Safety stock and lead time: how much inventory to hold when routes get longer

01Cycle stock, pipeline stock and safety stock

Inventory has different jobs, and it helps to separate them:

When a shipping route gets longer, as when Asia–Europe services were diverted around the Cape of Good Hope from late 2023, pipeline stock increases automatically. Safety stock increases only if variability increases, but in practice it almost always does, because disrupted routes are less reliable.

02The safety stock formula

The most widely used formula assumes that daily demand and lead time vary independently and roughly follow a normal distribution:

SS  = z × √( L × σd²  +  d² × σL² )
ROP = d × L + SS

d   = average daily demand          σd = standard deviation of daily demand
L   = average lead time (days)      σL = standard deviation of lead time (days)
z   = service factor for the target cycle service level
ROP = reorder point: place an order when stock on hand + on order falls to this level

The first term under the square root is demand uncertainty during the lead time; the second is the uncertainty created by the lead time itself, scaled by how much you consume per day. If lead time never varied, the second term would vanish and you would get the classic z × σd × √L.

03Service level and z

The cycle service level is the probability of not running out during a replenishment cycle. Each level corresponds to a z value from the normal distribution:

The cost rises steeply at the top. Choosing the service level per item, high for critical inputs that stop production, lower for items with substitutes, is one of the cheapest ways to reduce inventory without increasing risk.

04Calculate it

Try the calculator with your own data, then increase the extra days to see what a longer route does.

With the default values, demand variability alone would need around 340 units of safety stock, but lead time variability adds much more. That is the general pattern for long sea routes: reliability matters more than speed.

05What a longer route really costs

Suppose a route goes from 35 to 47 days, and its standard deviation from 6 to 9 days, because of a diversion. Three things change:

1. Pipeline stock grows by 12 days of demand. You pay for these goods earlier and hold them at sea longer.
2. Safety stock grows because both L and σL grow. The σL effect is usually the larger one.
3. Order timing shifts: the reorder point moves up, so the first orders after the change must be larger to rebuild cover. This is when shortages typically happen.

Working capital is the hidden cost. At an 8% annual cost of capital, every extra day of stock held costs about 0.02% of its value per day, which adds up quickly across many items and weeks.

06Measuring lead time variability properly

Most companies know their average lead time; few measure its spread. To estimate σL:

• Use the actual dates: order or departure date to receipt at your warehouse, not the supplier's promised date.
• Collect at least 20 to 30 deliveries per supplier and route if possible.
• Separate normal variability from disruptions. A rare 3-week delay is better modelled as a separate event than averaged into σL; that is where Monte Carlo simulation helps.
• Update after any route change, such as a chokepoint closure.

07When the formula is not enough

The normal-distribution formula works well for steady items with many deliveries. It underestimates risk when:

• Lead times have long tails. Disruptions make the distribution skewed, so a normal σL understates the late cases. Simulate instead.
• Demand is intermittent. Low-volume spare parts need other distributions, such as Poisson.
• One supplier can stop entirely. Safety stock covers variability, not a stoppage. For that, compare time to recover and time to survive in our single-supplier risk guide.
• Multiple components must all arrive. The probability that all are available is lower than for any single one; plan service levels at the product level.

08A practical routine

1. Classify items by criticality and volume; set a service level per class.
2. Measure d, σd, L and σL from the last 6 to 12 months.
3. Compute SS and ROP per item and compare with today's settings.
4. Re-run the calculation for the routes at risk with the lead time and variability of the alternative route.
5. Decide in advance which items get strategic stock if risk signals rise, and how much cash that requires.