← Long Nguyen
Data & AI · ADA (Data & AI) · 2022–Present

Unifying eCommerce data across 20+ global brands

Twenty platforms, twenty definitions of the same number — here's how I made them agree.

SQLPythonETL PipelinesDashboards

The problem

Our team's goal was to centralize reporting across 14 platforms so brands could analyze their data at a glance. Coverage varies a lot by platform — some give us Sales, Traffic, Media, Content, Affiliate, and Operations data; others only Sales:

PlatformSalesTrafficMediaContentAffiliateOperation
Shopee
Tiktok
Lazada
Coupang
HKTVmall
Momo
FoodPanda
Shopify
Amazon
Zalora
Robinsons
Growsari
Rakuten
Blibli

And even where platforms overlap, each reports its own metrics under its own definitions. Without a shared definition, you can't deep-dive or give a brand good insights — you're just comparing numbers that don't actually mean the same thing.

What I did

I used data analytics and user-behavior understanding to figure out what each number actually meant, then normalized those definitions into a bronze → silver → gold pipeline: bronze is the raw report from each system, untouched and messy; silver is one centralized, standardized database where the Analytics team does its work; gold is the aggregated, combined view that dashboards and clients actually see.

Trigger: scheduled daily pull from every platform API Scheduled — daily pull, 14 platform APIs Bronze: raw exports, 14 platforms — lots of tiny files BRONZE Raw exports, 14 platforms — lots of tiny files Shopee · Tiktok · Lazada · Coupang · HKTVmall · Momo · +8 more Transform: normalize fields, dedupe, map to one shared schema Normalize & dedupe → one standard schema Silver: one shared database, fewer & bigger blocks SILVER One shared database, fewer & bigger blocks id · platform · metric · value · date USED BY Analytics team Transform: aggregate and join across Silver sources Aggregate & join across Silver sources Gold: combined from multiple Silver sources GOLD Combined from multiple Silver sources Report tables USED BY Dashboards & clients Nestlé · Reckitt · L’Oréal · Mars · +26 more brands

The team then built dashboards on top of gold, not on 14 different raw exports.

What happened

200+ pipelines now run every day on that foundation, enabling dashboards for 30+ brands. A few I'm proudest of:

Reckitt & L’Oréal
Real-time campaign dashboardLets the team act within minutes on live campaign days.
Nestlé, Mars & Colgate
Buyer & platform behavior analysisUnderstands targeted consumer behavior per platform.
Mars Wrigley
Digital Shelf performanceImproves positioning and top-of-shelf placement via image and content.
200+ pipelines running daily, enabling dashboards for 30+ brands including Nestlé, Reckitt, L'Oréal, and Mars
← Back to all projects