Influencer commerce platform

Influencer marketing suffers from fragmentation: brands in one tool, creators in another, payments and analytics in more disconnected surfaces. This piece follows the arc from audit to deployed product (design system, feature architecture, integrations, iterative deployment) for a dual-sided marketplace serving brands and creators.

The work spanned multiple verticals, including merging an existing community platform into the new experience: both a greenfield build and a careful integration exercise.

The problem

On day one there were two codebases: a campaign management tool with rough early UX, and a standalone community app with its own identity and backend. The mandate: audit both, redesign the experience, and ship a unified product.

The goal wasn’t to build features. It was to make two disconnected products feel like one inevitable experience.

Phase 1: Design audit

The audit covered information architecture, visual consistency, interaction patterns, and technical feasibility across both products.

The audit drove prioritisation: a phased roadmap to stabilise brand-side flow, ship the influencer storefront, then merge the community product.

Phase 2: Brand-side

Authentication & onboarding

Login and signup support SSO (Google, LinkedIn) and OTP email. Brand signup became three clear steps with progress indicators and inline validation to cut abandonment.

Content management system

A hierarchical CMS for catalog structure: categories, subcategories, product attributes, media. CSV bulk upload for large catalogs and a real-time preview panel before go-live.

Form builder for lead collection

Configurable forms attached to products or campaigns (text, dropdowns, checkboxes, multi-select), validation, GDPR consent in schema by default. Submissions stored relationally with custom event instrumentation.

Campaign creation wizard

End-to-end wizard for goals, pricing models (CPC, CPV, CPP, CPCall), budget, targeting, and scheduling. Unique affiliate links per product-campaign pair with shortening and UTM injection for attribution.

Influencer discovery & collaboration

AI-assisted discovery filtered by niche, audience size, engagement, geography; invite flows with briefs; recommendation logic from historical campaign performance.

Campaign dashboard & fraud detection

Real-time monitoring for clicks, views, conversions, ROI at campaign and per-influencer level. Anomaly detection on traffic patterns and velocity to catch fraud before spend runs away.

Phase 3: Creator-side

Influencer onboarding & storefront

Social OAuth to verify handles and pull follower counts. Storefront builder from vertical templates (fashion, services, real estate, etc.) with visual editor; responsive storefronts linkable from bios. Template system component-driven so new layouts ship without creator-side code changes.

Campaign marketplace & analytics

Creators browse campaigns by niche, payout, category; accepted campaigns get scoped affiliate links; activity flows to creator and brand dashboards via UTM. Earnings consolidate campaign, storefront, and community revenue with payouts and settlement windows.

Phase 4: Community merge

The hardest phase: merging a standalone community app (own API, data model, and users) into the main product without breaking active usage.

Stack and architecture

Transactional data lived in Postgres; flexible hierarchies (e.g. CMS) in a document store. A custom event wrapper fed both product analytics and attribution to avoid double-counting.

Deployment & iteration

Incremental milestones with internal testing and limited creator beta. Feature flags for rollout and rollback. Post-launch: budget forecasting in the wizard, new storefront verticals, clearer settlement timelines, fraud threshold tuning after false positives.

Outcomes

Dual-sided marketplaces with embedded community are a different challenge than single-persona tools: every decision has dual implications for brand vs creator.

The hardest problems weren’t technical. They were about making two products with different histories feel like they were always meant to be one.

Audit-first prevented building on structural sand. Treating community migration as its own product (scope, timeline, comms for users) is a principle I carry forward.

Design direction drew inspiration from Jan Losert’s work. Due to NDA and compliance, actual product screens and client deliverables are not shown here.