Product Case Study

Product Data Platform (Product Catalog)

A 0→1 product data platform that normalized product data across five ingestion sources into one schema, taxonomy, and identity model.

APIs, SFTP, cloud storage, UI uploads, and e-commerce connectors each delivered product data in their own shape — with no shared way to resolve the same product, its variants, or its parent across feeds. The merchandiser browse/drilldown view below was one downstream consumer of the platform, and it became a cited differentiator across enterprise sales opportunities.

Product

0→1 product data platform

Impact

Cited sales differentiator

Stack

React + TypeScript + GraphQL

Outcome

$1.1M+ influenced ARR

Representative Product Demonstration — This page recreates the product experience using original HTML/CSS components and synthetic data. It illustrates the product concepts, UX, and technical implementation while avoiding reproduction of proprietary interfaces.

Try It

Browse a category. Open a product.

The experience below represents one downstream consumer of the product data platform. Behind it sits a normalized product model, ingestion framework, taxonomy, product-identity rules, and validation and transformation pipeline that also fed personalization, recommendations, and campaign targeting. Try it below — click through the category tree on the left, switch between list and grid view, or click any product or category to open its analytics panel. Everything below is a working, original recreation of the interaction model, running on synthetic data.

All Products — Last 30 Days

Total Revenue

$1.3M

+8%

Total Purchases

18.4K

+5%

Total Views

412.8K

+12%

Cart Abandons

9.8K

-6%
CategoryInterestRevenueSalesViewsCart Abandons

Case Study

The story behind the product

The Insight

Product data arrived through five different doors, and none of them agreed

APIs, SFTP, cloud storage, UI uploads, and e-commerce connectors each delivered product data in their own shape. Nothing enforced a shared schema or taxonomy, and there was no reliable way to answer the basic question underneath every downstream decision: is this the same product, a variant, or a different listing entirely?

The Product Thesis

A normalized product data platform, not another catalog UI

Create a normalized product data platform capable of ingesting multiple source types, enforcing a consistent schema and taxonomy, resolving product identity — same product, variant, or parent, with clear attribute inheritance and overrides — and providing a trusted foundation for downstream capabilities including personalization, recommendations, campaigns, and analytics.

The Outcome

The platform, proven by one of its consumers

The merchandiser browse/drilldown view was one downstream consumer of the platform — built to prove the underlying model held up under a real workflow. It became a capability prospects asked about directly, cited as a differentiator across enterprise sales opportunities and $1.1M+ in influenced ARR credited to solving the harder problem underneath: normalizing product data.

My Role

Sole product owner, concept to shipped capability

Product Leadership

  • Owned the ingestion strategy across five source types: APIs, SFTP, cloud storage, UI uploads, and e-commerce connectors
  • Designed the schema, 10+ level taxonomy, and product-identity rules — same product vs. variant vs. parent, attribute inheritance and overrides
  • Defined the category-drilldown and analytics-panel interaction model as the platform's first downstream consumer
  • Prioritized MVP scope, then extended the platform's reach into personalization, recommendations, and campaign targeting
  • Positioned the platform as a competitive differentiator in enterprise sales conversations

Technical Direction

  • Ingestion framework spanning APIs, SFTP, cloud storage, UI uploads, and e-commerce connectors
  • Schema design, validation, and transformation pipeline enforcing a consistent product model across sources
  • Hierarchical taxonomy (10+ levels) with attribute inheritance and override rules
  • Product identity resolution — matching and mapping the same product, its variants, and its parent across feeds
  • React + TypeScript front end with a GraphQL API layer, cursor-based pagination, and a multi-panel drillable interface for the merchandiser-facing consumer

Outcome

What happened

  • Normalized product data from five disparate ingestion sources into one consistent schema and taxonomy.
  • Resolved product identity — same product, variant, or parent — across every source feed.
  • Powered downstream capabilities beyond the catalog UI: personalization, recommendations, campaign targeting, and analytics.
  • Gave merchandisers a single browsable view of category and product performance instead of static exports.
  • Became a cited differentiator across enterprise sales opportunities.
  • Directly influenced $1.1M+ in ARR.

What I'd Build Next

  • Expanded Create Segment options — the criteria shown here, plus inverse segments (didn't purchase, didn't view, didn't abandon)
  • More analytics within the side panel, with more call-to-action options
  • Simple campaign creation directly from the catalog, via a campaign modal or slide-over
  • AI insights — product clusters, recommended segments, and product recommendations
  • A channel-sourced lens for the unified listing panel, toggled alongside the behavioral-signal view

Why This Matters

The hard problem was never the browse screen — it was making the platform agree with itself on what a product is, across five ingestion sources and a 10+ level taxonomy. This is what that looks like when product thinking and hands-on engineering come from the same person.