APTEKA.RU WEBSITE
Russia's #1 online pharmacy by order volume, Data Insight Top-100, 60M+ orders/year. Built the team's UX analytics practice from scratch, turning every interface decision into a measurable one. This enabled a redesign that lifted conversion by 5% and improved perceived appeal.
Scope
Design lead · UX analytics system owner / Design & visual direction / A/B testing strategy & perception benchmarking
Client
Apteka.ru
2026
Context
Apteka.ru's core business priority has always been assortment depth and delivery speed, and for years, that meant internal tooling took the lion's share of the team's attention. It was the right strategic call. But it also meant the website quietly accumulated debt: visual, structural, and experiential. By the time we looked up, the gap between where we were and where the market had moved was significant.
What made this moment different from simply knowing the site looked dated was that we had data to substantiate it, because several years earlier I had built a UX analytics system that gave us continuous insight into how users perceived and navigated the product. Perception surveys, support request monitoring, in-product surveys, our own A/B infrastructure, all of it running in the background, building an evidence base over time. When it came to making the case for change, we were not working from intuition alone.
The surveys showed "modern" and "beautiful" were underperforming, "beautiful" sat at just 14.10% of user responses. A homepage survey revealed that 64.7% of users identified advertising banners as the first thing they noticed on arrival, not the search bar or product content.
Navigation
That moment came from an external audit that flagged: "make the search bar more prominent." A modest recommendation. I used it as an opening to propose something considerably more substantial, a structured rethink of navigation architecture and visual language across the site. Rather than moving immediately to visual changes, we started with structure. We tested 3–5 different navigation concepts and ran corridor tests with real users to evaluate each one.
The final solution was closer to the existing structure than the more radical concepts we explored, but with a set of targeted changes that made a meaningful difference. The side catalog was moved to a horizontal navigation strip directly below the header, making product categories significantly more accessible. Information about the loyalty programme, promotions, and discounts was brought into the header and given visually distinct bright icons, which improved their findability considerably. Pharmacy storage time was added to the header as well, addressing a question our support request monitoring had flagged repeatedly as a recurring source of user confusion. And a small but effective detail: the current city display and the link to nearby pharmacies were merged into a single inviting phrase, "we'll deliver today to any of X pharmacies in [city]", turning a functional element into something that actually communicated value.
Visual direction
With navigation resolved, we turned to the visual language. The process began with a market study of Russian e-commerce, tracking what category leaders were doing, where visual trends were heading, and what was beginning to feel dated. We iterated through 3–5 directions before arriving at a concept built on three interlocking decisions: a move from flat to pseudo-3D iconography to introduce volume and tactility; an expansion of the palette beyond the signature blues, adding greens drawn from the Katren brand identity; and a richer, more layered gradient system that gave the product a sense of depth it had not previously had.
Results
We launched via A/B test across four variants, old and new design, each with and without the navigation addition, across approximately 10 million sessions. The new design performed best: conversion at 16.98% versus 16.15% in the control group, a gain of +5.1%. Registration rate moved from 0.90% to 0.95%. Average order value remained stable throughout.
We then ran the perception survey again using the same questions as the original benchmark, something that was only possible because that baseline had been established in advance. "Beautiful" grew from 14.10% to 16.69%, a gain of +2.59%. "Outdated" declined.
The successful update to navigation and visual language opened the next phase of work: moving from the ad-hoc page layouts that had accumulated over years to a consistent 4-column grid system across the site. The product page, historically one of the most structurally complex and commercially critical pages in the product, has been fully redesigned and is currently in development.
On the role of the analytics infrastructure
It is worth noting that none of this work was possible in isolation from the measurement system we had built. The perception benchmark existed because I had established the survey infrastructure and begun running it before any specific project required it. The navigation insight came from support request monitoring we had built and instrumented ourselves, enriched with platform data to distinguish web from mobile issues. The A/B test ran on our own internal platform. The project is a direct illustration of what becomes available when measurement infrastructure is treated as a design asset, by the time there were decisions to make, the data to make them well already existed.