Supplier lead times: why variability matters more than the average

Ask a buyer for a supplier's lead time and you'll usually get one number, often the one the supplier quoted when you first started working together. That number drives every reorder point you set. It's also frequently wrong, and even when the average is right, it hides the thing that actually causes stockouts: how much the lead time moves around.

Key takeaways

• Measure lead time from your own purchase orders: the date you sent the order to the date the goods were available to sell.
• Track the spread as well as the average. Two suppliers averaging 33 days can need very different safety stock.
• Lead-time variability is multiplied by your daily sales, so it hurts fast sellers most.
• Don't delete slow orders as "outliers". Find out why they were slow, and plan for the causes that will happen again.
• Reducing variability is often cheaper than holding the safety stock it requires.

Measure it properly

A lead time needs a defined start and end. Choose them once and apply them to every order.

Start: the date the PO was sent to the supplier. If you pay a deposit before production starts and suppliers routinely wait for it, you could start the clock at payment instead, but then you need to plan payment timing too. Sent date is the simplest and most honest.

End: the date the goods were received and available to sell. Not the ship date and not the date the container reached port. Stock in a 3PL's receiving queue still can't be sold.

Partial deliveries: use the date the bulk of the order arrived, or record both first and final receipt. Be consistent.

Then record it for every PO. A spreadsheet with PO number, supplier, sent date, received date and the difference in days is enough to start.

A worked example: same average, different supplier

Here are the last eight lead times, in days, for two suppliers making similar products.

Averages. Supplier A: 28 + 31 + 35 + 29 + 42 + 30 + 33 + 36 = 264, and 264 ÷ 8 = 33 days. Supplier B: 20 + 45 + 25 + 50 + 22 + 40 + 30 + 32 = 264, so also 33 days.

On the average alone, these suppliers are identical. They aren't.

Spread. The standard deviation measures how far a typical order lands from the average. For Supplier A, the differences from 33 are −5, −2, 2, −4, 9, −3, 0 and 3. Squared, they're 25, 4, 4, 16, 81, 9, 0 and 9, which sum to 148. Dividing by one less than the number of orders gives 148 ÷ 7 ≈ 21.1, and the square root of that is about 4.6 days.

For Supplier B, the differences are −13, 12, −8, 17, −11, 7, −3 and −1. Squared: 169, 144, 64, 289, 121, 49, 9 and 1, summing to 846. Then 846 ÷ 7 ≈ 120.9, and the square root is about 11.0 days.

Supplier B's lead time is more than twice as variable.

What variability does to safety stock

When both demand and lead time vary, a standard safety stock formula is:

safety stock = z × √(LT × σd² + d² × σLT²)

where d is average daily demand, σd is the standard deviation of daily demand, LT is average lead time in days, σLT is the standard deviation of lead time, and z is the service-level factor (about 1.65 for 95%). Our safety stock guide explains the formula in more detail.

Take a product selling d = 10 units a day with σd = 4, bought at a 95% service level.

Supplier A (LT = 33, σLT = 4.6):

• Demand term: 33 × 4² = 33 × 16 = 528
• Lead-time term: 10² × 4.6² = 100 × 21.16 = 2,116
• Total: 528 + 2,116 = 2,644; √2,644 ≈ 51.4
• Safety stock: 1.65 × 51.4 ≈ 85 units

Supplier B (LT = 33, σLT = 11):

• Demand term: 528, as before
• Lead-time term: 100 × 11² = 100 × 121 = 12,100
• Total: 528 + 12,100 = 12,628; √12,628 ≈ 112.4
• Safety stock: 1.65 × 112.4 ≈ 185 units

Same product, same average lead time, and Supplier B needs 100 more units of safety stock. The reorder points are 10 × 33 + 85 = 415 units for A and 10 × 33 + 185 = 515 units for B. At a unit cost of $12, that's 100 × $12 = $1,200 of extra stock permanently sitting on the shelf, for one product.

Notice where it comes from. With Supplier B, the lead-time term (12,100) dwarfs the demand term (528). The lead-time term is multiplied by d², so doubling a product's sales quadruples it. Variable suppliers are most expensive for your best sellers.

Why not just plan on the slowest delivery?

A common shortcut is to set the reorder point using the longest lead time you've seen. For Supplier A, the slowest order took 42 days, so 10 × 42 = 420 units. For Supplier B, the slowest took 50, so 10 × 50 = 500 units.

Those land close to the formula's 415 and 515 here, and the shortcut is far better than using the average alone. It has two weaknesses. It ignores demand variability, so a busy month during a slow delivery still catches you out. And it depends on a single order: one freak delivery sets your stock level for years, while a supplier who has simply never been unlucky yet looks safer than it is. Use it as a sanity check on the formula, not a replacement.

Don't throw away the slow orders

It's tempting to look at Supplier A's 42-day order and call it an outlier. Before you do, ask why it was slow.

• Predictable causes such as factory closures over Lunar New Year, peak shipping season or annual stocktakes will happen again. Don't delete them; plan for them by ordering earlier at those times of year.
• One-off causes like a port strike or a mis-shipped container might reasonably be excluded, but write down why.
• Causes on your side, like a late deposit, approved artwork sent late or a receiving backlog at your warehouse, are lead time too. The supplier didn't cause them, but your stock still arrived late.

How many orders do you need?

With two or three orders, any spread you calculate is mostly noise. Until a supplier has five or six received POs, use their quoted lead time plus a generous buffer, and treat the measured figures as an early signal rather than the answer. Keep updating as orders come in; recent orders matter more than ones from two years ago, especially if the supplier, route or product has changed.

Track on-time as well

Lead time and reliability are related but different. A supplier who quotes 45 days and reliably delivers in 45 is easier to plan around than one who quotes 30 and delivers anywhere from 20 to 50. Record the promised date on each PO and count how many arrived on time and in full. If 5 of 8 orders did, that's 5 ÷ 8 = 62.5%, which is a useful number to bring to a supplier review.

Reducing variability

Holding more stock is one fix. Often there are cheaper ones:

• Order earlier for known disruptions rather than holding extra stock all year.
• Ask for confirmed ship dates on every PO, and chase the day a date passes.
• Split long journeys into stages (production, transit, customs, receiving) so you can see which one varies.
• Fix your own receiving. If deliveries sit for days before they're put away, that's variability you control.
• Talk to the supplier with data. Showing eight orders and their dates is far more persuasive than "you're always late".

Where Stokrio fits

Stokrio measures lead times from the sent and received dates on your purchase orders and shows them next to what the supplier quoted. Its replenishment safety stock covers both demand and lead-time variability at your service level, and the inventory reports include supplier on-time in-full and lead-time variability. Even if you never use it, keeping the eight-row table above for each of your suppliers will change how you set reorder points.