Yandex shows businesses how to get into Alice's answers and measure search visibility

You cannot buy your way into Alice AI's answers or "game" them with a single trick. The main takeaway for businesses is much simpler: Yandex's assistant relies on regular search, so sites with strong SEO, a clear structure and accurate company data win. The channel is already too big to ignore: according to the author of the source material, 46.5 million people see Alice's quick answers in Yandex Search every month.

The key change came on April 7, 2026, when Yandex opened a new section in Webmaster, "Site visibility in Alice AI." Until then, companies mostly guessed whether their site appeared in generative answers; now they have working analytics. The section shows Share of Voice, the share of queries in which Alice mentioned the site, along with concrete example phrases and the sites you compete with for a place in the answer. The data covers the last three months and is updated weekly.

One important caveat: the statistics are built only from queries for which the site already ranks high enough in classic search results. If there is little data, the problem is usually not "Alice" but basic SEO.

Yandex's logic has also become clearer. After the "Alice, what's new?" conference on October 28, 2025, it became clear that the assistant increasingly works not as a simple link aggregator but as an editor: it searches on its own, cross-checks facts and assembles a detailed answer from different sources. So priority goes not to abstract sales pages but to material that already contains ready-made building blocks for an answer: a direct explanation on the first screen, tables, lists, comparisons, FAQs, product cards, and descriptions of services and organizations. If a page is full of general promises and short on specifics, Alice literally has nothing to extract.

For local businesses, data consistency matters separately: name, address, phone number, opening hours, categories and description must match across the website, Yandex Business, maps and directories.

The practical formula for businesses looks like this: build pages around users' real questions, not around a generic description of the service. Yandex advises following the EPOS framework: expertise, usefulness, originality and substance. In practice, the text must show real experience, answer practical questions, give prices, timelines, selection scenarios and limitations, and not just repeat keywords.

For online stores, product cards alone are especially insufficient. They also need an "explanatory layer": strong category pages, buying guides, model comparisons, reviews with details of use, and up-to-date availability and price data.

Additional signals matter too: link the site to Yandex Business, connect Metrica, submit product data, use Schema.org and keep reviews fresh. If a business opened a branch, changed delivery terms or launched a new service, it is better to publish such updates on the website rather than leave them only on social media.

For the market, this means something simple but unwelcome to many: the era of AI search has not abolished SEO but raised the bar for page quality. Ranking at the top helps, but it is not enough; the site must also be a convenient source for a generative answer. So companies should stop looking for a "secret optimization for Alice" and instead measure their visibility share in Webmaster, find queries where they are not cited as a source, and rewrite pages to include short answers, verifiable facts and current data. Businesses have no paid way to buy this kind of organic inclusion. Such sites are the ones that get a chance to become part of the new search layer.