Your data doesn’t need to be clean
Messy data is the starting point. Cleaning and classifying it is the product.
Anonymized Use Case · Cleaning & Soft FM
Nobody chose 30 toilet paper suppliers. They just couldn't see them.
In one cleaning business, 30 different suppliers invoiced for toilet paper within twelve months. Illustrative data, based on a real pattern in soft FM procurement - not a named client engagement.
The decision, once it's visible
Every line classified, the pattern was visible in days: here they are, cost, quality, and brand side by side, and the three worth keeping.
| Supplier | £/roll | Why |
|---|---|---|
| Gompels | £0.31 | Bulk price leader |
| Bunzl CHS | £0.33 | Contracted, Tork & Katrin |
| Pattersons | £0.38 | Regional next-day cover |
| Amazon | £0.41 | Spot buys, no terms |
30 suppliers found across twelve months. Three worth keeping, once the spend was classified and put side by side - yours to make, the table makes it obvious.
Off-contract buying
Contracted price, through Bunzl: £19.20 for a case of 36 rolls. Actually paid, through Amazon: £26.80 for the same case - 40% more. A cost risk and a compliance risk: client contracts specify products and suppliers, and every off-contract line weakens your position at renegotiation.
Same roll, five prices
The same product, Tork Advanced T4, priced per roll across five suppliers:
| Supplier | Price paid per roll |
|---|---|
| Supplier A | £0.31 |
| Supplier B | £0.36 |
| Supplier C | £0.41 |
| Supplier D | £0.47 |
| Supplier E | £0.52 |
A 68% spread on the same product - invisible, line by line, until classified.
AI-first, human reviewed
Every label checked. Classification holds up as new suppliers and invoices arrive.
Usage-based pricing
You pay for the lines processed. No heavy annual subscription.
They classified thousands of product lines in under a week. That would have taken our team six months.
Typical result in soft FM: 10-20% savings on addressable spend. The findings above are where it comes from.
Want to see this on your own spend?
Send a sample of your own invoice data and we'll show you the same kind of pattern in your own numbers.
