YK●

Case study 02 · Enterprise UX · Shipped

Designing
for judgment.

Reframing a dense moderation workspace so experts could make faster decisions without losing the context behind them.

My roleProduct Designer
UsersJunior & senior moderators
ScopeWorkflow, interface, measurement
ProductDigikala CMP

01 · Context

A tool built around fields, not decisions.

Moderators approve, reject, edit, and delete products under time pressure. The existing CMP spread evidence and controls across a dense interface, increasing cognitive load on every review.

Design tensionSpeed mattered, but removing context could increase incorrect approvals. The redesign had to reduce navigation—not judgment.

02 · Research

Watch experts build a mental model.

I combined a heuristic evaluation with contextual inquiry across categories and seniority levels. Moderators worked through real queues in think-aloud sessions, while seller-side research exposed the downstream cost of inconsistent rejection decisions.

2moderator experience levels
5high-frequency decision types
12key interactions instrumented

03 · Insights

Efficiency came from context.

01

Experts scan by risk

They do not read the page top to bottom; they jump to evidence that changes the decision.

02

Guidelines arrive too late

Rules outside the task interrupt flow and force memory-dependent judgment.

03

Editing and deciding compete

Corrective work obscures the primary approve-or-reject outcome.

04

Speed alone is unsafe

Review time needs quality guardrails such as reversals and repeated rejection reasons.

04 · Solution

A decision-first workspace.

I reorganized the information architecture around the review decision, then designed contextual guidelines, clearer duplicate detection, and a dedicated senior QA workflow.

Anchor the decision

Keep approve, reject, and reason selection visible and distinct.

Progressively reveal evidence

Prioritize product identity, imagery, attributes, brand, and category by risk.

Measure the real workflow

Track review time, edits, guideline use, rejection reasons, and reversals.

Less interface hunting.
More deliberate review.

CONTEXT AT THE POINT OF DECISION
Trade-offSpeed could not be optimized alone. Moderation time was paired with rejection, duplicate, and quality signals so the product would not reward careless decisions.

05 · Outcome

Faster—with a quality guardrail.

After launch, the number of products reviewed per moderator per hour increased by 10%. Duplicate detection improved, unnecessary navigation decreased, and the workflow became more consistent across experience levels.

+10%products reviewed per moderator per hour after launch
Measured beyond speedEvent taxonomy covered moderation time, rejection, duplicates, editing, guidelines, and QA behavior.
NEXT CASE STUDY

Sheypoor
Save Search

→