Grocery app redesign
I led the redesign of a grocery app with 5M+ monthly users. We changed it in parts over three years, alongside regular feature releases.
Perekrestok
A large grocery chain in Russia, part of X5 Group. The app handles delivery and pickup for 5M+ people a month.
What needed to change
The app had too much promo content, mixed styles and no clear hierarchy. There were no shared design principles. Perekrestok is a higher-end grocery brand, but the app didn’t feel like its stores.
Millions of people were used to how the app worked, so we couldn’t change things lightly. The architecture limited what developers could build, and work on the basics competed with new features for the same developers. We often updated single components rather than whole screens, so old and new patterns were in the app at the same time.
The principles I set
Put products first
Products and their quality get the most attention. Less room for interface elements, promos and decoration.
Calm as the main emotion
The interface shouldn’t rush, distract or overload people. It should help them get their shopping done.
One accent colour
Only key flows and main buttons get the accent colour. The rest is neutral, so products and key actions stand out.
More space, fewer styles
More air, fewer colours and type styles, so screens are easier to read. We chose colours and fonts with accessibility in mind.
Own graphics, no emojis
Our own icons and illustrations look neater and more premium than emojis, and fit a higher-end brand.
Simple flows
As few steps as possible to finish a task. But before removing a step, check what people already rely on.
I based the direction on product analysis, research, metrics, the brand guidelines and talks with product managers. The changes shipped at different times, so I used the principles in the team’s design reviews to keep them consistent.
A cleaner home screen, with the QR code still there
We left the loyalty QR code on the home screen at full size, although it was one of its biggest blocks. People were used to opening the app and showing the home screen at the checkout counter, and an extra tap would have slowed them down while a cashier and other shoppers were waiting.
We rebuilt the hierarchy around that habit, removed heavy blocks and made the main tasks easier to find, which also left room for later experiments.
More categories, without a much longer catalog
The top-level categories were too broad. People couldn’t always tell where a product belonged. Promos pulled attention away from navigation, and delivery terms and pickup were easy to miss.
We moved second-level categories to the first screen and grouped them. With about 10,000 products, that could have made the page very long. Compact tiles let us show more of the structure without making people scroll through a large grid.
We also made order modes and delivery info easier to see and cut down the promo blocks.
+9%
Category visits
+5%
Add-to-cart conversion
A product page that helps people decide
We looked into why people opened the product page and what they wanted to find out. The old page had a lot of secondary information, and the add-to-cart button was hard to notice.
We removed blocks people didn’t need, organised the useful information, made the main button easier to see and added recommendations. The rating block now shows a useful review. We also updated badges, ratings and reviews to match the new visual style.
+5.5%
Add-to-cart conversion, A/B test
+11.5%
Visits to product reviews
What the independent audit showed. In April 2025, an independent agency compared the main grocery delivery apps.
1st
In emotional UX
2nd
In rational UX
Users gave high marks to the visual quality, the navigation and the familiar placement of features. These were the areas we had worked on in the redesign.
The audit also showed gaps, such as limited personalisation. It doesn’t separate the effect of the redesign from everything else the team did.
What I would change. I’d bring in developers earlier to understand the architecture limits before choosing a solution. I’d also join backlog planning earlier, instead of pushing for improvements after the priorities were already set.
Next case
-33% order cancellations
Four experiments in an app with 5M+ monthly users. One cut cancellations due to forgotten items by 33%, and we rolled one back.