Safety Stock and Reorder Points: How Much Inventory to Keep Sitting at Your 3PL

There are two ways to get inventory levels wrong, and they fail in opposite directions.

Hold too little and you stock out: lost sales, a marketplace listing that loses ranking while it is unavailable, and customers who buy the substitute and stay there. Hold too much and you pay storage on goods that are not moving, tie up cash that the business needed elsewhere, and eventually face markdowns.

Safety stock and reorder points are how you pick a deliberate position between the two instead of discovering it by accident.

Start with the four inputs

Everything below rests on four numbers, and getting rough versions of all four beats getting one of them precisely right.

• Average demand. Units sold per day, across all channels, over a period long enough to be representative.
• Replenishment lead time. Not the factory's quoted production time — the full clock from raising the purchase order to the stock being pickable at your warehouse, including transit, customs and receiving.
• Variability. How much both of the above swing. A supplier averaging 30 days but ranging from 20 to 55 is a different planning problem from one that is reliably 30.
• Target service level. The share of demand you intend to fill from stock. This is a business decision, not a statistical one.

The reorder point, plainly

A reorder point is the on-hand quantity at which you place the next order. It has two parts:

Reorder point = (average daily demand × lead time in days) + safety stock

The first part covers what you will sell while you wait. The second covers everything that can go wrong during the waiting.

Worked through: if you sell 40 units a day and replenishment reliably takes 45 days, you will consume roughly 1,800 units while the order is in flight. Order at 1,800 and you are betting that nothing goes wrong. Safety stock is the size of that bet.

Sizing safety stock without a statistics degree

The textbook formula multiplies a service-level factor by the standard deviation of demand over lead time. The service factors are the part worth knowing:

Two things fall out of that table. First, the cost of coverage is not linear — moving from 95% to 99% costs about 40% more safety stock for four percentage points of protection. Second, 100% is not on the table at any price.

If the statistics are more than you need today, a defensible approximation is to hold enough to cover the difference between your average lead time and your worst realistic lead time. For the example above, a supplier that occasionally takes 60 days rather than 45 needs about 15 days of cover — 600 units. That is not optimal, but it is a considered number rather than a guess, and it can be refined later.

Set the level per SKU, not per catalogue. Your top ten items deserve a high service level; the long tail rarely justifies the carrying cost.

Where multichannel quietly breaks the maths

The formulas assume one demand stream drawing on one pool of stock. Most brands no longer have that.

Inventory split across a marketplace's own network, a 3PL warehouse and a retail account is three pools with three lead times and three demand patterns. A national average that looks healthy can hide a stockout in the pool that is actually selling. Safety stock has to be calculated per location, and the transfer time between locations is itself a lead time that needs cover.

This is the arithmetic that makes multichannel fulfilment harder than running one channel well — not the integrations, the inventory maths underneath them.

Review it on a cadence, not on an incident

Reorder points decay. Demand moves, suppliers speed up or slow down, and a number set in January is wrong by June. Three habits keep them honest:

• Recompute average demand monthly from a trailing window, excluding known one-offs like a promotion.
• Re-measure actual lead times quarterly against what your system assumes; the gap is usually larger than expected.
• Raise cover deliberately before peak rather than reactively during it, which is exactly the kind of thing a pre-peak outbound checklist exists to force.

What to expect from your fulfilment partner

A 3PL cannot set your service level — that is a margin decision only you can make. But it should give you the inputs: accurate on-hand counts you can trust without a physical audit, dock-to-stock timings that let you measure receiving as part of lead time, and alerting when a SKU crosses its reorder point rather than a report you have to remember to open. Getting those three from your provider is part of what a functioning e-commerce operations relationship should look like.

That alerting is the piece most often missing. AIDWMS watches on-hand positions against per-SKU reorder points across every location and raises the flag while there is still lead time left to act — worth a look at the platform if your current answer to "are we about to run out?" is a spreadsheet someone refreshes on Mondays.