Returns through the call center were costly and drove low CSAT and LTSA scores, reflecting customer perception and brand loyalty. Customers preferred online returns, so execs asked us to incentivize adoption by making them easier and more accessible.
Goals
Increase returns done online vs the call center (saves money)
Increase Customer Satisfaction (CSAT) scores with the returns experience
Improve Likelihood to Shop Again (LTSA) scores after a return
Results that speak - one year later
↑
ONLINE RETURNS
6x
Increase in returns done online vs call center. From 7% to 47%.
↑
RETURN COSTS
12%
Reduction in cost per return. From $33 to $29.
↑
CSAT SCORES
28%
Increase in customer satisfaction scores (58 to 74).
↑
LTSA SCORES
25%
Increase in likelihood to shop again scores (56 to 70).
17
50
28
791
29
Unclear return policy
Customers struggled to find return dates, driving call volume.
Waiting in line
Customers disliked waiting, especially for simple returns.
Lost receipts
Receipts were often lost, making purchase verification a challenge.
UPS returns
Without home printers, customers found printing labels frustrating.
Big & bulky items
Large-item returns often meant difficult store visits or customer support calls.
Limited return methods
Most items could only be returned in-store, limiting flexibility.
Testing revealed a consistent pattern: customers’ expectations were shaped by industry norms. I connected these mental models to the biggest issues in our returns experience.
⚡️ Best Buy displays a barcode for fast order lookups
🗓️ Target clearly calls out the order's return date
📦 Walmart offers multi-return methods, including large item pickup
🖨️ Amazon's QR code limits the need for printers
Before
👁️ Moving the CTA in the viewport incentivized online returns
⚡️ Instant in-store return access via barcode
After
Before
🗓️ Made the "return by" date prominent to support task completion
📦 80%+ of customers use UPS stores that accept the barcode — making it the priority
After
Before
⚡️ Quick access to the UPS barcode
🗓️ Emphasized the "return by" date instead of "initiated" date
After
→
→
→
→