Skip to content
Ailaaj Academy

Why projection uses forecasting sales velocity

In one line: the quantity we suggest to buy is driven by the same sales velocity we use to forecast — so procurement and forecasting never disagree.

The easy version we didn’t keep

The obvious way to project demand is to pick a date range, add up units sold, divide by the number of days, and call that the daily rate. It’s simple — and it’s fragile. A single promo week, a stockout (which suppresses sales), or a quiet stretch swings the average wildly. Two people picking two date ranges get two different “truths,” and the number you buy against has no relationship to the number your forecasting screen shows.

What we do instead

Projection asks the demand-forecast engine for each product’s sales velocity (SV) — the very same sv_xyz_merged the Demand Forecast feature reports. This is the part most procurement tools get wrong, and it’s a real differentiator: rather than a flat lifetime average, our SV is a more sophisticated, seasonality-aware velocity. It’s an XYZ-weighted blend of short, medium, and long windows (7 / 30 / 90 days) built to track seasonal swings and recent momentum — so a winter cough-syrup surge or a quiet summer for a seasonal line is reflected, without letting one odd week dominate or a long-dead tail drag the number down. Then the order quantity is simply:

order to cover Desired Inventory Days, minus what’s already on the shelf.

Why this matters on the floor

  • One source of truth. The pharmacy manager who trusts the forecast can trust the PO — it’s literally the same SV underneath. No “the forecast says X but the buy says Y.”
  • Stable numbers. Because SV smooths across windows, the suggested quantity doesn’t lurch every time someone changes a date filter — there is no date filter to change.
  • Honest gaps. A product with no recent movement has no velocity, so it isn’t suggested. Projection focuses your buying on what actually sells, and leaves new or dead SKUs for a deliberate decision.
  • Scope that matches reality. You can project across several stores/warehouses at once; velocity and stock combine, so a shared supplier order reflects total demand, not one bin.

The trade-off we accept

Fixed forecasting windows mean you can’t hand-pick an arbitrary sales date range for a one-off “project just last week” question. That’s deliberate: the whole point is that the buy agrees with the forecast. If you genuinely need a bespoke window, that’s a forecasting question — set it there, and the PO follows.