OMNICHANNEL DETAIL
One sale, one story. Status and next step on top; history that explains itself; the right modules in the right order for Point, QR, and Payment Link.
/// Session replays exposed back-and-forth loops and ID-copy workarounds. Heat/click maps concentrated on secondary data while the next step under-performed. CX tickets clustered around status, what to do, and where the history lives. In IA tests, people consistently completed tasks faster when:
•Status + Next step were pinned and visible at all times.
•The timeline highlighted incidents and collapsed routine noise.
•Names and actions stayed identical across ML and MP.
•Modules reordered by channel (same semantics, different priority).
On mobile, the pinned header significantly reduced time-to-action. The evidence gave us a clear bet: one sale, one story, with a decision at the top and a history that explains itself.
Role: Senior UX Designer — discovery to handoff; PM owned post-launch measurement.
Stakeholder alignment from day zero. Led working sessions with ML & MP stakeholders across levels (designers, PMs, tech leads, directors/VPs) to align goals, constraints, and decision guardrails.
Problem framing & success criteria. Reconstructed the legacy context (ML vs MP views, channel nuances), defined the decision-first objectives, and agreed on what “good” looks like for adoption, comprehension, and time-to-action.
Insight gathering & synthesis. Consolidated inputs from usage data, CX, prior research, and stakeholder interviews into a clear root-cause map and opportunity areas (status/next step hierarchy, timeline clarity, naming drift, channel order).
Principles & direction. Co-defined with stakeholders the guiding principles (pinned Status + Next step, incident-aware timeline, per-channel module order with consistent naming) and the trade-offs we were willing to make.
Wireframing & IA exploration. Produced multiple IA options and task-based validations; iterated low/mid-fi until convergence on a layout that was both fast and learnable across Point, QR, and Payment Link.
High-fidelity delivery. Shipped final UI with full states (empty/error/no-permission/success), mobile considerations (one-thumb reach), accessibility AA, and performance budgets (p95 ≤ 400 ms per block).
Executable handoff. Delivered build-ready packages per channel (flows, states, naming glossary ML↔MP, “can vary / never changes” table, anti-patterns), plus minimal UI/API contracts and event hooks.
Governance & rollout support. Facilitated cross-squad reviews, documented decisions, and supported feature-flag/canary rollout; ensured observability was wired so the PM could own post-launch measurement on adoption and behavior change.
/// No single source of truth. The squads who built Sale Detail (ML) and Activity Detail (MP) were gone; specs were scattered, terminology had drifted, and past decisions weren’t documented.
/// Two stories for one reality.
•Mercado Libre leaned commercial/order (items, delivery, refunds).
•Mercado Pago leaned collection/financial (auth/capture, reversals, disputes, settlement).
Same events, different surfaces and CTAs—users had to stitch a single truth.
/// Real jobs behind every visit.
1.What’s the current status?
2.What’s the very next action?
3.What’s the trace so far (timeline + incidents)?
These weren’t obvious at the top.
/// Channel-driven hierarchy.
•Point: receipt + on-device retries dominate.
•QR: attempts/authorization windows matter.
•Payment Link: checkout states + link edits.
A single fixed layout penalized at least one segment.
/// Platform & data realities.
Web vs app constraints (one-thumb reach); asynchronous events (finance/logistics) arriving out of order → UI must show eventual consistency clearly.
/// Naming drift.
“Refund,” “reversal,” “chargeback,” “dispute” were used inconsistently across teams. We needed a shared glossary: one action = one name = one icon.
/// Constraints.
PII redaction and auditability; no dark patterns in action copy; reuse of design-system components; telemetry gaps to be closed with instrumentation by design; gradual rollout behind flags per channel/region.
/// Signals it was broken.
Heatmaps focused on secondary details; low interaction with next step; replays showed pogo-sticking and copying IDs; CX tickets repeated the same three questions.
/// Why now.
The omnichannel push made inconsistencies painfully visible (sellers using Point + Link had to relearn the page). Reducing handling time and raising seller trust required a unified decision architecture.
/// Assumptions to validate (design bets).
Pin status + next step → faster time-to-action; surface incidents in a collapsible timeline → instant narrative; consistent naming/CTAs across ML/MP → less friction; reorder modules by channel (keep names) → relevance without chaos.
/// 045WRK
/// When checking a sale, users can’t immediately see status and the next step, nor reconstruct a trustworthy narrative. They jump between blocks/products, face inconsistent naming, and pay a channel learning tax—driving up handling time and support load.
/// Session replays exposed back-and-forth loops and ID-copy workarounds. Heat/click maps concentrated on secondary data while the next step under-performed. CX tickets clustered around status, what to do, and where the history lives. In IA tests, people consistently completed tasks faster when:
•Status + Next step were pinned and visible at all times.
•The timeline highlighted incidents and collapsed routine noise.
•Names and actions stayed identical across ML and MP.
•Modules reordered by channel (same semantics, different priority).
On mobile, the pinned header significantly reduced time-to-action. The evidence gave us a clear bet: one sale, one story, with a decision at the top and a history that explains itself.
