SEO & Search

Alice AI blocks now appear in 15% of Yandex's product search results

The share of product queries with an Alice AI block in Yandex search results grew from 0.1% in April to 15% in July, Ashmanov and Partners found.

2 min read

Алиса AI активно встраивается в «коммерческий» поиск Яндекса, а коммерция ― в Алису AI
Алиса AI активно встраивается в «коммерческий» поиск Яндекса, а коммерция ― в Алису AIAI-generated

The share of product queries for which Yandex shows an AI block with Alice's “quick answer” has reached 15% — up from one query in a thousand in April 2026. These figures were reported by Ashmanov and Partners, an agency that has been tracking “search with Alice” for almost a year since its launch. In July, the neural block was often placed above all other search results.

What is changing in commercial search results?

Quick AI answers visually compete with links to online stores' product pages, but functionally they do not replace them yet — they serve an informational, supporting purpose. Yandex separates informational and commercial scenarios: the “full” answers in the Alice AI tab already include sections about stores and products, but the company has not introduced commercial elements in quick answers directly in search. This may be due to the fact that Alice's answers in search results directly affect the buying audience and the search engine's monetization.

Quick answers to commercial queries are almost always offline — they are not reassembled in real time and may remain unchanged for weeks, which is maximally economical for Yandex. According to the researchers' forecast, the share of such answers will continue to grow at the expense of high-frequency queries without an unambiguous intent — for example, “Dyson hairdryers” or “a car from China”.

How has source selection become stricter?

After the June update, Alice began using fewer queries when generating an answer: in April, one original query produced roughly four derivative queries, while in July — slightly more than two. The source selection criteria became stricter in 2026:

  • 5 pages per query;
  • more than half of sources — from the top 5;
  • more than 90% — from the top 10.

After the June 25 update, product sections in full answers became noticeably more numerous. Commercial sections now come in two types — “store” sections and “product” sections. Store sections are formed based on the sources selected by Alice (sometimes literally only one or two end up there). Product sections follow a different logic: the source page is merely a starting point for generating queries to Yandex.Market, and product cards are taken from Yandex's service. The store whose page was used as the source gets no advantage: a link from the card may lead to a competitor.

What does the data about shoppers say?

According to a survey by Ashmanov and Partners, online shoppers pay attention to the AI block 30% more often than to regular search results. Some more figures:

  • 69% of respondents agree that AI services speed up the search and study of products;
  • 74% use AI answers in search engines;
  • two thirds of respondents made purchases immediately after interacting with AI: 14% often, 52% sometimes;
  • for 38% of respondents, the purchased item often or almost always matches the neural network's recommendation, and for another 45% a match happens sometimes.

In terms of prevalence, AI services still lag behind marketplaces and search engines, but among their users they lead in four of seven popular scenarios: searching for purchase ideas, comparing products, studying specifications, and finding good prices. A new product search scenario is taking shape, where quick AI answers become an entry point into a dialogue with Alice AI, and competition within the full answer is waged both for a spot on the list of recommended stores and for inclusion in the product section, where prices and ratings play the key role.

Research methodology: data from Alice's answers was collected manually, Yandex and Google search results — automatically; a methodology for statistically assessing the influence of factors on getting into the set of answer sources, developed at the Search Analytics Laboratory, was applied.

Original: t.me

Share this article:

More from Екатерина Ларина

Related articles

« Previous articleNext article »