Supplier lead times you can actually plan with

The number you were given is not the number

Every reorder calculation depends on a lead time. Most stores get that lead time the same way: they asked the supplier once, at the start of the relationship, and wrote down what they said.

The problem is not that suppliers lie. It is that "our lead time is three weeks" is a description of a good outcome under favourable conditions, quoted by somebody in sales, about a factory they do not personally run. It is a best case dressed as an average, and you are using it as an average.

The consequence is predictable. Your reorder point is set with a number that is achieved perhaps half the time, so you go out of stock roughly as often as the supplier is late, which feels like bad luck and is actually arithmetic.

Lead time is four numbers, not one

The quoted figure usually covers one segment of a chain with several. Break it up and the sources of variation become obvious.

• Order to acknowledgement. How long between you sending a purchase order and the supplier confirming it. Often days, occasionally weeks, and almost never counted.
• Acknowledgement to ready. The manufacturing or picking time. This is the number they quoted you.
• Ready to shipped. Waiting for a container, a collection, a consolidation. Highly variable and entirely outside their production control.
• Shipped to received. Transit, customs, and your own receiving time before the stock is actually sellable.

A supplier quoting three weeks is usually quoting the second segment. The real elapsed time from your decision to sellable stock can easily be double that, and the difference is where stockouts come from.

Measure what actually happened

The fix is unglamorous: record two dates per purchase order and let a few months pass.

Date you sent it. Date the stock became sellable. That is it. Two columns, one row per order, per supplier.

After six or eight orders you have something better than any quote: your own distribution. Not just the average, which is the least useful statistic here, but the spread. The average tells you what to expect. The spread tells you what to protect against, and protection is what safety stock is for.

The number worth planning against is not the mean. It is closer to the worst case you are willing to accept, which for most stores means something like the slowest of the last several orders rather than the typical one.

Why the spread matters more than the average

Two suppliers both average four weeks. The first delivers in four weeks almost every time. The second delivers in two weeks half the time and six weeks the other half.

On paper they are identical. In practice the first is far easier to work with, because you can hold less safety stock and still not run out. The second forces you to plan for six weeks, which means carrying weeks of extra stock and paying for it, on every line they supply.

This is worth saying out loud to suppliers, because most of them have never heard it framed this way. Consistency is worth money to you, and a supplier who understands that a reliable five weeks beats an erratic four has been given a way to be more valuable that does not involve dropping their price.

A consistent five weeks is cheaper to buy from than an unpredictable four. Very few suppliers have ever been told this.

Feeding it back into the reorder point

Once you have measured lead times, the reorder point stops being guesswork. The mechanics are in how to set a reorder point, and the short version is that you reorder when remaining cover falls below the lead time plus a buffer.

Two things change once the lead time is measured rather than quoted. The number goes up for most suppliers, which feels like carrying more stock and is actually carrying the right amount. And the buffer becomes justifiable per supplier rather than a flat guess, so the reliable ones stop being penalised for the unreliable ones.

The same logic runs through low stock alerts: an alert threshold derived from a real lead time fires when there is time to act, and one derived from an optimistic quote fires when it is already too late.

The conversation to have with a slow supplier

Data changes this conversation completely. Without it you are complaining about a feeling. With it you are describing a pattern.

Go in with specifics: the last six orders, dates sent, dates received, the range. Ask what drives the variation, because they usually know and have often never been asked. The answer is frequently something addressable, such as batching your order into a production run that happens on a fixed cycle, or a shipping consolidation that waits for other customers.

Then ask what would make it more predictable. Ordering on their cycle rather than yours. Larger, less frequent orders. A standing forecast so they can produce ahead. Several of these cost you very little and shift the variation substantially.

And ask what they need from you, because the honest answer is sometimes that your own purchase orders are late, ambiguous or repeatedly amended, which is a problem you can fix in an afternoon. The mechanics of that are in a purchase order process for a store without an ERP.

Seasonality is not variation

One thing to separate carefully. A supplier who is reliable for ten months and slow for two is not unpredictable. They are seasonal, and seasonality is plannable in a way that randomness is not.

So record the month alongside the dates. If the slow orders cluster around a factory shutdown, a national holiday period, or the run-up to peak, you do not need a bigger buffer all year. You need a different buffer for those weeks, and a reminder to order early ahead of them.

Getting this wrong in the safe direction is expensive: you carry stock for twelve months to survive two.

Lead time is not fixed, and you can move it

Stores tend to treat lead time as a property of the supplier, like their address. Quite a lot of it is a property of the arrangement, and arrangements are negotiable.

Order on their cycle. Many suppliers produce in batches on a fixed schedule. An order that arrives two days before a run gets made in that run; one that arrives two days after waits for the next. Knowing the cycle can remove a week without asking anyone to work faster.

Give a rolling forecast. Even a rough one, updated monthly, lets a supplier hold materials or produce ahead. Forecasts you are prepared to be held to are worth more than accurate ones you keep private.

Order less often, in larger quantities. This trades working capital for predictability, which is a bad trade for some stores and a good one for others, but it is a trade rather than a fact.

Ask about the shipping leg separately. A large part of the variation on imported goods is consolidation waiting rather than production. Sometimes paying for a dedicated collection removes more time than any production change would.

None of this requires leverage. It requires knowing which segment the time is actually going into, which is why the four-part breakdown earlier is worth doing before you negotiate anything.

Who owns this, and where it lives

The measurement has to belong to someone or it stops after three entries. It is a small job, a minute per purchase order, and it is worth naming rather than assuming.

More importantly, the resulting numbers have to be somewhere the person placing the next order will see them. A lead time analysis living in a file nobody opens changes nothing. The number belongs next to the decision it informs, which in practice means in the room where purchasing gets discussed, updated when an order lands.

The same applies to what you learn about a supplier's behaviour: that they always ship late in August, that they respond faster to one contact than another, that their acknowledgement takes a week. That is institutional knowledge, and when it lives only in the buyer's head, it leaves when they do.

What this looks like in Store Huddle

A purchasing room where each purchase order is a task with a due date set from the measured lead time, not the quoted one. The task closing when stock is received gives you the two dates without anyone maintaining a separate spreadsheet, because the record is a by-product of doing the work.

Products attached to the conversation mean the item being reordered is identified rather than described, which matters when a supplier has four variants with similar names. And a note pinned in the room about a supplier's real behaviour is readable by whoever places the order next month, including if that is somebody new.

Common questions