Apex

The tool priced every Pro sale, and the only way in was an email.

A Pro price is not a weekend promotion: it sits in the database for years as an evergreen offer, across every Lowe's store nationwide. But the only way to set one was to build the numbers in Excel, email the sheet to a developer, and wait for them to key it in by hand.

Nobody could see a mistake until it was live: one wrong decimal reached Lowes.com, every store, and everything downstream.

Pricing knowledge was also never captured in the system. Which pricing strategies applied to which stores, and what discounts were available, lived in analysts' heads. That knowledge disappeared whenever someone left or went on vacation.

Meet Dana, paint pricing analysts

Dana is a corporate analyst who focuses on Paint for Lowe's. When a new vendor adds 50 new paint colors, she builds a spreadsheet listing every SKU, price, and promotional offer that can apply to any of 1700+ Lowe's stores.

She knows this catalog better than the system does, and none of it is written down anywhere the system can read.

Shes does this 3-5 times a month, finishing her part quickly, but then she has to wait 2-3 weeks for the information to go live. When the live price is wrong, she finds out at the same time everyone else does.

The 5 decisions that created a source of truth

I validated every row before it could go live. Uploads pass through a gate that checks each row and catches errors before anything writes to the system.

I kept the work in Excel. Excel is where analysts already work and where vendor negotiations happen, so I built templates and a drag-and-drop upload instead of moving them into a new tool. Most of their process stays exactly where it was.

I turned a downstream limit into a guardrail. The processing tool we handed off to could only handle 500 rows at a time and dropped the rest with no warning, so a 700-row upload would look successful while 200 prices quietly vanished. I surfaced the limit in the interface and capped uploads so nothing could disappear unnoticed.

I let analysts edit offers in the app, but not prices. Offers, the pricing strategies layered on top, can be rewritten or removed in-app to match current strategy. Item prices stay in Excel.

I required a human to confirm every override. The simplest rule would let the newest upload overwrite whatever came before. That's fine for a routine line item, but dangerous for an offer that has held for years. So I surfaced every conflict before the commit step and made someone confirm each one. The tool proposes. The human commits.