The SQI is one of the most discussed—and least transparent—metrics on the Russian-language internet. Website owners track it like an exchange rate, agencies include it in case studies, and Yandex advertising programs use it as an eligibility threshold. Yet there is no official calculation formula: Yandex deliberately keeps the methodology secret to prevent manipulation.
Let’s look at what is known about the SQI from Yandex’s documentation and practical experience, which factors really affect the metric, and how to improve it without questionable tactics—using our own project as an example.
What is the SQI
The SQI (Site Quality Index) is a metric Yandex introduced in 2018 to replace the old TIC (Thematic Citation Index). While TIC was calculated almost entirely based on backlinks, the SQI is a comprehensive, machine-learning-based assessment that takes dozens of signals into account, from traffic to user behavior on the site.
The main difference from TIC is that the SQI is not static. It is recalculated using a continuous stream of data, rather than once a month based on the link graph. As a result, changes to a site—both positive and negative—are reflected in the metric more quickly than before.
You can check your site’s SQI in Yandex Webmaster after verifying ownership, or by using third-party checking services.
Why the SQI matters
- Eligibility for advertising programs. Yandex Direct and the Yandex Advertising Network have SQI thresholds for certain placement formats.
- An indirect trust signal. There is no direct correlation of “higher SQI → higher rankings,” but a high SQI is usually associated with sites that have their technical, behavioral, and commercial factors in good shape—in other words, the same things that matter for ranking.
- A benchmark for agencies and advertisers. It is often used as a quick, straightforward metric when choosing a site for placement or evaluating a competitor.
Which factors really affect the SQI
Yandex officially confirms only the general principle—that it takes “behavioral, technical, and commercial factors” into account—without disclosing their weighting. However, based on practical experience and the data tracked in Yandex Webmaster, we can identify the key groups.
1. Behavioral signals from Yandex Metrica
This is probably the most significant group of factors. Yandex can see visitors’ actual behavior: time on site, pages viewed, return visits, and bounces. Without Metrica installed (or with incorrectly configured goals), Yandex simply has less data to use in its calculations—and that is a disadvantage.
What to check:
- The Metrica counter is installed on every page, including dynamic and programmatically generated pages;
- Goals are configured for key actions (registration, payment, application)—this provides a signal not just about traffic, but about its quality;
- Webvisor is enabled, at least selectively—it can help you see whether pages are “spamming” users with empty content that they close within a second.
2. Technical factors
Loading speed, mobile responsiveness, correct HTTP headers, valid HTML, and working robots.txt and sitemap.xml files are a standard set of factors that are also checked in a technical audit. None of this is specific to Yandex: what is good for Core Web Vitals and Google is also good for the SQI.
Site availability deserves separate attention. If a site goes down periodically or responds slowly, both users and crawlers notice—which means this affects both behavioral metrics and the technical assessment directly. Ongoing uptime monitoring and SSL certificate monitoring cover this area: it’s better to learn about downtime from an alert than from a drop in the metric two weeks later.
3. Commercial factors
For sites that sell products, services, or subscriptions, Yandex separately evaluates their “commercial quality”: whether company details are provided, contact methods are clear, prices are up to date, and terms are transparent. This does not mean you need a physical office, but it is a good idea to provide:
- Contact details that are easy to find on the site (email, messaging app, and, if possible, a phone number);
- The legal entity’s or individual entrepreneur’s details in the site footer or on a separate page;
- An up-to-date page with prices or plans, without “contact us for pricing”;
- A public offer and privacy policy if the site accepts payments.
4. Content factors
Uniqueness, regular updates, and the absence of duplicate and “empty” pages. There is an important nuance that is often overlooked: it’s not the number of pages that matters, but the share of useful pages. A site with 200 substantive articles usually looks better to Yandex than a site with 10 000 pages, 95% of which are near-identical templates with minimal differences.
5. Backlink profile
Backlinks used to be the main factor (TIC); now they are one of many. Quality, relevant backlinks can still have a positive effect, but aggressively buying links from low-quality sites on link marketplaces may have no effect or even cause harm if the profile looks unnatural (a sudden increase in the number of links over a short period is a noticeable pattern).
Case study: how we cleaned up our own site
Recently, we found a classic problem with programmatic SEO pages on reChecker.ru: every site check through our audit tool automatically published a /report/[домен] page that was then indexed. Over several months, more than 10 000 such pages accumulated, and 97% of them had been checked exactly once—that is, someone entered a domain once and never returned. Worse, some of them were “audits” of completely unrelated sites that an auxiliary script had accidentally reached.
To Yandex (as to Google), this looks exactly like a sign of low quality: tens of thousands of near-identical pages with no user engagement, no return traffic, and no behavioral signals. We:
- Disabled the automatic generation of such pages for random domains;
- Excluded from indexing (
noindex) all reports without signs of genuine interest (based on the number of repeat checks for a given domain); - Kept in the sitemap only the pages with genuine engagement—about 260 out of 10 737.
Having a large number of pages is not inherently bad. The problem is when most of them provide no value but are still indexed. The same principle applies to any site: catalogs, tags, and automatically generated landing pages.
What you can actually do to improve the SQI
There is no clear step-by-step guide with guaranteed results—otherwise, everyone would manipulate the metric. But there is a set of actions that consistently have a positive effect:
- Connect and correctly configure Yandex Metrica with goals—without it, Yandex has less data about users’ actual behavior;
- Verify your site in Yandex Webmaster and monitor the diagnostics section—it shows indexing issues before they affect the metric;
- Conduct a technical audit and fix critical issues: broken links, missing HTTPS, and slow loading;
- Check whether the site has programmatically generated pages with low uniqueness and no engagement, and exclude them from indexing if it does;
- Provide the company’s full contact details and registration information in a prominent place on the site;
- Publish content regularly—once every one or two weeks is enough; consistency matters more than frequency;
- Monitor site availability to avoid losing behavioral metrics due to outages;
- Don’t buy links in bulk or all at once—build the backlink profile gradually using relevant sources.
How long to wait for results
The SQI does not update instantly: changes in user behavior and the site’s technical condition need to accumulate before the algorithm takes them into account. In practice, noticeable changes appear within several weeks of systematic improvements, not the next day. This is normal: the metric is calculated in a way that prevents it from being changed drastically by a one-off action—only sustained improvements to the site itself can do that.
If you need to track how a site’s technical and behavioral metrics change over time, you can set up a weekly SEO report. It shows trends in key metrics without requiring you to run an audit manually each time.