Poshmark
App Redesign
A ground-up rethink of the Poshmark mobile experience, simplifying discovery, listing, and social commerce for millions of users.
A ground-up rethink of the Poshmark mobile experience, simplifying discovery, listing, and social commerce for millions of users.
Poshmark built its reputation as the go-to social commerce marketplace for fashion resellers. But by FY2025, the cracks were showing. DAU was declining, Gen-Z adoption was lagging, and competitors like Depop and Vestiaire Collective were eating into the core seller demographic.
The timing for a comprehensive platform redesign was not arbitrary. A new engineering platform had become available, enabling a redesign at scale for the first time. The competitive window was closing. This was the highest-priority initiative for the year.
Growth had stalled. Sellers were leaving. The app felt dated, and a 4-minute listing flow was the single biggest reason new sellers never came back.
Seller activation had a 4x higher ROI impact than any other initiative in the roadmap. If sellers don't list, buyers don't buy. The funnel is entirely supply-side dependent, fixing seller experience was the only lever that could move GMV at scale.
I was the design manager responsible for the end-to-end redesign initiative. My role split between direct hands-on design work on the strategy and core flows, and leading the team across the broader surface area. Every major cross-team design decision ran through me.
That was the north star. Simple enough to align a cross-functional team of 30+ people. Specific enough to use as a decision filter. Every design choice, every feature prioritisation call, was evaluated against it.
Reduce listing friction to the point where it feels effortless. Target: under 90 seconds from tap to live. This meant redesigning photo upload, auto-fill intelligence, and step consolidation in parallel.
Make the feed feel personal, not algorithmic noise. Surface listings based on actual browsing behaviour, followed sellers, and trending styles, not just recency. Buyers return when discovery feels like serendipity.
Social proof at scale. Seller ratings, social signals, and community validation needed to be visible without cluttering the UI. Trust converts browsers into buyers, especially first-time purchasers.
Ship the new listing experience as the first milestone. The most impactful, fastest-to-validate change, and the one that unblocks seller growth. Proven ROI creates internal momentum for Phase 2.
A design system that scales across all surfaces, not just the redesign. Adopted by all 4 product squads (Seller Experience, Buyer Discovery, Social and Community, and Payments and Trust) within 3 months. The goal was to make future design work 3x faster and to eliminate cross-squad inconsistency permanently.
We ran qualitative interviews, session recordings, and funnel analytics across 30+ sellers and 15 buyers. Not all findings were surprising, but the ones that were fundamentally shifted our priorities.
We assumed the listing flow's step count was the core problem. It wasn't. Session recordings showed sellers were crashing out before step 2, at photo upload. On mobile, the upload flow was slow, crashed mid-session, and lost progress silently. Fixing step count without fixing photo upload would have shipped a faster path to the same abandonment point.
Power sellers weren't an edge case, they were the backbone of the marketplace. Designing the listing flow for a casual first-timer would have optimised for the 95% while potentially degrading the experience for the 5% who kept the platform alive. We designed for power sellers first, then extended accessibility to new users.
Buyers consistently asked for better search and discovery. We scoped it out of Phase 1. Seller activation had 4x higher ROI, and without supply, better discovery is irrelevant. We documented the buyer requests and committed them to Phase 2, rather than letting them dilute Phase 1 execution.
Analytics pointed to the home feed as a top-3 priority. Engineering estimated 6 months of backend work to properly personalise it. We made a deliberate call to ship a "good enough" feed in Phase 1, manually curated signals, and invest in algorithmic personalisation in Phase 2. Speed of learning beat perfection.
Every major design direction was tested against at least one alternative. These weren't aesthetic choices, they were strategic calls with measurable stakes.
Our hypothesis was that fewer screens meant faster completion. We designed and tested both: a single-page form that showed all fields at once, and a 5-step wizard that broke the flow into focused stages. The single-page approach felt faster to us in testing, but sellers felt overwhelmed by the visual weight of all required fields simultaneously.
The IA team proposed consolidating three navigation patterns into a single bottom tab bar. The PM pushed back, the existing nav had known session depth metrics attached to it, and any restructure risked regression. Rather than accepting or rejecting the pushback, we ran a controlled A/B test across a 10% traffic split for 3 weeks.
The existing Figma component library had grown organically, inconsistent naming, undocumented variants, and tight coupling to the old visual language. Option A was to extend and clean up the existing system. Option B was to start fresh with a properly tokenised foundation. Option A was faster in the short term; Option B was the right call for 18+ months of scale.
The rollout was deliberate and phased, not because we were being cautious, but because learning from each phase improved the quality of what came next. Each phase had clear success criteria before it was marked done.
The highest-ROI change shipped first. New wizard flow, rebuilt photo upload with background retry logic, AI-assisted category tagging. Shipped to 100% of users after a 4-week ramp. Listing completion rate improved 31% in the first 6 weeks.
New bottom tab nav rolled out with A/B test validation. Feed personalisation v1 using manual curation signals, full algorithmic personalisation scoped to Phase 3. IA restructure eliminated the hamburger menu entirely.
Design system adopted across all 4 product squads (Seller Experience, Buyer Discovery, Social and Community, Payments and Trust) within 3 months. 500+ component variants with full accessibility coverage. Documentation and Figma handoff tooling built alongside, though this was underinvested early and had to be retrofitted.
Casual sellers (first-timers) and power sellers (50+ listings/month) had fundamentally different needs. The listing flow had to accommodate both without separate products, progressive disclosure was the solution.
500+ component variants had to meet WCAG 2.1 AA. We built accessibility checks directly into the design system documentation, each component shipped with a pass/fail status before it was available for use.
The redesign hit or exceeded every success criterion. The listing flow improvement alone drove the strongest seller retention lift in Poshmark's history for a single product change.
New seller activation rate improved 25% within 90 days of the Phase 1 launch. The photo upload fix alone accounted for roughly half of that gain.
The design system was adopted by all 4 product squads (Seller Experience, Buyer Discovery, Social and Community, Payments and Trust) within 3 months, ahead of the 6-month target. Cross-squad design consistency improved measurably in the following cycle.
App store reviews mentioning "listing" shifted from predominantly negative to predominantly positive. Power sellers specifically cited the speed improvement as keeping them active on the platform.
We sequenced the feed work after the listing flow, which meant buyers didn't see the improved experience until Q3. In hindsight, we should have parallelised both tracks from day one, the listing team and the feed team were largely independent, and the delay cost us 6 months of compounding data from the feed improvements.
We shipped a well-documented design system in Figma. What we didn't account for was the engineering lift to actually consume it, tokenisation, component mapping, and developer documentation were all retrofitted after launch. I'd invest in that infrastructure from day one and treat handoff tooling as a first-class deliverable, not a follow-up task.
The hardest part of this project wasn't making the right design call, it was getting the room to agree. Stakeholder alignment, PM pushback on the nav, engineering scoping debates: these took more calendar time than the design work itself. The lesson: start alignment earlier, document decisions immediately, and treat consensus-building as a design discipline.
Great design delivered poorly documented costs the team weeks in implementation. We treated documentation as the end of the process. It should be the middle, built alongside the design work, not after. If I ran this again, documentation would have a dedicated designer from Phase 1.
The hardest part wasn't making the right call, it was getting the room to agree. That's the real work of design leadership.
The complete Poshmark 2.0, every flow, every screen, shipped.
