Returns ran at 38%, and 71% of those cited fit. The size chart was accurate, comprehensive and used by almost nobody, because it required measurements the customer did not have.
STUDY 13 / 30 · ANONYMISED · NDA · D2C & COMMERCE · INDIA · SERIES B
D2C & COMMERCE · INDIA Size selection as a conversation, not a chart.
RoleLead designer
Timeline10 weeks
Team2 designers, 1 PM, 1 researcher, 2 frontend
VerticalApparel D2C · 4.2M monthly sessions
Replaced measurement entry with reference comparison - the customer names a garment they already own that fits, and the system reasons from that.
Fit-related returns fell by more than a third. The reference garment turned out to be the only body data most customers can supply reliably.
-38%FIT-RELATED RETURNSvs baseline
74%FIT-FINDER COMPLETIONvs 6% size chart
4.3 / 5SIZE-CONFIDENCE AT PDPfrom 2.1
THE ARGUMENT
Why the obvious solution was wrong.
The study matters because the product problem was reframed before the interface was polished.
Size charts fail because they ask for information the customer does not possess. Almost nobody knows their chest measurement in centimetres, and the ones who do measured themselves incorrectly. The chart is accurate and useless - a well-designed answer to a question the customer cannot answer. Meanwhile every customer has a wardrobe full of calibrated reference objects and no interface had ever asked about them.
The fit finder asks three questions: name a brand and garment you own that fits the way you like, how you prefer it to sit, and what you are shopping for. The reference garment resolves against a database of known measurements, and the system reasons the equivalent. Where confidence is low it says so and recommends the safer size explicitly. The mechanism works because it converts an unanswerable question about the body into an answerable one about a wardrobe, and because customers are far better at describing preference than dimension.
THE INTERFACE CRAFT
The interaction, rendered as a working product surface.
The specimen below is code-native and uses the study's own design logic. The client interface remains protected.
D2C & COMMERCE
YOUR FITM / REGULAR
Confidence 86%
DESIGN DECISIONS
Positions we would defend.
Each decision names the principle and the product consequence, not a stylistic preference.
Ask what the customer knows
An accurate chart requiring unavailable input is a designed failure. Reframe the question until it is answerable.
The wardrobe is the dataset
Every customer owns calibrated reference objects. No fit interface in the category was using them.
Name the safer size
In a fit decision, an honest hedge converts better than a confident guess, because the alternative to buying is not buying.
PRODUCT LEADER READOUT
What transfers, and what should remain specific to this product.
A case study is useful when its operating principle travels without turning the original interface into a template.
Read the operating condition
For Apparel D2C · 4.2M monthly sessions, the transferable lesson is not a copied screen. It is the condition the interface had to make legible: Name a garment you own that fits well. The system resolves it against known measurements and reasons the equivalent. Rebuild that visibility for your own roles, risk, terminology, and operating cadence.
Protect the design rule
An accurate chart requiring unavailable input is a designed failure. Reframe the question until it is answerable. Keep that rule in the acceptance criteria, component states, and production QA record so later visual cleanup cannot erase why the interaction exists.
Measure behaviour after ship
The evidence record is -38% for fit-related returns, vs baseline. Recreate the baseline and outcome window before rollout, segment the result by role and context, and state clearly what the measure cannot prove.
RESEARCH RECORD
The work behind the interface.
These artefacts connect the final interaction back to the evidence and product model that produced it.
Return-reason coding
Coded 4,000 free-text return reasons; 71% were fit and of those 62% specified a direction, which the model now uses.
Size chart usage instrumentation
Measured actual size-chart interaction at 6% open rate and under four seconds median dwell.
Reference garment database
Built the resolution layer across 340 brands' published measurements with the merchandising team.
Confidence threshold test
Established the confidence level below which recommending the safer size outperformed recommending the best guess.
“Nobody knows their measurements. Everybody knows which jeans fit. It seems obvious now.”
Head of Product, apparel D2C · under NDA
NEXT STUDY · 14 / 30 · D2C & COMMERCE