When I began leading the search & discovery product team at Yandex Market*, I focused on defining our e-commerce search strategy. Data from Yandex Search indicated a user shift: they were moving from traditional search engines like Google to marketplaces when researching or buying products. Our past approach to search had been linear – optimising machine learning algorithms to provide the best list of items for specific queries, with a goal to maximise orders. I believed that in order to revolutionise our search experience, we needed to delve deeper into understanding the intent and context behind each user's query. It isn't just about the query itself and its results, but rather about understanding the user's end goal or the "job to be done."


Here we classified the higher-level questions users ask our search engine. Since we didn't aim for a comprehensive list with an ideal classification, we focused on discerning the primary scenarios, enabling us to map out the key steps users take during their purchase decisions.

Users' decision-making steps might not always manifest explicitly in their search queries. However, we needed to understand the thought processes during the purchasing journey to provide a more helpful, context-based search experience.

Intent types are the key part of this document. We classified 5 key intents that might be traced in all search queries. I've provided some examples here to visualise each case. It's important to note that a search query doesn't always equate to user intent. Nevertheless, by analyzing the search query, we could make a guess what the intent is. Subsequently, our team created an ML algorithm that helped us classify search queries into intent types.

When discussing intent-based search, I used examples from various product categories to make it easier to communicate the vision to teammates and stakeholders.

Once we identified key intents and pain points users are facing, our product discovery team generated an initial set of solutions that served as a basis of hypotheses to test.

Some more examples to help the team be aligned.

This strategy led to numerous experiments and A/B tests that aren't typical for search experiences. By understanding user context and intent, our search and discovery team added tens of millions of additional annual GMV (Gross Merchandise Volume) to the company's growth.

Vazgen Badalyan

This is a user intent diagram that served as the foundation for our AI/ML product strategy.

User Intents for Product Strategy in Yandex Market Search

Sign up to see more