How much history do you need?

Twelve to twenty-four months of usable history gives the best result. With less, the system falls back to simpler models and is honest about lower confidence.

Can we feed in forward-looking events?

Yes. Promotions, contracts, project demand, end-of-life. They appear as adjustments on top of the statistical baseline — visible, dated, attributed to a user.

How is forecast accuracy measured?

MAPE and bias, tracked per item and rolled up per class. The system flags items where accuracy is degrading so you can investigate before it costs you a stockout.

Inventory & Purchasing

uPurchase

Demand forecasting and purchasing, in one screen.

uPurchase is the demand-forecasting and inventory-intelligence layer Business Central never had. It classifies every SKU, forecasts demand by location, sizes safety stock against a chosen service level, applies supplier calendars and lead times, weighs carrying cost against stockout risk — and then turns the whole picture into concrete, explainable purchase suggestions. Every recommendation answers the question "why" before you have to ask it: the forecast, the safety stock, the lead time and the current position all live on the same row as the quantity. Edit any number and the projected stock curve updates before anything is committed to Business Central. Purchasers get to challenge the numbers instead of trusting them blind, and controllers see service level, stockouts and excess as daily KPIs rather than monthly autopsies.

uPurchase demand chart with historical, future and forecast curves

Functionalities · 10

Advanced tier

uPurchase · Demand Forecasting

Forecasts that learn from what actually happened.

Statistical and pattern-based forecasts trained on your real consumption. Updated daily. Always visible — the trend, the seasonality, the confidence band — so you can sanity-check a number before you commit money to it.

Multiple models, best fit chosen

Croston for intermittent, Holt-Winters for seasonal, simple moving average for stable. The system picks the model that matches your data — and tells you which one it picked.

No black box

Open any forecast and see the historical actuals, the chosen model, the seasonality and the trend. Sense-check before you trust.

Inline overrides with reason

Saw a one-off promotion in the data? Mark it as an outlier. Expect a contract to start in October? Add a forward adjustment. The model respects what you tell it.

What you see

A forecast is a hypothesis.

Most forecasting tools produce a single number and ask you to trust it. uPurchase produces a number plus the working — the model, the assumptions, the historical fit. When something looks wrong, it usually is, and you can correct it inline before it becomes a stockout or an excess.

Live forecasts per item, per location — always current with the latest Business Central data
Automatic model selection per demand pattern (stable, seasonal, intermittent, lumpy)
Outlier detection — promotions, one-off projects, data errors
Forward overrides for known events (campaigns, contracts, end-of-life)
Forecast accuracy tracked over time, per item and per category
Side-by-side actual vs forecast chart across your full data history — look back as far as you want

uPurchase predicted inventory against safety stock and reorder point

When this matters

Signals you'd reach for this.

Seasonality is "common knowledge"

Everyone knows Q4 is heavier — but the reorder points do not. A model that respects seasonality fixes that automatically.

Promotions break next month's plan

A successful promo distorts demand history. Without outlier handling, you over-order for months afterward.

Long-tail items have no signal

C items move twice a year. Standard models give you garbage. Intermittent-demand models exist for a reason.

Lead times are treated as a constant

Most systems use one lead-time number per supplier. Real lead times drift, slip and bunch. Without that variance in the model, safety stock is either guesswork or a buffer for the worst case.

FAQ

The questions everyone asks first.

Still wondering? Ask us directly →

Both. For finished goods we use the sales signal; for components and raw materials we explode demand through the BOM from open production orders, the master schedule and dependent items. A manufacturer's forecast that only looks at the sales line is missing most of the picture.

See uPurchase in a 30-minute demo.

A real screen-share with someone who built it. No slides.

The uTools suite