Sites. Yandex puts an estimate for the behavior

More recently, a wave of rumors and speculation about the influence of behavioral factors on the ranking of search engine results flashed across the Internet.

Sadovsky vaguely hints that they say we use all available information to improve the quality of the search. But in general, from the statements of the “Yandex” themselves, it clearly does not follow that the behavior of users is indeed taken into account when generating search results.

Well-known researchers of algorithms unequivocally claim that they affect. There are services such as "user", created with the aim of cheating behavioral factors.
But still it is not clear what, how and does it really affect.
')
Maybe this is just a flash mob created by Yandex to whip up webmasters?


Why does Yandex need it? Obviously:

• Improve search
• Reduce the impact and the possibility of existing cheating by optimizers
• Reduce staff costs (assessors) with the help of users
• Making the webmasters and optimizers finally pay attention to improving the site’s user experience and learn how to work with visitors, websites and content, and not manipulate the issue

Where can the data come from:

• Yandex Advertising Network
• Yandex.Direct
• Yandex.Metrica
• Yandex Bar
• Open statistics LiveInternet
• Actually, the issue
• Webmaster.Yandeks.ru

What factors can be considered:

• Site depth
• Time spent on the site (document)
• The bounce rate (in different cases is treated differently, usually the percentage of visitors who immediately left the site)
• Exit points
• Voting click (the visitor went to the site and never returned to the search)
• Snippet CTR
• Returns to the site
• Compliance with sites-standards in the subject and the region with good performance on behavioral factors.
• Achieving the goals specified in Yandex.Metrica site owners

Factor accounting problems

The most important is the dual nature of the interpretation of behavioral factors:

• The depth of visits may indicate that the visitor likes the site, and that you figure it out, so you have to go further;
• The time spent on the site can be quite small if a visitor came, for example, to a page with a description of the product, looked at the characteristics, price, and went to see the best offers (but the return factor from the search to the site may work if the visitor did not find a better offer )
• The bounce rate on some topics is very high. In the "iron" and industrial subjects - from 40-50%. On medical sites it reaches 80-85%. And this is from really convenient and friendly sites. How to deal with refusals for one-page sites?
• CTR is no longer an indicator of the success of advertising campaigns. Younger people open up a bunch of tabs at once and then consistently close them in search of what they want. Those who are older, first study the search results and only then go to the sites. On the other hand, a site that is behind the 20th position in a highly competitive subject has little chance of getting a click on a request. And only if the search engine compares its behavioral characteristics with the TOP-10 sites and finds them similar, it will have a chance to rise higher.
• Targets marked in the Metric by the site owner may not be set correctly.

The metric is not everywhere (roughly - in 2010 there are 4 million domains on the RuNet according to the RU-Center and only 7.7% of sites with the metrics according to the Interlabs).
How to take into account unknown factors on other sites? Perhaps a comparison of known values ​​with standards will help here. It turns out that of the whole list of factors (maybe I'm wrong?), Only voting clicks and the percentage of returns to the site can be taken into account?

Besides all this, there are other questions:

• Different subjects are characterized by different types of behavior, because completely different people work in them.
• Different behavior depending on the type of requests (transactional, informational, navigation), sites (individual documents) and even the time of day and time of year. The more complex the request, the less visitor satisfaction with the sites in the issue.
• Different interpretation of the meaning of the query by the search engine and the user
• Different issue of different people depending on the region, browser, resolution ...
• How can behavioral factors be related to each other?

What's next?

Someone seriously engaged in usability audit of their clients, not knowing what makes the site just some kind of "average" form, because we also need a marketing audit, which involves working with text and graphic materials on the site.

Someone is trying to find patterns and putting experiments on 1-2 sites, making wrong conclusions.

Someone refuses Metrics altogether, forgetting that there are plenty of other sources of data.

It will be great if Yandex will have the opportunity to let the users themselves vote for the sites directly in the issue and form their TOP by moving resources to the right places and deleting the “leftist”. Although it also will not save from cheating.

There is a feeling that in recent years, Yandex has been casting the bait for webmasters about using various promising ways to improve search and new factors, and then follows the discussion of their influence among webmasters and develops the most successful thoughts.

What else to see

Sergei Protasov's report (Rambler) at the conference “Search Engine Optimization and Website Promotion on the Internet 2009” November 26-27, 2009.