01.
CONTEXT
Walmart recently launched in South Africa.
Everything post-purchase was still being figured out - returns especially. With limited data, I leaned on Makro (Walmart’s wholesale retail brand in South Africa), competitor flows, and real customer conversations.
RESEARCH SOURCE
Walmart
Massmart
Parent
Subsidiary
Massmart
Makro
Existing SA operations
Makro
Returns Data
Baseline research
02.
THE PROBLEM
Inconsistent timelines
Refunds ranged from 24 hours to weeks. Pickups happened same-day or never.
Zero visibility
After handoff, the system disappeared. Users had no idea where their item was.
30%
Over-reliance on support
~30% of users needed help - not because returns are complex, but because the system didn't guide them.
During this project I used AI tools to synthesise research, generate IA explorations, critique flows, and speed up wire-framing. If you're curious about the behind-the-scenes workflow, I documented the full process.
03.
WHAT I LEARNED
Research → Insight
System disappears after handoff
Users didn't know who picks up, where products go, or what happens if something gets lost.
Logistics feels chaotic
Multiple couriers for one order. Different pickup days for different items. No coordination.
Returning is not intuitive
Users struggled to find orders, initiate returns, and understand the steps involved.
Speed ≠ Trust
Even fast returns felt unreliable. The issue was predictability, not velocity.
04.
CURRENT SCENARIO
67%
check return policy before buying
55%
drop off if returns feel difficult
10–15%
of Makro orders get returned
05.
UX VISION
Not just "make returns easier"
Make them predictable. Make them visible. Make them feel reliable.
Dynamic
Adapts to each return scenario.
Visible
Users see every step of the process.
Transparent
Users see every step of the process.
Self-serve
Minimal dependency on support
07.
EDGE CASES
Designing for when things go wrong
A robust return system isn't just about the happy path - it's about handling failures gracefully.
Pickup fails
Auto-reschedule with user notification. Fallback to store return after 2 attempts.
QC rejects return
Clear explanation of rejection reason. Option to escalate with photo evidence.
Partial refund
Transparent breakdown of deductions. Users understand before they commit.
08.
IMPACT
METRIC
BEFORE
AFTER
Return initiation
10-15 mins
<2 min
Resolution time
1-14 days
3-5 days
Support dependency
-30%
↓ 40–50%
Pickup failures
15-25%
<5%
Other works
View all works
Enterprise Data Experience · Web
Designing connected workflows for discovering, understanding, governing, and trusting enterprise data across 18 business domains.