At the Optimization-2021 online conference, Mikhail Slivinsky, head of the webmaster service at Yandex, spoke about metrics that help improve the quality of search.
One of such metrics is Proxima, which includes various aspects: commercial, audience loyalty, and others.
The Proxima metric takes into account different aspects of quality. In particular, it includes the aspect of expertise and trust, which Yandex initially took into account when ranking sites in the field of medicine, finance, and legal services.
These areas are difficult to understand, and it is important for a search engine to provide answers for such queries that users can trust.
Then the aspects of expertise and trust were applied in the field of navigation queries.
In the future, the search engine will begin to take into account expertise and trust when ranking other topics.
Other signals that are important for ranking and that Yandex began to consider recently are signals of value for the user . Sites that are of no value to users are at risk.
You can check if a site has value by answering the following questions:
- What is the expertise of the author/service provider?
- Do users distinguish this site from many others?
- Would this site have appeared if there were no search engines?
In addition, Yandex began to take into account “antiquality” signals. A few weeks ago, webmasters noticed a decrease in the rankings of their sites and received a comment from Yandex support that the site was associated with fraudulent actions.
This is the result of running an algorithm that is included in the ranking technology. Yandex lowers the position of sites that users may consider unscrupulous.
Mikhail also talked about the search quality metric “Surplus” . The idea of the metric has built-in logic to solve the user’s problem.
The surplus characterizes the success of solving the problem by the user during the search and is calculated as the sum of the results of interactions with the elements of the search results.
Successful interactions increase the surplus, unsuccessful ones decrease (and the more, the more attractive the element is).
According to Mikhail, the new metric is needed, since Proxima is insensitive to the size of elements, it is about individual elements and not a general assessment of user interaction.
A key assessment can be made through the Surplus metric.