// CASE_STUDY

Vitrinde

Marketplace platform for the Azerbaijan market

Vitrinde had demand on the buyer side — but seller onboarding was a cliff, not a ramp.

MARKETPLACEUXONBOARDING
Vitrinde preview
ROLE
Frontend Developer · UX Lead
DURATION
2023 — 2024
STACK
React · Next.js · TypeScript · Form Systems
PHILOSOPHY
Think first · Build once · Scale forever
CH_01
01/05
// THE HOOK

Sellers Were Leaving Before They Started

Vitrinde had demand on the buyer side — but seller onboarding was a cliff, not a ramp.

Analytics showed a sharp drop-off after the third step of seller registration. Support tickets repeated the same confusion: unclear requirements, inconsistent form behavior, no progress visibility.

The instinct was to redesign the UI. I pushed back — gently — and asked: do we understand why sellers abandon, or are we guessing?

This wasn't a styling problem. It was a trust and clarity problem wrapped in a technical one.

Beautiful forms don't convert. Clear, predictable systems do.

CH_02
02/05
// THE INVESTIGATION

Following the Drop-Off Trail

I combined quantitative funnel data with qualitative seller feedback before proposing any solution.

I instrumented each onboarding step with event tracking and mapped the exact fields where users hesitated or abandoned.

Three interviews with sellers who completed registration and three who didn't revealed the pattern: uncertainty about document requirements and fear of losing progress on refresh.

The strategy document I delivered prioritized progressive disclosure, persistent draft state, and a unified form validation language across every step.

PHASE A

Funnel Analysis

Identified step 3 document upload as primary abandonment point — 61% exit rate.

Next phase
PHASE B

Seller Interviews

6 structured sessions surfaced trust and clarity as root causes, not aesthetics.

Next phase
PHASE C

Solution Blueprint

Multi-step wizard with draft persistence, inline validation, and requirement tooltips.

CH_03
03/05
// THE FRAMEWORK

An Onboarding Engine, Not a Form Page

I designed a reusable onboarding framework that could power seller registration and future vendor flows.

The wizard was built as a state machine: each step had defined entry conditions, validation rules, and exit criteria. No step could be skipped without explicit user intent.

Form state persisted to local storage with server sync on blur — sellers could leave and return without starting over.

Validation was centralized: one schema definition drove both client feedback and API payload shape, eliminating the classic 'it looked fine on frontend' integration bugs.

STEP_ENGINEState Machine

Deterministic step progression with guard rails and rollback support.

FORM_COREValidation Layer

Shared schema — one truth for UI errors and API contracts.

PERSISTENCEDraft Sync

Auto-save on interaction, recovery on return — zero data loss anxiety.

FEEDBACKProgress UX

Visual step indicator, requirement hints, and success micro-confirmations.

CH_04
04/05
// THE SYNERGY

Translating Between Business and Build

The marketplace team spoke conversion. Engineering spoke endpoints. I became the translator.

Business stakeholders wanted more fields upfront — 'collect everything now.' Data showed that approach killed conversion.

I presented the funnel data in their language: each extra field at step 3 cost an estimated 12% completion. That reframed the conversation from opinion to evidence.

For the backend team, I delivered step-by-step API contracts aligned to the wizard state machine — no more 'we'll figure out the payload later.'

Business / Ops

Conversion impact dashboards

Agreement on progressive data collection over big-bang forms.

Backend

Step-scoped API contracts

Parallel development — frontend and API shipped in sync.

Support

Onboarding FAQ co-written from interview insights

Ticket volume on registration dropped within the first month.

CH_05
05/05
// THE IMPACT

More Sellers, Less Friction

The onboarding system turned registration from a barrier into a guided experience.

A/B testing on the new wizard versus the legacy flow ran for four weeks. Completion rate, time-to-first-listing, and support ticket volume were the primary signals.

The framework became reusable — the same onboarding engine powered a secondary vendor category months later with minimal new code.

Onboarding completion38% increase
Support tickets (registration)45% reduction
Time to first listing30% faster
Wizard reuse on new flows80% code reuse
  • Research turned a UI debate into a data-driven decision.
  • Systems thinking made onboarding scalable beyond the first use case.
  • Clear communication channels prevented siloed rework.
// NEXT_STEP

Ready to build something together?

I bring the same system-thinking approach to every project — from architecture to team alignment.