Matt Cutts Describes How Google Uses Human Quality Raters
IX Brand SEO Services Company serves small businesses throughout the Houston area and around the US through offices in Sugar Land and Wharton, TX. We focus on local search engine optimization, online reputation management for collecting good customer reviews, and business process improvement for using customer feedback information. Please give us a call at 979-531-8300 or send an email using the contact form.
Searchengineland.com featured an article about a recent Google video describing its use of human raters. Matt Cutts, the head of Google’s Webspam Team did the video. It’s great information and it illustrates one aspect of Google’s continuing efforts to improve the quality of its search results — from the perspective of a human. Here’s the video:
The question the video is intended to answer is how Google uses human raters as a part of its algorithm. In other words, how does Google use people paid by Google as a part of its algorithm. Here are the main points of the response:
- Google does not use human raters to influence its algorithm in a direct way. In other words, a human rater paid by Google does not influence your website’s placement in search rankings.
- Google does use human raters to test, validate and verify its algorithm.
- Google’s hundreds of raters have classified a large number of websites as good, not good, webspam, etc.
- Google engineers use that bank of evaluated domains to test changes to its algorithm. In other words, if a test modification to the algorithm results in higher quality domains being placed higher in results, then that’s a good thing and the algorithm change might be considered for additional study and implementation.
- The next step in evaluating an algorithm change might be the use of side-by-side testing. A human rater is presented with two sets of search results, side-by-side, for a particular query and then is asked to pick the better set of results, providing comments about the reasoning for the choice. One side would likely be the results from the current Google algorithm and the other side would be the search results derived from the modified algorithm being tested. The human rater doesn’t know which side is which — derived from the current algorithm or the test algorithm.
- Next Google’s engineers will look at the search results rated much better or much worse by the human testers. If the testers didn’t see much of a difference between the quality of one side and the other, then the test algorithm wouldn’t be implemented. Matt Cutts says that Google is looking for “outlier” differences — in other words very significant improvements from use of the test algorithm.
- If the new algorithm is rated a significant improvement by the human testers, then a small live test is conducted to see if people in the real world click and interact more with the search results delivered by the test algorithm. If so, then it’s a good thing!
- The human testers sometimes have difficulty distinguishing spam from good websites, so Matt says the human testers are not a substitute for the expertise and intuition of Google’s engineers. However, the human testers can have a significant impact on the testing of algorithm changes and they help Google identify the algorithm changes that are a big improvement or at least an improvement in the validity and relevance of search results.
- Bottom line — the human raters don’t have a direct impact on search engine results, but they do impact the testing and validation of algorithm changes.
If you have questions about this concept or you’d like to discuss your search marketing situation, please contact me using the web form or call me at 281-343-3284. IX Brand SEO Services Company serves small businesses with search engine optimization and online reputation management help in the Houston area including Fort Bend and Harris County. Thanks!
What are you doing to get your website higher Google rankings?