Safety stock and reorder point
Your safety stock was calculated on assumptions that no longer fit. We recalculate it from how much your demand and your delivery times swing today.
- Free 30-minute call
- No commitment
- You own the code
In short: A safety stock is the extra stock you hold to cover swings in demand and supplier lead time. The reorder point is the stock level that triggers a new order, early enough that it arrives before that buffer runs out.
The problem
Ask your team where the safety stock numbers come from. Usually they were set years ago, by rule of thumb: the average demand of a few weeks, or the same percentage of demand for every item. Since then your suppliers, your customers and your demand have all changed. The figure on the screen still looks official. It’s been wrong for a long time.
A wrong buffer fails in one of two ways. It protects you from risks that no longer exist, and ties up cash in stock for demand that has settled down. Or it leaves you exposed, because demand or delivery times now swing more than when the rule was written. Either way you pay, in cash or in empty shelves.
Signs your safety stock needs a rethink
Some items have weeks of buffer and never run out. Others run out regularly, even though the system shows safety stock. Every item has the same weeks of cover, whatever its demand or its supplier. Orders go out too late for a slow supplier, so the item sells out before the delivery arrives. Or they go out so early that the delivery comes while the safety stock is still untouched. And you can’t say how often each item really runs out.
Why simple rules fail both ways
A flat “two weeks of cover” or a blanket 95 percent service target fails in both directions. Items with steady demand get too much buffer, although a stockout costs little there. Items whose demand swings a lot get too little, although a stockout hurts there. That’s exactly where buffer would pay off. The total looks reasonable, even though it’s wrong item by item.
What changes
Today, every item carries the same few weeks of cover, set years ago and never checked. With our model, each buffer is sized from today’s risk on that item. How much does its demand swing? How much do its supplier’s delivery times move? Safety stock optimisation answers both for every item, and you can see the reason behind each number.
Fixing it rarely means more stock overall. Mostly, you move buffer from where it’s wasted to where it’s needed, which can free cash and improve availability at the same time. A service level safety stock calculation, done item by item, makes that move visible.
What you get
We rebuild your safety stock and reorder points from your own demand history and your suppliers’ delivery records. Each buffer is sized to a fill-rate target, the share of demand you want to serve straight from stock. That target reflects what a stockout costs on that item. Before anything goes live, we test the policy on a holdout: past demand the model never saw.
- A model written around your own safety stock and reorder point rules.
- Results measured in the numbers you already track.
- It slots into how you already work, and you can check every answer it gives.
How it works
Forecasts, stock and decisions about your warehouse network go into one model instead of three spreadsheets. By network we mean all your warehouses, distribution centres and branches, and which site supplies which. The model plans forecasts, stock and network decisions together in a single run. So a fix in one warehouse no longer creates a new problem at another site.
Two sources of uncertainty
Safety stock absorbs two separate uncertainties. The first is demand variability: how much actual demand swings around the forecast. The second is lead time variability: how much the time from order to delivery moves around its average. They interact, because a longer lead time means more days of demand to cover and more time for demand to drift.
A probabilistic safety stock model treats both as distributions: every value each can take, and how likely it is. From both, it works out how far demand can swing in total before the next delivery arrives. That’s how big the buffer gets. Spreadsheet rules skip this step and work with single averages, so they can’t tell an erratic supplier from erratic customers.
A worked example: when the supplier is the problem
The textbook formula, which assumes the two swings are independent, is short. Safety stock is a service factor z times the square root of two terms added together. The first is lead time times the variance of daily demand. The second is daily demand squared times the variance of the lead time.
Take one item. It sells 40 units a day, give or take 10. The supplier takes 9 days on average, give or take 3. At a 95 percent cycle service level, meaning no stockout in 95 of 100 order cycles, z is about 1.64. The demand term alone asks for about 49 units of buffer. Add the late deliveries, and the answer is about 203.
So a rule that only looks at demand holds a quarter of what this item needs. Now say the supplier gets reliable, give or take one day. The buffer falls to about 82 units. Before you buy more stock to cover a shaky supplier, find out what better delivery reliability is worth. The formula tells you.
Fill rate versus cycle service level
Simple rules ignore the most important distinction: fill rate versus cycle service level. Cycle service level is the chance you don’t run out between one order and the next. At 95 percent, you run out in 5 of 100 order cycles. It says nothing about how many units you’re short each time.
Fill rate is the share of demand you serve from stock, counted in units. An item that rarely sells can have a low cycle service level and still a high fill rate, because its rare stockout costs only a unit or two. A best seller with the same cycle service level can lose many sales each time it runs out. Fill rate shows the sales you lose, and that’s almost always what you care about.
The reorder point
The reorder point is the stock level at which you reorder. It’s the demand you expect over the lead time, plus safety stock. When stock falls to that level, you order. Estimate the lead time wrongly, and you order too late or too early. Too late, and the item sells out before the new delivery arrives. Too early, and the delivery comes while the safety stock is still untouched, so you hold more stock than you need.
So the reorder point and safety stock have to come from the same demand and lead time distributions. Reorder point calculation software that treats them separately gets both wrong. We calculate them in one step, so the trigger fits the buffer.
Where safety stock fits
How much safety stock you need depends on how far your demand forecast is off. So a good demand forecast comes first. If the forecast improves, you need less buffer. A buffer of the right size must also be in the right place, such as the central warehouse instead of every branch. For that, we use stochastic optimisation. It plans against many possible weeks, the bad ones included, instead of a single forecast.
Common questions.
Something else? Ask us directly.
About 1.64, if you mean a 95 percent chance of no stockout per replenishment cycle. That’s the cycle service level. For 90 percent it’s about 1.28, for 99 percent about 2.33. If you mean 95 percent of demand served from stock, the fill rate, the z-value is different and usually lower. It also depends on your order quantity, so check which target you mean first.
Add a second term under the square root. Safety stock is z times the root of lead time times demand variance, plus average demand squared times lead-time variance. The second term can easily dominate. An item selling 40 a day with deliveries swinging by three days needs far more buffer than its demand alone suggests. Measure lead times from your goods receipts, not from the contract.
Whenever its inputs move, and at least every quarter for most items. Check your best-selling items and items with changing demand every month. Recalculate straight away when a supplier changes, a lead time jumps or a product enters a new season. A safety stock you set years ago and never checked again usually no longer fits today’s demand.
Sometimes it can, and if that’s the case, we’ll tell you. Standard software is built for an average operation, though. Yours runs on its own rules: your contracts, your exceptions, the things you know by heart. A custom model writes those rules in, and that’s where the better plan comes from.
You don’t need either to start. We build the model from the data you’ve already got, even if it lives in three spreadsheets and one person’s head. You get a system you run yourself. We train your team until they no longer need us.
You see a first result with the prototype, before you commit to anything bigger. We start with a free 30-minute call about your operation and the model that would fit it. Then we build a working prototype and run it on your numbers.
Similar pages.
Terms on this page
Every abbreviation explained in plain words, in our article “Metrics and methods”.
Tell us what you’re planning.
In a free 30-minute call with a specialist, you find out whether optimisation can help.
- Free, no commitment
- You talk to a specialist directly
- A clear go or no-go