Release $16M in Working Capital in 10 Weeks with One Data Scientist

Ten weeks, one data scientist, and AI forecasting turned a 100 store furniture distributor into an inventory-planning machine. If you run businesses where working-capital drag and showroom stock-outs gut margin, this solution is repeatable.

When you sell couches in 16 states, warehouse stock is EBITDA. Yet until recently this retailer relied on averages, gut feel, and Excel to decide how many recliners to buy. The result was a one-two punch:

• $34M tied up in slow moving SKUs clogging DCs

• Lost sales from empty slots on high velocity lines

5 years of POS, promo-calendar, and logistics data told the same story: over bought for the wrong times and SKUs while best-sellers vanished during holiday rushes. Analysts fought the noise with ad-hoc spreadsheets, but misses piled up and trust in Supply Chain cratered.

I dropped in one senior data scientist. Week 1 we piped five years of order history, promotions, and lead-times into the platform. Weeks 2-6 we trained gradient-boosted trees and Prophet ensembles to handle lumpy, intermittent, and smooth demand, every demand pattern seen across sectors. By week 7, nightly retrains and drift alerts were live. By week 10 the forecasts flowed straight into the replenishment dashboard, where analysts could place buy orders and supply chain could manage inventory levels.

Key features:

• SKU/DC forecasts for 10k items, 10-26 weeks out

• Fully automated MLOps with human-readable explanations

• Self-serve dashboard so merchants could sanity-check the math

The system beats human forecasts by ~15 percentage points, shaving $16M a year off carrying costs while capturing revenue the chain used to leave on the showroom floor. Inventory turns improved by low single digit percentages, dead stock shrank by the same order of magnitude, and the cash released is available to de-lever or fund new efficiency plays.

Just as important, Supply Chain doesn't have a credibility problem anymore. Merchants stopped second guessing the numbers, analysts became strategic partners, and leadership finally treats demand planning as a growth lever, not a cost center.

The solution backbone we deploy is sector-agnostic: whether the portfolio company staffs retail floors, machine shops, 3-shift plants, or regional DCs, the architecture ingests historical sales data, pricing, and demand signals, then surfaces actionable inventory recommendations that save over/under costs that tank EBITDA. Cutting carry costs is a margin play but it also frees cash that can flow to capex, bolt-on acquisitions, or straight to the balance sheet, outcomes any PE operator will recognize.

Want the playbook? Would love to chat.