A product was unavailable for five days, and the report shows zero sales. A model that reads those zeros as weak interest may recommend too small an order. Separate observed sales from demand that had no chance to appear.
This workflow requires an enabled logistics module and suitable data sources; check availability in your account.
A practical sequence
Mark the days when the product was actually sellable. Track missing dates, promotions, returns and stock corrections separately.
Compare a forecast with a simple baseline and past periods. For sparse products, show a broad range or insufficient history instead of a precise point estimate.
Turn the estimate into a reorder proposal using supplier lead time, minimum order quantity, pack multiple and cash limit.
Expose infeasible choices. If timing or cash cannot prevent a shortage, the team must choose a trade-off rather than hide it.
A worked example
Suppose 100 units sold during 20 in-stock days, followed by five out-of-stock days. An average of four across all 25 days understates observed sales velocity; five across in-stock days is still only a starting point when promotions or seasonality matter.
What to verify
A forecast is a decision range, not a demand guarantee. Supplier lead-time reliability needs observed history; a manual assumption must be labeled as such.
Check historical availability before preparing a replenishment plan in Yasnora AI.
