumes the content and perhaps clicks deeper into the site for more information, the search engines judge this to be a successful referral as validated by the reader.
Like the Facebook and relationship study, these analysis tools require understanding huge volumes of user history and data. However, the ability to analyze large data sets is much more possible than ever before, using powerful data sorting systems and technologies.
In order to continue providing the highest quality search results, Google launched Hummingbird to provide better and more relevant results to questions and queries that users are now submitting. Google has seen growth in the types of search requests that use longer, more conversational questions. Understanding what the user really wants is key to delivering great results. Distinguishing between what the user asks for and what the user wants is a tricky game, but Google has determined that it is one worth winning.
Longer, more conversational queries are not necessarily the only types of results that the search engine giant is trying to handle better. Readers using search terms like “animal rescue information” might also find results valuable that include “dog rescue” or “cat rescue information.” Returning the dog and cat oriented rescue websites might actually be more relevant than a result for a company specializing in getting squirrels out of chimneys. A user asking for animal rescue results may be looking for cat or dog rescue sites.
Using history and readership statistics to determine that a “cat” is actually an “animal” when used in the context of “animal rescue” seems to be a main thrust of the Hummingbird update. Understanding the actual meaning of the user query is key to delivering the best quality results.
In light of the revelations about Facebook and successful romantic relationships, it is possible to gather huge amounts of information and build models that can reasonably predict value or behavior and to understand much more about what search engine users really want.