The economics of a warehouse network: the U-curve, the square-root law, and break-even

11 min read

Most decisions to open a second warehouse start with a location and a single number: monthly rent. Someone finds a cheap warehouse for rent in a province, and the meeting revolves around “this location is cheap”, and a few weeks later a multi-year contract is signed. The problem is that rent is the most visible cost but often the smallest in the cost picture — and once signed, it becomes a fixed burden that is difficult to reverse. A warehouse network should not be decided based on the cheapest cost, but on the lowest total cost at an acceptable level of service.

A second warehouse rental contract is a multi-year bet, not just a cost line

The decision to open a new warehouse is a strategic, long-term, and difficult-to-reverse decision — similar to a bet on geographic demand rather than a simple operating cost. When an SME distribution business justifies a second warehouse based on rent (“this location is cheap”) or emotions (“need to be closer to the North”), they are using the wrong signal to make an irreversible decision. “Winning” in this equation is not about having the cheapest rent or the fastest delivery in every region — but about having the lowest total network cost per order while still meeting service commitments.

The difference between large enterprises and SMEs lies not in intelligence but in tools. Large distribution companies run network optimization software like anyLogistix or AIMMS to simulate dozens of warehouse placement scenarios before committing capital. SMEs only have a spreadsheet and the owner's intuition. This article doesn't discuss expensive software; it brings the same thinking that the software operates on — total cost, break-even point, demand focus — to the hands of Excel users, so a multi-year decision is made based on numbers rather than impressions of a location.

Four mistakes that make a second warehouse a permanent fixed burden

Four repeated mistakes turn a potentially profitable warehouse into a structural loss. The first is deciding based only on rent: choosing the cheapest location and missing the real reward of a new warehouse — transportation savings — or worse, discovering it barely exists in that location. The second is ignoring inventory duplication: the second warehouse quietly adds storage costs due to the safety stock that must be maintained at two points, and this hidden cost can erase the transportation savings if not accounted for.

Thirdly not calculating break-even volume: opening a warehouse requires a certain threshold of orders per month to offset the additional costs; if the service area does not generate enough orders to exceed that threshold, the loss is already programmed from the start. optimizing service over total cost: chasing fast delivery in all areas at the cost of killing profit margins, instead of seeking the lowest total cost while still meeting the committed service threshold. The common point of all four: each mistake is a missed number. The rest of the article puts these numbers back together.

Eight minimum concepts to decide on a warehouse network

A correct network decision requires eight correct concepts — no more — and each concept is a number that can be put on a spreadsheet. Total network cost is the fixed warehouse cost plus shipping cost plus processing cost plus inventory storage cost; this is the only number used to judge a network decision. Serving cost (cost per order) is the total network cost divided by the total number of orders — a standardized measure that allows comparison of different scale scenarios. Shipping fee is the unit price per order per line, and is the largest lever in the entire network. Fixed Warehouse Cost : including rent, labor, utilities, and setup allocation — the burden that a new warehouse adds every month, regardless of whether there are orders or not.

The remaining four concepts frame the decision. Service level is the rate of orders meeting SLA plus weighted average lead time — a constraint that all options must overcome before considering costs. Break-even production is the new fixed cost divided by the transportation savings per order, while break-even month is the setup cost divided by the monthly savings — two numbers that turn a strategic bet into a specific threshold. Finally, center of gravity is the weighted location that minimizes transportation: each region “pulls” the warehouse towards itself in proportion to its order volume. These eight concepts are the entire grammar of the warehouse network problem.

U-shaped curve: adding a warehouse cuts transportation costs but increases inventory and fixed costs

The dominant thought framework for the entire problem is the trade-off between three cost components. In Supply Chain Management: Strategy, Planning, and OperationChopra and Meindl point out that total logistics cost equals transportation cost plus inventory cost plus fixed cost. As the number of warehouses increases, transportation costs giảm decrease because products are closer to customers, but inventory and fixed costs tăng increase because additional facilities must be maintained and a portion of the inventory is duplicated. The sum of these two opposing trends forms a U-shaped curve — and the correct network is the one that minimizes tổngthis, not any individual component.

Inventory levels increase according to a law that few people consider. According to the square root law of inventory (Maister, 1976; the portfolio effect — Zinn, Levy & Bowersox, 1989), total safety stock is proportional to the square root of the number of warehouses. Going from one to two warehouses, safety stock increases by approximately √2 ≈ 1.41 times — less than double, so the fear of "doubling inventory" is often exaggerated; but it is a real inventory cost, repeated every month, that transportation savings must offset. This is the quantified "doubling of inventory". In terms of location, the centroid method — a classic facility location model used in anyLogistix and AIMMS — indicates that the optimal point is where the total weighted distance of transportation is minimized according to demand: place warehouses where demand pulls, not where land is cheap.

Calculate the break-even point before signing, not after regretting

Break-even analysis turns a vague strategic bet into a specific production threshold that can be verified before signing. Break-even production volume equals new fixed costs divided by transportation savings per unit; break-even time equals setup costs divided by monthly savings. These two simple divisions answer the question that most decision-makers never ask: how many units per month must the service area of the new warehouse generate for savings to offset the additional costs? If the geography of demand does not create enough production volume to exceed that threshold, the plan should be eliminated right from the spreadsheet, not after a multi-year contract has been locked in.

A few benchmarks help to calibrate intuition. Transportation is often the largest logistics cost, accounting for about half or more of total network costs (Chopra & Meindl) — so the real leverage lies in placing warehouses to shift as many transportation invoices as possible, not in cutting rent. Inventory costs are typically quoted at around 18–25% per year of inventory value; a practical model can use around 1.8% per month (≈ 21.6%/year) as a default, and make that number adjustable. Best practice is summarized in four principles: decide based on total landed cost and and service level at the same time; calculate break-even production volume before signing; locate according to demand centroid; and keep all assumptions visible and adjustable. The common mistake of SMEs is to overlook both inventory duplication and fixed costs, and never calculate the production threshold to recover capital — and that is why rental contracts become a burden that transportation savings never recover.

Why this tool has exactly those pages: each page blocks a mistake

When thinking is clear, the structure of a warehouse network decision model is no longer a list of features — it is a decision chain physicalized into pages, with each page existing to prevent one specific mistake. The table below places each page next to the framework principle it implements and the mistake it eliminates.

Summary: no page in the tool is redundant, and no page is arbitrary — each page exists vì the total cost mindset requires it, and each page blocks exactly one mistake that has turned the second warehouse into a burden. Read horizontally, this table is the decision process for the warehouse network flattened into executable steps on a spreadsheet.

Want to see this thinking in action with sample data? An interactive dashboard runs directly in your browser at canvas.beup.space/warehouse-network.

Reference: Chopra & Meindl — Supply Chain Management: Strategy, Planning, and Operation (total logistics cost & network model) · Maister (1976), Zinn, Levy & Bowersox (1989) — square root law of inventory / catalog effect · Center of Gravity model, used in anyLogistix / AIMMS.

Frequently asked questions

Should you open a second warehouse?

You should open a new warehouse when the total logistics cost per order for the new setup is lower than the current one and the service level remains the same — not just when you find a cheap rental location. The way to verify this before signing is to calculate the break-even point: if the new service area generates enough orders to cover the increased fixed costs and doubled safety stock, the new warehouse is worth opening; if not, it's a structural loss.

What does the total logistics cost of a warehouse network consist of?

It consists of four parts: transportation costs (delivering goods to customers), fixed warehouse costs (rent, labor, utilities, setup allocation), order processing costs, and inventory storage costs (frozen capital on shelves). Transportation is usually the largest cost, accounting for about half or more, and is the lever that can be moved when repositioning a warehouse. The total network cost is the only number that should be used to judge a warehouse decision.

How many orders per month are needed to break even on a new warehouse?

The break-even point is equal to the new fixed costs divided by the transportation cost savings per order. For example, an additional warehouse with high fixed costs that only saves a few thousand dollars per order will require a very large volume to break even — if the area doesn't generate that many orders, the option should be eliminated. Calculate this number on a spreadsheet before signing, don't estimate after committing.

What is the square root law of inventory?

The square root law (Maister, 1976) states that the total safety stock of a network is proportional to the square root of the number of warehouses. Going from one to two warehouses, safety stock increases by about √2 ≈ 1.41 times — more than one warehouse but less than double. Practical consequence: the fear of "doubling inventory" is often exaggerated, but the additional inventory is still a real storage cost, repeated every month, that transportation savings must offset.

Where should the second warehouse be located?

It should be placed at the center of gravity — the location that minimizes the total transportation distance, weighted by order volume, where demand areas "pull" the warehouse towards themselves the strongest. This is the classic location model used in network optimization software. The correct location is where demand pulls, not where the cheapest rental is — a misplaced warehouse will never generate enough transportation savings to break even.