Reorder points have a particular way of going wrong. Someone calculates them carefully at implementation, they’re correct for perhaps six months, and then they stay exactly where they are while the business around them changes. Three years later they’re driving purchasing decisions based on a demand pattern that stopped existing.
The Static Problem
A conventional reorder point combines expected demand over the lead time with a safety buffer. Both inputs are entered as fixed values and both are wrong within months.
Demand shifts as products mature, customers change, and markets move. Lead times drift as suppliers get busier or improve. Variability changes as your customer mix changes. None of this is visible in a static field, and there’s rarely a process that catches it — reorder points get reviewed when something goes wrong, which means after the cost has been incurred.
The other static weakness is uniformity. Applying the same policy across a class of items ignores that individual items within it behave completely differently.
What Adaptive Calculation Does
It recalculates continuously from current data rather than holding a fixed value.
Demand over lead time comes from a forecast that reflects recent movement, seasonality, and trend rather than a historical average. Lead time comes from the measured distribution of that supplier’s actual deliveries, not the quoted number. Variability is measured directly and recently. Safety stock derives from the service level you’ve set for that item’s importance rather than from a blanket rule.
So an item whose demand has been climbing for three months gets a rising reorder point automatically. A supplier whose deliveries have become erratic triggers a larger buffer without anyone noticing and intervening. A seasonal item’s reorder point moves ahead of its season rather than after it.
Guardrails Worth Having
Fully automatic adjustment can misbehave, and a few constraints prevent it.
Rate-limit the movement. A reorder point that doubles overnight on one unusual week creates its own problems. Damping changes to a sensible percentage per period keeps the system stable.
Set floors and ceilings for critical items, so no calculation can drive a line-stopping component below a minimum you’re comfortable with.
Flag large changes for review rather than applying silently. A recommendation to triple a level is either an important signal or a data problem, and both deserve a look.
Exclude known anomalies. A one-off bulk order shouldn’t permanently reset a reorder point, and the system needs a way to be told that.
The Measurable Difference
The gain isn’t only lower inventory. It’s that stock ends up in the right places. Static systems typically over-stock declining items and under-stock growing ones simultaneously — so total inventory looks reasonable while service suffers and dead stock accumulates. Adaptive levels fix both directions at once, which is why the service and inventory numbers can move together rather than against each other.
ticktick.ai recalculates reorder points continuously per item with rate limiting, floors, and review thresholds on significant changes.
