Employees submitted purchase requests into a single form and waited. Requests routed by category codes chosen by people who did not know them, so 44% were misrouted and every misroute cost roughly nine days.
STUDY 30 / 30 · ANONYMISED · NDA · PROCUREMENT · US · FORTUNE 500
PROCUREMENT · US The request form that decides which process you are in.
RoleLead designer
Timeline11 weeks
Team2 designers, 1 PM, 2 frontend, 1 process SME
VerticalProcurement intake platform · 62,000 employees
Replaced code selection with plain description and let the system infer the route, policy path and approvals - with the requester shown their route and its rationale up front.
Misrouting largely ended. Cycle time fell substantially and the procurement team stopped spending its week rerouting other people's requests.
6%MISROUTED REQUESTSfrom 44%
11 daysMEDIAN CYCLE TIMEfrom 27
-84%PROCUREMENT TIME ON REROUTINGreclaimed
THE ARGUMENT
Why the obvious solution was wrong.
The study matters because the product problem was reframed before the interface was polished.
An intake form that requires an employee to select a commodity code is asking a question only the procurement team can answer, of a person who submits two requests a year. They guess. The guess routes the request into the wrong policy path, where it sits until someone notices, and the nine-day cost of that is invisible in every metric the function reports because the clock is measured from correct receipt.
The requester now describes what they need in their own words and answers three or four situational questions. The system infers commodity, policy path, threshold and approval chain, and - critically - shows the requester the route it selected, why, and what will happen next with an expected duration. Where the inference is uncertain it says so and asks one disambiguating question rather than guessing silently. Misrouting fell from 44% to 6%, and procurement recovered most of a week per person.
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.
PROCUREMENT
EVALUATION / OPEN
Price
Risk
Delivery
DESIGN DECISIONS
Positions we would defend.
Each decision names the principle and the product consequence, not a stylistic preference.
Never ask for a code
A commodity code is procurement's vocabulary, not the requester's. Asking for it guarantees a 44% error rate.
Show the route
A requester who can see their path does not call to ask about it. Status transparency is a capacity intervention for the procurement team.
Ask once when uncertain
One disambiguating question is cheap. A silent wrong guess costs nine days.
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 Procurement intake platform · 62,000 employees, the transferable lesson is not a copied screen. It is the condition the interface had to make legible: The requester writes what they need in plain words. Commodity, policy path and threshold are inferred rather than asked. Rebuild that visibility for your own roles, risk, terminology, and operating cadence.
Protect the design rule
A commodity code is procurement's vocabulary, not the requester's. Asking for it guarantees a 44% error rate. 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 6% for misrouted requests, from 44%. 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.
Misroute cost analysis
Traced 300 misrouted requests end to end; median additional delay was 9.2 days and 71% originated at code selection.
Requester vocabulary study
Collected how 60 employees actually described 20 common purchases; almost none matched the commodity taxonomy.
Inference model design
Built the description-to-route model with the process SME and validated against 2,000 historical correctly-routed requests.
Status transparency test
Measured chase-call volume against route visibility; showing the route and duration cut inbound status queries by 76%.
“We were asking sixty thousand people to know our commodity taxonomy. They never did and they never will.”
VP Procurement Operations, Fortune 500 manufacturer · under NDA
NEXT STUDY · 01 / 30 · AI TOOLS