Back

varunmashru@gmail.com

+91 8238659055

Designing a first-in-class user experience of an AI-powered e-commerce distribution operations optimization platform

Context

This AI‑powered e‑commerce distribution operations optimization platform helps Amazon FBA sellers optimize spend, content, and operations. When I joined, the product had PMF signals and an active user base, but needed a scalable design system, a coherent information architecture, and a higher‑fidelity UX to sustain growth and broaden adoption.

Objective

Evolve the MVP into a production platform that:

  • exposes AI insights continuously and transparently

  • clarifies what the AI decided vs what the user decided

  • gives expert users granular control without overwhelming newcomers

  • compresses time‑to‑insight with dense but readable visualizations

Role & Collaboration

I led design across IA, systems, and data visualization while partnering daily with product and engineering. I conducted and synthesized user interviews and feedback sessions, and collaborated with leadership and operations to align on business goals and terminology.

Approach

Systematize and scale

  • Built a shared design language: tokens, components, patterns, and documentation for consistent handoff and faster iteration.

  • Reworked the IA to reflect how sellers think: catalog (ASIN/SKU), spend, content quality, forecasting, and AI recommendations.

Make AI legible and trustworthy

  • AI is present on every critical screen with inline guidance, next‑best actions, and rationale.

  • Clear attribution of actions and outcomes: what the model changed, what the user changed, and the impact of each.

Compress time‑to‑insight with purposeful density

  • Designed dashboards that surface the “first 30‑seconds aha”: status, outliers, and trends before deep‑dive.

  • Used sparklines, segmented timelines, stacked bars, and guided tables to show state, change, and causality without clutter.

Signature Modules

  • AI Content Builder (early 2023): A from‑scratch experience for generating and auditing product content when few credible references existed. Balanced AI suggestions with human controls, versioning, and publish‑readiness checks.

  • PPC Budget Strategy Visualization: A timeline view of strategy by week, with planned vs actual performance and a forward forecast to inform next‑week allocation. Connects spend to outcomes, not just spend to spend.

  • Product Content Quality & Publishing Dashboards: Grouped by ASIN/SKU, with diagnostics for missing or weak content, items not pushed to Amazon, and prioritized fixes. Converts a messy catalog into an actionable queue.

Research & Validation

  • Conducted iterative user interviews and feedback cycles with power users and operators.

  • Used testable prototypes to tune thresholds, visual density, and copy until users reliably reached the intended insight path.

Outcomes (Qualitative)

  • Faster decision loops due to always‑visible AI guidance and clearer provenance of changes.

  • Higher confidence in AI recommendations thanks to transparent rationale and human override paths.

  • Team velocity gains from the design system and cleaned‑up IA, reducing rework and enabling parallel delivery.

Impact

The platform’s UX evolved from PMF‑level to a differentiated, production‑grade experience suited to expert decision‑making. The platform’s clarity and control became a selling point with customers, while the design system and IA reduced friction for continuous feature growth.