How to calculate average daily usage (ADU)

In short: past ADU is the usage over a rolling window of past days divided by the number of days in the window, counting days with no usage as zero. Forward ADU does the same with forecast or known future demand, and blended ADU weights the two. A shorter window reacts faster but is noisier. A demand adjustment factor scales ADU for a planned change such as a promotion.

Past ADU

Past ADU = total usage in the last N days ÷ N

Usage means what actually left the buffer: shipments or sales for a finished item, issues to production for a component. Divide by every day in the window, including days with zero usage. Dividing only by days that had usage overstates ADU for slow movers.

A worked example

A made-up item used 20 units a day for 60 days, then 30 a day for the most recent 30 days. Its buffer uses a 10-day decoupled lead time, medium factors (50% and 50%), no MOQ and a 7-day order cycle, so with these settings the top of green works out to 24.5 × ADU.

The same item gets a buffer anywhere from 572 to 858 units depending only on how ADU is calculated: a 50% spread. The 90-day window still remembers the slower months, while the 30-day window has already caught up with the higher rate.

Choosing the window

• Shorter windows react faster to real changes in demand but also chase noise, so buffers move around more.
• Longer windows are steadier but lag: after demand steps up, the buffer stays too small for a while.
• A rolling window of around 90 days is a common place to start. Check how often items fall into red or sit above green, and adjust from there.
• Recalculate on a schedule (daily or weekly) so the window keeps rolling forward. Stale ADU is one of the main reasons buffers end up too lean or bloated.

Forward and blended ADU

Forward ADU uses forecast or known future demand over a coming window. It suits new items without history, planned ramp-ups or ramp-downs, and seasonal items heading into their season. Its weakness is the forecast itself: a biased forecast becomes a biased buffer. Blended ADU weights past and forward, for example 50/50, to temper either one. Whatever you choose, use the same method across items with similar behavior so buffer status stays comparable.

Demand adjustment factor (DAF)

A demand adjustment factor multiplies ADU for a known, temporary change: a promotion, a product launch, a planned phase-out or a seasonal ramp. With a DAF of 1.3, the 90-day ADU of 23.33 becomes 23.33 × 1.3 = 30.33. Apply the factor ahead of the change by about one lead time, so the extra stock arrives in time, and remove it when the change ends.

Practical checks

• Units must match the lead time. With lead times in calendar days, use calendar-day ADU. With working days, divide usage by working days.
• One-off orders distort history. A single large, unrepeatable order inflates ADU for the whole window. Many teams leave it out of the history and treat it as a spike when it comes.
• Intermittent items (many zero days) get a small ADU but lumpy demand. Watch their variability category, not just ADU.
• Short history: if an item has less history than the window, the missing days count as zero and ADU is understated. Use forward ADU or a manual ADU until enough history builds up.

The free planner calculates past ADU from imported usage history with a window you set, accepts a manual ADU for new items, and applies a demand adjustment factor per item. To see how ADU feeds the zones, try the buffer zone calculator. For what happens after the buffer is sized, see net flow position.