Margin visibility before every purchase,
across 30,000 classified product lines
FARO is an off-price retailer sourcing inventory from major European brands. Each purchasing cycle required classifying around 6,000 product lines per container, across five containers, into FARO's own category hierarchy. By hand. Before a margin estimate was even possible. Pearstop automated the full classification pipeline and linked it directly to the POS and retail direct data warehouse, giving buyers accurate margin and sell-through estimates before committing to any purchase.
FARO sources inventory from multiple major brands, each with their own product codes, naming conventions, and data formats. The same product could arrive described six different ways depending on the supplier. Before any purchasing decision, every line item had to be mapped into a consistent schema so that margin, sale price, and expected sell-through could be calculated.
Inconsistent supplier codes
Every supplier used different product identifiers and naming structures, with no common reference across sources.
Fragmented POS data
Retail direct and point-of-sale data arrived inconsistently formatted, making it hard to link back to classified inventory.
No pre-purchase visibility
Without classified data linked to sales history, margin estimates were guesswork rather than a number you could act on.
Four steps from raw manifest to margin estimate
Raw supplier manifests
Hundreds per container, with inconsistent codes and naming conventions across suppliers.
AI classification engine
Trained on FARO's own categorised history; maps each line to the FARO product hierarchy.
Low-confidence review
Items below the confidence threshold flagged for human sign-off; only the genuinely ambiguous cases.
Margin estimate
Linked to POS and retail direct data warehouse; margin, sale price, and weeks-to-clear visible before purchase.
- Two people spent days manually classifying each container spreadsheet before a decision could begin
- Inconsistent results; the same product coded differently depending on supplier and who processed it
- No link to POS or sales history; margin was a best guess, not a deliverable number
- 30,000 lines classified automatically; full purchasing cycle processed in hours, not days
- Consistent hierarchy across every container and every cycle, regardless of supplier coding format
- Classification linked to POS and retail direct data; margin and sell-through visible before commitment
Automated processing at a fraction of the cost
Based on 2 FTE dedicated to manual classification, compared to Pearstop automated processing at equivalent volume. Indexed to manual cost baseline.
From classification engine to decision-ready intelligence
The follow-on project connected the classification engine directly to FARO's data warehouse, pulling in retail direct and POS data to give buyers decision-ready intelligence before any purchase is committed.
Weeks-to-sale by category
How long each product type sits in store before selling, broken down by category, warehouse, size, and supplier.
Sell-through modelling
Weekly progression from 100% inventory to clearance; break-even forecasting before a container is purchased.
Discount pattern analysis
Standard deviation of discount rates per category; identifying where markdowns are being over-applied.
Margin before commitment
Estimated margin per container calculated from classification plus POS history, before the purchase is signed off.
Weekly remaining inventory from arrival in store to clearance
The output buyers use to estimate container value before purchase. Each curve shows the pace at which a product category moves from 100% stock-on-hand to clearance.
Illustrative data based on the analytics structure built for FARO. Actual curves vary by category, season, and store location.
"We had thousands of product lines that needed to be categorised before we could even begin to understand our costs. Pearstop classified 95% of them in under a week. That would have taken our team six months and still wouldn't have been this accurate."
