Safety stock

Without it you plan for an average week, and an average week is not the one that empties your shelf.

A rough way to work it out

Take the busiest week of sales you have had for that line in the last year. Take a typical week. The gap between them, multiplied by your lead time in weeks, is a sensible buffer.

Sell 20 a week normally, 35 in a good week, two-week lead time. The buffer is 30 units.

It is not exact. It is much better than a flat number applied to everything.

Vary it by line

Hold more where being out hurts: your fast sellers, anything with a long lead time, anything a customer would go elsewhere for rather than wait.

Hold less on slow lines. A buffer on something that sells one a month is dead stock with a friendly name.

Where it goes wrong

The buffer is set once and never revisited. Sales change, suppliers get slower, and a number chosen two Christmases ago is still sitting there.

The other fault is treating safety stock as untouchable. It exists to be eaten into during a bad week. A buffer you never dip into is not protection, it is capital sitting still, and the stockout it was meant to prevent never happens because the line was overstocked the whole time.