P&G Canada
Cracking Nielsen’s ETL: From 20 Days to 3
Rebuilt P&G Canada’s Nielsen POS pipeline so analysts walked into Walmart reviews with fresher market data than the retailer’s own buyers.
Founder’s track record · built while at P&G Canada. The in-house operator experience Clarivant is built on.
The starting point
Nielsen POS data was the lifeblood of category management at P&G Canada — and its UI extraction took up to 20 days when it worked at all. Timeouts and session failures meant analysts babysat extractions a third of their week, and the data reached sales and planning two or three weeks stale.
The method
I bypassed the UI entirely: reverse-engineered Nielsen’s backend database, opened a direct ODBC connection their tooling didn’t document, and built automated KNIME pipelines on scheduled VMs. Retry logic, row-count checks, and checksum comparisons recovered failures and caught retroactive updates automatically — and the output dropped straight into P&G’s reporting templates, killing a full day of reformatting.
The result
The SLA fell from 20 days to 3 — an 85% cut in data latency. Category managers walked into Walmart reviews with data fresher than the retailer’s own buyers had, contributing to P&G winning Category Captaincy in Oral, Feminine, and Baby Care and roughly $5M in annual POS uplift. The pipeline ran for years afterward as invisible infrastructure.
“Arturo rapidly learned new technologies and applied them to business problems — redesigned our bulk data extraction to be 10× faster within two months, then enabled others to scale the approach.”