Choosing a plant or warehouse site: five criteria and a break-even

A paint manufacturer (hypothetical; I am not describing any real company) has two plants, in Long An and Binh Duong, serving 15 dealers across the Mekong Delta. Each plant ships directly to each dealer, usually on part-loaded trucks. Someone proposes renting a warehouse in Can Tho. Where to put the warehouse, or whether to do nothing at all, is the question for this post.

In what order should a location decision go?

Location funnel: from strategy to site has four phases, narrowing as it goes (Chopra, 2019). Phase one sets competitive strategy and the supply chain design: what to do in-house, what to outsource, and whether to grow by acquiring, building or partnering. Phase two chooses the regional configuration: demand by region, tariffs and tax incentives and the response time wanted, which together say which regions, how many facilities and what capacity. Phase three picks a set of potential sites in each region, looking at hard infrastructure such as transport and utilities, and soft infrastructure such as labour and community attitudes. Phase four fixes the site by estimating demand and cost at each one.

Chopra makes two cautions: companies competing on cost often pick the cheapest place even if it is far from the market, while companies competing on responsiveness place facilities near customers, even where it costs more. Plants tend to stay for ten years or more, whereas warehouses the company does not own can change within a year (Chopra, 2019). The post on efficient or responsive supply chains helps settle phase one. The paint maker in the example sells steady goods, so phase one leans towards cost, but dealers complain about slow delivery.

What are the five groups of criteria?

Five groups of location criteria lists five groups (Slack et al., 2022, ch. 5):

The most memorable part is the caution that goes with it. Slack and colleagues name two pitfalls: looking at one site only, and forgetting the option of staying put. They advise always keeping "do nothing" as the base case, because the cost of moving and the risk of inconveniencing customers are high, and a location decision is hard to undo once taken (Slack et al., 2022, ch. 5). In the example, option A is to stay put and ship directly from the two plants.

Counting lanes: direct shipping or through a warehouse?

Direct shipping or distributor warehouse compares two networks. Direct shipping from the manufacturer keeps stock pooled in one place, so inventory is low and availability high, but transport is expensive because the manufacturer is far from the customer and orders with several sources arrive in several shipments. A distributor warehouse needs more inventory because pooling is lost, but far less than a network of retail stores; full truckloads go into the warehouse and orders are bundled into one delivery (Chopra, 2019). It also raises facility cost a little and does not suit very slow movers. Chopra notes that most companies combine several networks rather than choosing one.

A simple count shows why. If each plant ships directly to each dealer, m plants and n dealers need m × n lanes. Through one warehouse there are m lanes in and n lanes out: m + n. With 2 plants and 15 dealers, direct shipping is 30 lanes and the warehouse route is 17. This is my own count, valid when every plant serves every dealer; it is not a figure from the book.

Fewer lanes does not mean cheaper: a warehouse has fixed costs. Hypothetical numbers for the paint maker: direct shipping costs VND 600,000 a pallet because trucks run part-loaded. Through the warehouse, full trucks go in (VND 150,000 a pallet) and consolidated orders go out (VND 290,000), a total of VND 440,000. The rented warehouse costs VND 90m a month, plus VND 25m for capital tied up in extra stock and labour: VND 115m fixed. Each pallet through the warehouse saves VND 160,000, so break-even is VND 115m ÷ 160,000 ≈ 719 pallets a month.

At 900 pallets a month the warehouse saves about VND 29m a month. But a 20% fall in volume, to 720, wipes out the saving. So the question is not "is the warehouse cheaper?" but "what is the lowest volume I trust enough to commit to a lease?" Low, unpredictable volumes are a different case: Chopra says direct shipping suits high-value items with low and unpredictable demand, where customers are willing to wait (Chopra, 2019).

What happens when you score the five groups and change the weights?

Three options: A, stay put and ship directly; B, rent a warehouse in Can Tho; C, rent a warehouse in Long An, near the plants. Scores run from 1 to 5, higher is better, and I set them. The cost scores for B and C assume volume around 900 pallets a month, above the break-even:

Two sets of weights (per cent, in the order of the five groups): set one is 20, 25, 25, 15, 15; set two is 35, 10, 25, 15, 15, for a firm that is short of capital and not pushed by customers on speed.

Set one picks B and set two picks A, from the same scores. That is what scoring is really for: not to produce the answer but to expose disagreement. If the sales director and the finance director give different weights, the meeting should argue about the weights, not about each score. The library gives the five groups but no weights or scale; the scoring is my method, not Slack's or Chopra's.

Going deeper: what did I read, and what did I not?

I did not read Chopra (2019) or Slack and colleagues (2022, chapter 5) directly. The four-phase funnel, the cautions on plants lasting ten years and warehouses changing within a year, the comparison of direct shipping with a warehouse, and the five criteria groups come from the framework summaries already published on the site. The m × n versus m + n count, the 719-pallet break-even, the score sheet and both weight sets are mine, recomputed in code. Every figure for freight cost, rent, pallets and scores is an assumption; do not copy them into your own plan.

The biggest limit: the example treats the freight cost per pallet as fixed, while in practice it moves with the season, fuel prices and load size. It also ignores the taxes and incentives of each province, which Chopra lists in phases two and four. The break-even also uses one average cost per pallet. If the warehouse changes the service level as well, for instance next-day delivery to dealers who now wait three days, that benefit is not captured by cost alone, which is where the market-factors group in the score sheet matters. A lease longer than the period over which you trust the volume turns the break-even into a commitment, which is why the future-flexibility group matters too. On the same supply chain theme, see Choosing a supplier: the seven Cs and total cost.