Demand forecasting built for how ecommerce stock actually moves

Forecasts run on products, not listings. Sales from every channel feed one history per warehouse. Automatic picks the best of 12 methods by backtest — it is not AI.

Hero

How forecasting works

  1. Connect channels and import ~12 months of orders (24 optional).
  2. Link listing SKUs to products (pack quantities supported).
  3. Map locations to warehouses.
  4. Forecasts appear per SKU as soon as there are ~4 weeks of data; Automatic needs 8+.

Twelve methods

Automatic (default) · Recent velocity · Weighted recent · Trend · Seasonal · Trend + seasonal · Slow mover · Manual override · Similar SKU · Event calendar · Bundles and kits · Size and variant split

Always applied: exclude out-of-stock days; cap one-off spikes; show method + recent accuracy; FBA warehouse forecast is never added on top of main-warehouse demand when transfers are already counted.

Replenishment maths

Reorder point = daily forecast × lead time + safety stock
Safety stock = Z × σ_daily × √lead time (default service level 95%)
Every product gets a stockout date and an order-by date, recalculated daily.

CTA

Start free trial · See multi-warehouse · See AI assistant (explains forecasts; does not invent them)

See what runs out next

14-day free trial. No card needed. Or get a free stockout report first.