Introduction
Yandex Eats' desktop site hadn't changed since 2018, some users mistook it for a phishing page. This case covers a year-long, ground-up redesign journey that closed that gap and became the template the team reused when launching in new markets.
Role
Product Designer
Sole designer on the project. Owned the redesign end to end: from audit and research through UI, design system, and developer handoff. Worked closely with a product manager and a development team
Team
Product Designer
Product Manager
Development Team
Design Lead (review)
Timeline
May 2022 — April 2023
Context
Business
Yandex Eats operates in one of the largest food-tech markets in the CIS — Yandex's food delivery business grew GMV by roughly 50% year-over-year by late 2023, and served as a major acquisition channel for Yandex's broader e-commerce ecosystem, generating around 15% of its new customers.
Problem
In the mobile-first era, the desktop faced the set of problems that have become a modern classic: random typography, an outdated color palette, components nobody had touched in years. At the same time, the rest of Yandex's ecosystem, including the Yandex Eats app, had moved on. In interviews, several users said the outdated look made them briefly wonder if the site actually belonged to Yandex — one specifically said it looked like it could be a phishing page. For a food delivery product, that's not just a look problem: it's a trust problem right before someone enters their payment details.
Recognizing this gap, I saw an opportunity to rebuild trust in the desktop experience — and bring it in line with everything else in Yandex's ecosystem
Constraints
The redesign had to stay within Yandex's existing brand system (typography, color, patterns), and address a growing pile of technical debt — inconsistent, hardcoded UI patterns that had accumulated for years without a shared desktop design system.
Work Scope
I owned the end-to-end redesign of the desktop experience: from discovery and interaction design to UI, design system, and developer handoff. It covered: homepage, restaurant and store pages (including product cards), checkout, tracking and order history, an international landing page, the address/map selection modal, search, and the site footer.
Research
With limited resources allocated to a lower-priority platform, I focused on a small set of moderated user interviews and a full UX audit of the existing desktop experience.
Watching people actually place an order revealed far more than a survey ever could.
Painpoints
No confidence stepping in
A significant part of users mentioned weak trust signals before even placing an order during the interviews. No stores on the homepage, no address confirmation on entry, an outdated logo, and other small omissions made the site feel generic rather than recognizably Yandex.
Comparing before committing
Comparing options was part of almost every ordering flow. Since the cart lived on a separate page, users constantly bounced between browsing and reviewing their selections.
Loading that looked broken
No skeleton states meant a slow connection didn't read as "loading" — it read as frozen.
Discovery left behind
Finding the right place or dish often took more effort than it should. Search lacked the filtering users expected, making even targeted searches unnecessarily slow.
Key Decisions
Keep comparison uninterrupted with a persistent cart
Users constantly compared running totals across dishes and stores while deciding what to order. Hiding the cart behind a separate page, as many Western delivery apps (Wolt, Uber Eats) do, would have interrupted that behavior. Instead, I proposed a three-column layout with categories, the menu, and a persistent cart that kept the total visible.
This wasn't a feature request. It emerged from observing ordering habits during interviews and was later adopted by both the PM and design lead.
Trade-off
At smaller desktop resolutions, the persistent cart competed too much with menu space, so it collapsed into a standard cart icon below a defined breakpoint.
Keep attention on the selected product while showing more
The PM initially proposed a YouTube-style layout with upsell items in a persistent side rail. I argued that once someone opens a product, their attention should stay on that product, not compete with additional recommendations. The design lead sided with this approach.
Keeping upsell in a horizontal row below the description preserved focus, simplified implementation, avoided awkward empty space on shorter product pages, and allowed significantly more recommendations than a side rail. The design lead agreed with this direction.
The first draft
The final version
Give sponsored products their own destination
Stores could buy ad placements inside Yandex Search: clicking one dropped users into a modal version of the product card, opened on top of the store's menu. The monetization feature itself was a business decision, not mine. But the existing implementation had two real problems:
— Users lost track of which store they were even in
— Modal sometimes failed to open at all, silently dropping paying advertisers' traffic into a generic store menu instead of the product they'd paid to promote.
Engineering flagged the second issue to me directly. Talking it through with them, it became clear this wasn't a quick fix — the modal itself was the problem, not a bug in it. That's what pushed me toward designing a full new page instead of patching the existing one: a design with breadcrumb navigation to keep users oriented, and a denser upsell grid than the standard product card.
The reasoning behind the density
A regular product page helps someone decide if this item is right for them; an ad landing page's job is to surface as much of the relevant catalog as possible to someone who's already shown intent by clicking. It's the same building blocks, product card and upsell, just tuned to different densities because the two pages are trying to do different things.
The solution shipped and later helped support new advertising partnerships.
Other shipped work
Tracking + order history merged into one screen
Few users found their way back to tracking once they'd navigated away, and maintaining two separate pages carried ongoing dev cost. I proposed merging them behind the profile icon over the lead's preference for keeping them separate; we agreed to A/B test it, but I left before it shipped.
Checkout
A structural and visual cleanup of the existing flow. The team decided together to place a Yandex Plus subscription banner here — strong visibility, extra screen real estate, and a natural moment to offer a discount via cashback.
Global entry point for new markets
Instead of maintaining separate country-specific pages, I designed a single entry point that could scale with future market launches. Users could check service availability, switch languages, and download the app from one consistent experience.
Desktop design system
Comprehensive desktop design system covering components, states, edge cases, and developer documentation.
Result
Project completed in 2023. Internal analytics are no longer available, so the outcomes below reflect long-term product adoption rather than historical metrics.
From forgotten platform to reusable foundation
Adopted
The team reused the same layouts and patterns when launching Yandex Eats in new markets instead of designing new experiences from scratch.
Standardized
Shared components and patterns replaced years of duplicated UI and gave the desktop product a maintainable design foundation.
Endured
Key decisions such as the persistent cart, information architecture and product pages are still part of the experience.
Reconnected
Instead of feeling like a phishing or a forgotten product from another era, desktop became a consistent part of the same family.