Skip to content

Inventory optimisation

Too much stock ties up cash, and too little loses sales. Our model sets, for each item, how much you hold and when you reorder.

  • Free 30-minute call
  • No commitment
  • You own the code
Typical industries
Retail and e-commerce, Spare parts and MRO
Problem type
Supply chain
You receive
A documented model, run on your data

In short: An inventory optimisation sets three things for each item: how much reserve to hold (safety stock), when to reorder and how much. It weighs how much money is tied up in stock against how reliably you can supply your customers.

The problem

On Monday, you open the stock report. One warehouse is full of items that haven’t sold since spring. A branch is out of a best-selling item, and a customer order waits. The reorder rule in your ERP is the same for every item: enough stock for a few weeks. It was set years ago and hasn’t been checked since.

Inventory is the most expensive asset most businesses carry, and often the least optimised. Too much, and your cash is tied up in stock that ages until it’s obsolete. Too little, and you lose sales at the shelf or on the production line. The right answer differs for every stock-keeping unit (SKU), meaning each item you stock. One blanket rule loses money in both directions at once.

Signs your inventory policy is out of date

One warehouse is full while another runs out of the same item. More and more cash is tied up in stock, but you still can’t deliver from stock more often. Every SKU gets the same reorder rule, whatever it earns. So you hold too much of items nobody would miss, and too little of items customers ask for every day. Each branch holds its own buffer and never counts on another branch to help out with stock. Together, the branches hold more reserve than they need. Old stock sits until you have to throw it away or sell it at a loss, while best-selling items run short.

What changes

Today, every item gets the same few weeks of cover, and each warehouse sets its own stock. Our model gives each SKU its own policy: how much to hold, when to reorder and how much to order. Important best sellers get more buffer, and cheap items that rarely sell get very little. The model puts the buffer where it protects the most, say in the central warehouse instead of in every branch.

What you get

We build the model on your own demand history and your SKU hierarchy, the way your range is grouped. A stockout means an item has run out. You tell us how much that hurts for each item, and the model weights it that way. Before anything goes live, we test the policy on a past period of demand that we kept out of the build. The test shows how much cash the policy frees up before you commit to it.

  • A model written around your own inventory optimisation 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.

The full replenishment policy per SKU

Most teams reach for safety stock first, the buffer kept against surprises. It’s only one input. The full replenishment policy per SKU answers three questions. How much do you hold, when do you reorder, and how much do you order? Day to day, three numbers carry out the policy: the safety stock, the reorder point, which is the stock level at which you reorder, and how often you order.

The textbook answer for order size is the economic order quantity (EOQ). Order rarely and in bulk, and holding costs rise. Order often and in small amounts, and ordering costs rise. EOQ finds the quantity in between. But it assumes demand never changes. It doesn’t know how much room your warehouse has or which minimum quantity your supplier asks for. A dynamic order quantity adapts to rising and falling demand, to seasons and to your suppliers’ minimum order quantities.

Service levels by SKU criticality

SKU criticality belongs in the policy as an input. Not every SKU deserves the same service level, meaning how reliably you can deliver it from stock. In ABC terms, a ranking by revenue, A items earn the most and C items the least. Say an A item drives a third of revenue. It should get more buffer and a higher fill-rate target, a bigger share of its orders shipped straight from stock.

A C item that rarely sells runs lean. Even with an occasional stockout, the cash it frees up beats the rare lost sale. Intermittent SKUs sell only now and then. The model takes your demand forecast for them as an input. One common way to make it is Croston’s method. It estimates separately how often an item sells and how much sells each time. More stock means fewer stockouts. You settle this service level inventory trade-off for each SKU, not once for the whole range.

What one more point of service level costs

Safety stock grows faster than the service level you ask for. In the textbook model, it’s a service factor times the spread of demand over the lead time. That factor is about 1.28 for a 90 percent chance of getting through a cycle without a stockout. It’s 1.64 for 95 percent and 2.33 for 99 percent.

Take an item whose lead-time demand swings by a standard deviation of 100 units. It needs about 128 units of buffer at 90 percent, 164 at 95 and 233 at 99. The last four points cost almost twice as much buffer as the five before them. Put every SKU on 99 percent, and you pay that steep end of the curve on items nobody would miss.

Multi-echelon inventory optimisation

Most companies run multi-echelon networks: stock held at several levels, such as a central warehouse feeding regional ones. Planning each location on its own doesn’t work here, because the levels depend on each other. A unit of safety stock is worth more at the central level. There, it can cover demand swings for any location below it. Spread thinly across every branch, it covers far less.

Multi-echelon inventory optimisation recognises that stock one level up can replace stock one level down. It places the buffer where one unit of reserve prevents the most stockouts. If every location optimises for itself, the system holds too much. Solved as one model, the network can reach the same service level with less cash tied up.

The MILP model behind it

The method is mixed-integer linear programming (MILP). Some decisions can’t be split: you either stock an item at a warehouse or you don’t. The rest are ordinary quantities. Here, the model decides three things: where you hold stock, which service level each SKU gets and how much warehouse space it uses. The goal is to keep the cost of holding stock plus the cost of stockouts as small as possible. Every answer must also keep your rules. One of them: each branch gets its goods only from the warehouse that supplies it.

It’s the same class of model that solves scheduling and routing problems. So inventory optimisation software built this way uses the same method as routing and scheduling models. You don’t add one more spreadsheet that is connected to nothing. You can question the model and change it as your business moves.

Retail, e-commerce and spare parts

Inventory optimisation pays off most where a stockout costs more than one sale. In retail and e-commerce, the classic failure is the wrong stock in the wrong store, so the optimisation works per location as well as per SKU. In spare parts and MRO, short for maintenance, repair and operations, one missing part can stop a production line. There, a part matters more the longer the production line stands still without it.

Common questions.

Something else? Ask us directly.

There’s no single right number, and that’s the point. Buffer grows steeply near the top: going from 95 to 99 percent needs far more stock than going from 90 to 95. So set high targets where a stockout stops a production line or loses a key customer. Let cheap, slow, replaceable items run lower. The model then shows what each target costs in stock.

Your ERP’s material requirements planning (MRP) works out what to order. It uses the values you type in: how much reserve to hold (safety stock), at what stock level to reorder (reorder point) and how much to order (lot size). It doesn’t tell you whether those parameters are right. Inventory optimisation sets them, item by item, from your demand, lead times and service targets. Your ERP keeps running the orders, now on better numbers.

When you hold stock at more than one level, and one level feeds the other. A central warehouse supplying regional ones is the classic case. Then safety stock in the central warehouse covers the demand swings of all regional warehouses. If you plan each site on its own, you hold too much stock. With one warehouse and direct deliveries, a single-site model is enough, and we’d tell you so.

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.

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