Stop Guessing. Reorder What Actually Sells.
You've got cash tied up in stock that hasn't moved in nine months — and you're stocking out on the items your best customers actually buy. Your buyer's gut feel was right ten years ago. Now there are 4,000 SKUs and Excel can't keep up. We build forecasting that looks at your real sales velocity, lead times, and seasonality, and tells you what to reorder, in what quantity, and when.
The promise
Where it hurts
The real friction we hear about.
Your buyer is drowning
They're trying to track lead times, sales velocity, seasonality, and supplier MOQs across thousands of SKUs in their head and a few spreadsheets. Some SKUs get attention; most get reordered to the same level they always have been.
Reorder points were set years ago
Min/max levels in your ERP were set when you launched the product line. Sales patterns have changed. Lead times have stretched. Nobody's gone back to retune the rules.
Working capital is locked in slow movers
A 5% slice of your SKUs is doing 80% of your sales. The other 95% is sitting in racking, costing you in cashflow, warehouse space, and shrinkage.
You don't want a £40k SaaS subscription
Enterprise demand-planning tools are designed for retailers with 50+ stores and a forecasting team. SMB pricing on those products either doesn't exist or doesn't make sense for your margin.
How it works
What we actually build.
Pull your real sales history
From Shopify, WooCommerce, Sage, Xero, or your bespoke trade counter system. We've done this for trade counter retailers, ecommerce shops, and B2B suppliers.
Model demand per SKU
Seasonality, growth trend, weekly pattern, customer concentration. A typical model finds three to five clusters of behaviour across your SKU base — "everyday consumables", "trade-counter season peaks", "long-tail bespoke" — each with different reorder logic.
Layer in lead times and supplier reality
Lead times vary by supplier and season. MOQs. Container fill optimisation. Your real cash-flow rules ("don't spend more than £X with that supplier this month"). These constraints shape every recommendation.
Recommend, don't auto-order
You get a weekly reorder recommendation per SKU, ranked by criticality and confidence. Your buyer reviews and approves. The system learns from their overrides — which suppliers they actually trust, which slow movers they refuse to let stock-out.
Proof
Real outcomes, not slideware.
Sales velocity model across the SKU base. Buyer time on reorder planning reduced significantly while stock-out rate on bestsellers dropped.
Industry fit
Where this lands hardest.
Pricing
Fixed fee, phased delivery.
£4,800–£12,000 for build, ~£250–£600/month depending on data volume
From £4,800
Typical stock forecasting build
- Workflow audit + spec phase
- Build, integrations, and tuning to your data
- Live deployment in your environment
- Monthly support and accuracy monitoring
Questions
Things people ask us about this.
01 How is this different from min/max levels in our ERP?
Min/max is a static rule. Real demand isn't static. AI forecasting looks at the last 12-24 months of sales velocity, weekly patterns, growth trend, and lead times — and recommends a dynamic reorder quantity that updates as the data updates. Min/max says "always have 30." Forecasting says "next week you need 47, the week after 22, because here's the pattern."
02 What if our sales history is messy?
Most are. Returns, refunds, promotional bumps, one-off bulk orders that distort the average. We handle the cleanup as part of the build — anomaly detection, outlier removal, holiday adjustments. You don't need perfect data; you need data, and we'll work with what you've got.
03 Can it handle bespoke / one-off items?
It flags them separately. Bespoke items don't get forecast — they get a separate workflow for material reservation against confirmed orders. The forecast covers your stocked SKUs; the bespoke side stays human-led.
04 Does it integrate with our existing ERP?
Yes — we read sales and stock from your system, and we can write reorder POs back if you want that level of automation. Most clients prefer to keep the human approval step, with the system generating the suggested PO ready for one-click send.
05 How long until we see results?
Recommendations are live in 6-8 weeks. The financial impact (lower working capital, fewer stock-outs) typically shows up over 3-6 months as old slow stock works through and new reorder patterns take effect.
Related
Other AI patterns we deliver.
Let's see if we can help.
A 15-minute chat with Chris & Kay. No slides. No pitch deck. You tell us what's on your plate; we follow up by email with real thinking.