Introduction
In the rapidly evolving landscape of online grocery shopping, efficient customer service is paramount. Everli, a fast-growing European scale-up, recognized the need to enhance its customer support infrastructure to keep pace with its growth. This case study delves into the challenges faced, the strategic design interventions implemented, and the subsequent impact on operations and customer satisfaction.
Context
Everli’s mission is to simplify grocery shopping by eliminating common hassles like waiting in lines and carrying heavy bags, ensuring seamless delivery experiences. However, rapid growth brought challenges. The existing customer support tools, initially designed for simple inquiries, struggled to manage the increasing complexity and volume of issues. This led to a fragmented system with unreliable processes, leaving operations agents without real-time updates, limited visibility, and inefficient workflows, culminating in delays and customer dissatisfaction. The pressing challenge was to design a solution to streamline issue resolution and optimize operations.
OVERVIEW OF THE NEW EXPERIENCE
Goals and impact
The goal was to redesign a centralized order management system for faster, more efficient operations. Previously, managing delays and reassignments was slow and frustrating. The new system enables real-time tracking and instant issue resolution in a single tool, allowing quick operator intervention.
Impact
✅ 20% more orders processed—effortlessly.
✅ 8% NPS boost—fewer delays, greater reliability.
✅ +5% of reported issues resolved in under 1 minute
Role and responsibilities
As a Product Designer, I led the UX strategy to resolve unassigned and delayed orders—aligning teams, defining priorities, and shaping the MVP. I collaborated with stakeholders and engineers to deliver scalable solutions, balancing agents needs with business goals. I owned the end-to-end design process, from research and analysis to interaction design, system integration, and testing.
Defining the problem
🔴 Lost Priorities: 70% of agents struggled to spot urgent orders, leading to delays and misallocations. High-priority tickets? Stuck in the queue.
🔄 Tool Overload: Agents juggled 4+ systems per order, wasting 6–8 minutes just switching screens. Log data confirmed 30% of their time was spent navigating, not solving.
💔 Trust on the Line: 62% of negative feedback pointed to delivery delays. Analyzing 5,000+ complaints made one thing clear—slow orders cost customer confidence.
UNASSIGNED ORDERS (image 1.0)
UNASSIGNED ORDERS DETAILS (image 1.1)
Getting started
Previously, agents had to search manually for unassigned orders—causing delays. Now, a dedicated tab surfaces them instantly, improving clarity and response time (image 1.0). Shopper responses are tracked in real time, showing who was notified, how, and when—enabling faster decisions without switching tools (image 2.0). Shared notes in the table reduce miscommunication by surfacing delivery constraints or bonuses at a glance (image 1.1). The UI uses visual hierarchy—icons and badges ease scanning, while hover popovers reveal extra detail only when needed. Notes act as a shared memory, with attribution ensuring continuity across shifts. Together, these elements support faster triage and confident, real-time coordination.
SHARED NOTES (image 2.0)
Teamwork made Visible
The Shared Notes (image 2.0) feature is a key improvement for agent collaboration in managing problematic orders. Previously, communication happened through an external messaging tool, which led to delays, frequent misunderstandings, and difficulties in identifying which order the issue referred to and how urgent the note was.
With Shared Notes, each agent can:
- - Read notes left by colleagues, gaining instant context and the latest updates on the order.
- - Write new notes to add details, report on follow-ups, or suggest next steps.
- - Flag urgent issues by setting a High Priority level, drawing immediate attention to critical matters.
This shared visibility centralizes communication, removes unnecessary steps, and significantly reduces the risk of errors—greatly improving both the speed and quality of operational response.
(image 2.0)
(image 2.1)
Flexible assignment
- If orders go unassigned, agents get real-time insights to match them with the right shoppers—no guesswork. The system highlights availability, flags conflicts, and prevents double bookings (image 2.0). Shoppers are auto-sorted by an algorithm based on multiple factors, showing the best options first. Agents can view full order details (image 2.1) already assigned to each shopper, enabling more informed, context-aware decisions. Built-in chat and call buttons allow instant action, keeping orders on track without delays.
Fallback when auto-assign fails
Key features like real-time shopper geolocation, availability and performance indicators, and direct contact options help agents step in when automatic assignment fails. They can quickly spot active shoppers—even outside the delivery area—assess reliability, and reach out with incentives to secure manual order acceptance, reducing delays and maintaining service continuity.
(image 3.1)
From phone calls to one tap
The new rescheduling feature in the OMS (image3.1) allows agents to intervene in critical situations when no shopper is available to complete a delivery. With a clear view of available time slots—based on store opening hours and shopper availability—agents can quickly reassign the delivery. Once a new slot is selected, the customer automatically receives a notification via the app to confirm the updated time, eliminating the need for phone calls and significantly reducing handling times and the risk of late communication.
Lessons learned
1. Design as facilitation
A product’s success isn’t just about good UX—it’s about engaging all teams and keeping everyone aligned throughout the process.
2. Iteration beats assumption
The best solutions come from testing and adapting, not guessing. Progress happens through learning, not leaps of faith.
3. Urgency over engagement In customer service tools, speed and clarity matter more than rich interactions. Get to the point, fast.
Biggest Challenge
A great tool is useless if no one wants to use it.
The issue? Experienced agents were reluctant to switch—new tools meant more effort.
The solution? Make the transition feel natural, engaging, and truly beneficial for them. 1. Agents First
From day one, I involved them through shadowing sessions, usability tests, and ongoing feedback—making them feel part of the change.
2. A Smooth Transition
The interface felt familiar, and the rollout was gradual—avoiding the shock of something entirely new.
3. Tailored Support
Live Q&As, hands-on guides, and a dedicated help channel built trust and reduced the learning curve.