Yana Yantres
Yana Yantres
Led end-to-end design of a new AI-driven tool at Semrush. Conducted research and 20+ user tests to understand and build the product from 0 to 2,000 active users.
About Semrush
Semrush is a leading SEO and digital marketing platform helping businesses improve their online visibility.
Context
Semrush has powerful data, but getting to an answer takes too many steps. Users were getting lost before they got value.
AI raised the bar.
Marketers started expecting faster answers, less manual work, more automation. The tools they used daily were changing. Semrush needed to change with them.
The goal
Make complex SEO tasks simpler and faster, without adding more complexity to the platform.
My Role
I led the project from discovery through launch, driving research, product strategy, MVP definition, UX design, validation, and delivery.
Discovery
Round 1
Alpha testing, 3 initial workflows. Mixed results, but the pattern was more valuable than the outcomes.
EEAT Assessment worked. It gave users genuinely new data they couldn't get anywhere in Semrush, instant value, zero friction.
Keywords and Backlinks Opportunities didn't. Both reassembled data that already existed in Semrush tools. Experienced users saw no meaningful difference.
The finding that changed everything:
"Users are interested in the outcome, not in the process of how we generate it. These features might be a nice-to-have — unless the solution guarantees new actionable data."
Alpha research summary
Round 2
New workflows testing
10 in-depth interviews • 40 min each • 3 SEO specialists + 7 non-SEO users
9 out of 10 users accepted AI-powered, non-SEO workflows as a natural fit for Semrush, validating the product's potential beyond the core SEO audience.
The finding that changed everything:
Research insight
AI workflows earn user trust not by using AI, but by showing something the user couldn't see before. This became the filter for every subsequent design decision.
MVP Definition
Discovering uncovered areas of automation opportunities.
To prioritize, we assessed workflow frequency of use, skill requirements, and time cost. This process revealed three focused workflows.
The landing page presents a library of predefined SEO jobs.
Each workflow clearly communicates:
Users wanted to understand what happens after clicking Run Workflow.
To make the process transparent, the setup shows:
The goal is not to show more information, but to help users make progress.
Design Principles
If users could find it elsewhere in Semrush, we rethought it.
Users always asked "what do I do now?" We stopped linking out. We put the answer next to the data.
Test concepts before building them, not to spend a lot of time.
The product validated AI workflow automation as a scalable direction for Semrush.
Without a single marketing email, launch announcement, or in-product push:
As a result, the product evolved around guided workflows designed to reduce manual work, simplify research, and help users move from analysis to action faster.
Not every workflow generated meaningful adoption.
Some concepts that seemed promising showed limited demand after launch, reinforcing an important principle:
The challenge is not building more workflows. It's building the right workflows.
As new workflows were introduced, users had more reasons to return to the product and incorporate it into their day to day work.
The success of the initial release led to continued investment in new workflows, some of which later achieved retention rates above 80%.
The learnings from this project also influenced future AI initiatives across the Semrush ecosystem.
