Supply chain network design
Your warehouses and distribution centres were planned for the customers and volumes of years ago. We redesign your supply chain network for the business you run today.
- Free 30-minute call
- No commitment
- You own the code
In short: A supply chain network design decides how many plants, warehouses and distribution centres you need, where they are and which customers each one serves. It looks for the set-up with the lowest transport and site costs that still delivers to your customers as fast as you promise. It works with today’s demand and with several ways demand could develop.
The problem
Your supply chain network is your plants, warehouses and distribution centres and the transport between them. It was planned years ago, for the volumes and customers of that time. Since then, more customers order online, in smaller parcels and with shorter delivery deadlines. You buy more from suppliers nearby, in Europe instead of Asia for example, and spread your purchasing across several countries. Some large customers have moved, and their goods now travel further. The warehouses are still where they always were.
Those old decisions still set your costs today, but they only partly fit your business. Every quarter brings avoidable transport cost and duplicated stock, and some customers wait longer than you promised. Supply chain network design fixes the structure behind those costs. It looks at where your sites are, whom they serve and how goods flow between them.
Signs your network no longer fits
Your trucks drive long distances to customers, even though one of your other warehouses is close by. The same items are stored in several warehouses at once and tie up money there. Some warehouses are full, others half-empty. Some customers wait longer than you promised, because they are served from a warehouse far away. And every year finance asks whether you really still need all your warehouses. You can’t answer with numbers, because nobody has worked out what one warehouse fewer would cost or save.
Why comparing two options isn’t enough
The questions are easy to ask and hard to answer properly. Should six distribution centres (DCs) become four? Where should the next one go, and how big should it be? Which customers should each site serve? Should you run the warehouse yourself or buy space from a third-party logistics provider (3PL)?
Each of these decisions changes total landed cost and delivery times at the same time. Landed cost is everything it costs to get goods to the customer. Close a distribution centre, for example, and you save rent, but your trucks drive further. A spreadsheet that compares two options side by side can’t see the third option that beats both.
What changes
Today, you compare two or three network options in a spreadsheet, one assumption at a time. Our model weighs every site, every customer and every transport route in one calculation. You see which sites to keep, open or close, and which customers each one serves. And you see how well that design works if volumes rise or a big customer moves.
What you get
We replan your network. We start from where your customers are and how much they order, what your sites and transport cost, and what you promise on delivery, such as next-day delivery. One model decides everything together: which sites you run, which site serves which customers and which routes the goods take. Then we test the design against the demand scenarios that matter to you, such as higher volumes or a new large customer. A simulation checks it before you invest money or sign a lease.
- A model written around your own supply chain network design rules.
- Results measured in the numbers you already track.
- It slots into how you already work, and you can check every answer it gives.
Why a centre-of-gravity calculation isn’t enough
The classic first step is a centre-of-gravity calculation. It finds the point from which the distances to your customers are shortest in total. Customers who order large volumes count for more. In optimisation, this is called the Weber problem. It’s a fine sketch for a first conversation, and we’d use it that way too.
For a decision, it leaves too much out. It measures straight lines, which in Switzerland can run through a mountain or across a lake. It places one site at a time and knows nothing about rent, labour, capacity or the fixed cost of opening a site. Our model starts from the candidate sites you could actually lease or buy, with road distances and the costs each site carries.
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.
DC location optimisation
Underneath, you’re solving a facility location problem: where to put sites so demand is served at the least cost. The p-median model opens a set number of sites, the p in its name, and keeps the volume-weighted distance to customers as short as possible. The p-centre model instead keeps the farthest customer as close as possible.
Fixed-charge models add a fixed cost for every site you open. In practice you need this version, because a DC costs rent and staff whether or not it’s full. On top of location, the model adds capacity limits and covering constraints, rules that every customer is reached within a set service time. The model plans several years one after the other. So you don’t size the network for the volumes of one single peak year.
Why mixed-integer optimisation
The engine underneath is mixed-integer optimisation. Some decisions are yes or no: open this DC, use this lane. Others are quantities, such as how much flows through each lane. Combining both lets the model keep rules like these. A hub, where goods change trucks, only pays off above a minimum volume of goods. And a DC only serves customers within a set distance.
No calculation with averages can keep those constraints. Mixed-integer linear programming (MILP) shows you what each rule costs. For every proposed network, you see which rule is binding, meaning it limits the plan, and what it would cost to relax it.
Testing the design against many futures
Nobody knows the demand of the next few years exactly, so we test every design against several scenarios. Multi-scenario analysis takes the plausible developments, such as much higher volumes, demand that moves from one region to another or a new large customer. It solves the network for each one, then ranks the candidates by how well they work across all of them. You end up with a network you’d still be happy with if the year surprises you.
Optimisation produces the design, and simulation checks it. Network simulation runs a design with your data from the past few years. It also runs Monte Carlo scenarios: many random but realistic demand patterns. You watch cost, delivery times and warehouse use move over time, before you commit.
Where distribution network optimisation matters
For manufacturers, a plant closure or a new production line rewrites the whole flow from plant to DC to customer. For the wider supply chain, the network sets which costs and delivery times you can reach at all. Stock levels and replenishment are planned inside those limits. Put the sites too far from your customers, and even the best inventory planning can hardly make up for it. Warehouse network design software that only looks at one decision at a time can’t fix that.
Common questions.
Something else? Ask us directly.
There’s no fixed number. It depends on your costs and on how fast you promise to deliver. More sites mean shorter last-mile trips and faster delivery. They also mean more rent, more staff and more stock, because each site holds its own buffer. With fewer sites it’s the other way round: lower fixed costs, but longer trips and slower delivery. Our model works out the cost for every number of sites and every possible location. You see which set-up costs the least in total and still delivers as fast as you promise.
A year or more of shipments by customer location, with volumes and weights. Transport rates or invoices for your main lanes. The costs and capacities of today’s sites and of any candidates. And your service promises, such as next-day delivery to a region. It doesn’t need to be perfect. Where data is missing, we use clear assumptions and show how much the answer depends on them.
It’s a good first sketch, but not a decision. It finds the point from which the straight-line distance to all your customers is shortest in total. Customers with large volumes count more. It ignores roads and mountains, and it ignores rent, wages, capacity and the fixed cost of a site. It also works out only one site at a time, so it can’t show how several warehouses work together. Use it to start the conversation, then test the sites you could lease or buy, with what each one costs.
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