Introduction

Yandex Eats is a food delivery service connecting customers, restaurants, and couriers. When something goes wrong with an order — a late delivery, a missing item, a payment issue — support operators investigate and resolve it through an internal admin system.

The order page is the core of that workflow: it brings together information about the customer, courier, restaurant, payment, and delivery status. But years of piecemeal development had turned it into a fragmented interface where even finding basic information took time.

This case covers rebuilding the order page around how support actually diagnoses a problem, cutting information lookup time roughly in half.

Product Redesign

GETX

2025—2026

Other cases

Redesign

GETX

2025—2026

Other cases

Context

Business

Support quality was becoming a business problem: customer reviews increasingly mentioned slow or unhelpful support, while high operator turnover meant new people regularly had to learn a complex internal system.

At the same time, Yandex Eats was preparing to expand into new geographies. The admin had grown piecemeal without design involvement or reusable patterns — difficult to learn, maintain, and scale.

And this wasn't a niche internal tool: every support operator used it, and every customer request passed through it accordingly.

Problem

Years of piecemeal development had left the order page without a consistent structure. Information was placed wherever there was room for it: high-use and obsolete elements competed for attention, related data was split across the page, and visually prominent sections often had little relationship to how frequently operators needed them.

Operators had the data they needed, but the page gave them no clear path through it

Constraints

There was no existing design system: patterns had been hardcoded independently over time. The timeline was tight relative to the number of known issues, and there was internal resistance to investing additional development effort into an internal tool.

Work Scope

The support order page — customer, courier, timing, order, payment, and restaurant information — plus search, courier tracking, core actions, and the design system underneath it.

Research

I started by learning how operators actually worked: a week immersed in support workflows, operator training, and real support call recordings.

After that, I followed my research with:

10 operator interviews

10 operator interviews

Prioritization survey

Prioritization survey

Timed usability testing

Timed usability testing

10 operator interviews

Prioritization survey

Timed usability testing

The interviews surfaced a broad set of issues, but the timeline made solving everything unrealistic. I used a prioritization survey to identify what mattered most to operators, then benchmarked the existing admin through timed usability testing — measuring how quickly they could analyze an order and find key information. This gave me a baseline to validate the redesign against later.

2. Bringing courier tracking into the workflow

Keep attention on the selected product
while showing more

Courier location was critical when investigating delays, but previously required leaving the admin and opening Yandex Go.

I integrated tracking directly into the order page, but kept it on demand rather than loading the map by default.

  1. Building a system instead of another one-off page

I designed a reusable component system alongside the new order page, covering the core components, states, patterns, and developer specifications needed across the admin.

The development team initially pushed back on introducing the additional process during an already time-constrained project. The PM and I made the case in terms of development reuse, easier maintenance and upcoming geographic expansion, rather than visual consistency alone.

Other shipped work

Search & secondary pages

I simplified search by keeping frequently used parameters upfront and moving rarely used fields into advanced options. Status and change-history pages already worked well, so I kept their logic and refreshed them with the new component system.

Better modals

I introduced task-focused modals for actions that needed more space without taking operators away from the order page, including courier tracking with key delivery details and a new date and time picker.

Real-time change notifications

I added notifications for order updates that happened while an operator was working on a ticket, reducing the risk of acting on outdated information.

Some future work

Dark theme

Frequently requested by operators working full shifts in the admin. It was prioritized as the next design-system improvement after the core order page was shipped.

Keyboard shortcuts

An interview insight led to a proposal for shortcuts for repetitive actions. The direction was approved, but implementation was left for a later iteration.

Further design system rollout

The new system was designed to extend beyond the order page. The next step was bringing more internal admin surfaces onto the same components and interaction patterns.

Result

Faster

I repeated the original courier-delay test with a different group of roughly 20 operators, using the same scenario and questions.

For instance, in one scenario operators had to determine:

Is the order late? → By how much? → Why? → Where is the courier? → What should happen next?

The average task achievement time per step dropped from 13–14 seconds to 5–8 seconds.

Shorter

The page itself also became roughly 40% shorter after obsolete and redundant information was removed and the remaining content restructured.

Systematized

Reusable foundation for subsequent admin pages

I left Yandex before longer-term operational metrics became available, so unfortunatley I don't attribute changes in support KPIs or business performance to the redesign. The result I could measure directly was clear: operators could find the information needed to diagnose the same support scenario in roughly half the time.

I design clear, thoughtful interfaces for complex digital products, combining strong product thinking with attention to interaction and visual detail

I design clear, thoughtful interfaces for complex digital products, combining strong product thinking with attention to interaction and visual detail