Efficient or responsive? Match the supply chain to each product

Sport Obermeyer's forecasts of skiwear demand could be wrong by as much as 200 per cent. Campbell Soup shipped more than 98 per cent of its demand straight from stock (Fisher, 1997, p. 108). Both ran supply chains, and Marshall Fisher (1997) argued that they needed opposite ones. If you sell a steady line and a line of new designs through one chain, you are probably serving one of them badly. I read the strategic fit matrix in Fisher's own article and in chapter 2 of Chopra's textbook, tried it on an invented coffee roaster, and say below where it breaks.

What does Fisher's matrix say?

Many broken chains come from a mismatch: functional products need an efficient chain, innovative products a responsive one (Fisher, 1997, pp. 106, 109). According to the site's framework summary, the two mismatched cells are an efficient chain for an innovative product (stockouts on hits, mark-downs on misses) and a responsive chain for a functional product (paying for capability that goes unused).

Fisher gives a supply chain two jobs. The physical job is making, moving and storing goods. The market-mediation job is matching the volume reaching the market to what customers want to buy; its costs arise when supply exceeds demand and goods are sold at a loss, or when supply falls short and orders are lost (Fisher, 1997, p. 107). Functional products mostly incur physical costs. Innovative products mostly incur market-mediation costs.

Is your product functional or innovative?

Classify by the shape of demand, not by what the product is. Fisher says clothes, ice cream, coffee, cookies and computers can all be sold in a basic form or an innovative one (Fisher, 1997, p. 108). The table gives the levels he found typical.

Contribution margin is price minus variable cost, divided by price (Fisher, 1997, p. 107). Fisher himself calls these levels guidelines based on what he found typical (Fisher, 1997, p. 109), not cut-offs. Use the table as a question: for most rows, which column do your numbers lean towards?

Why is an efficient chain so costly for an innovative product?

Because the big loss sits in stockouts, not in unit cost. Fisher takes an innovative product with a 40 per cent contribution margin and a 25 per cent stockout rate: the contribution lost is 40% × 25% = 10 per cent of sales, an amount that usually exceeds pre-tax profit. For a functional product, 10% × 1% is only 0.1 per cent (Fisher, 1997, p. 110). From this he concludes that each dollar invested in responsiveness for an innovative product usually returns more than a dollar (Fisher, 1997, pp. 109-110).

The two firms from the opening show both ends: Campbell has only 5 per cent new products a year, Sport Obermeyer 95 per cent new designs (Fisher, 1997, p. 108). Early orders from Obermeyer's 25 largest retailers brought its forecast error down to about 10 per cent (Fisher, 1997, p. 115). Fisher writes that this approach halved the cost of overproduction and underproduction and raised profits by 60 per cent (Fisher, 1997, p. 116), and he cites his own 1994 co-authored article for the case, so it is not independent evidence.

What should a small coffee roaster do with two product lines?

Invented example. Rang Hẻm Sáu is a four-person roastery in Ho Chi Minh City with one 15 kg roaster. The name and all figures below are my own invention. Line one is a house blend, "Sáu Đậm": 38 cafés and offices reorder weekly, weekly sales stayed between 180 and 220 kg for 12 months, and revenue is about VND 2.5 billion a year. Line two is Tet gift boxes: three new designs a year, sold in the five weeks before Tet, about 1,500 boxes at VND 400,000 each with a variable cost of VND 240,000.

Today both run through one efficient chain: one big roasting run a week, a three-month green-bean contract with the cheapest supplier, boxes packed in December on last year's forecast. Last year design A sold out in nine days. Design B sold less than half of what was made, and the rest went at a discount in February.

I read Fisher's "stockout rate" as the share of demand left unmet; the article does not define it. The gift box has a quarter of the blend's revenue (VND 600 million against 2.5 billion) yet loses six times as much to stockouts. Fisher names three ways to cope with uncertainty: reduce it, avoid it and hedge what remains. He also advises accepting it rather than forcing sharper forecasts: for innovative products, "uncertainty must be accepted as good" (Fisher, 1997, p. 114).

Holding 300 kg of green beans for three months, with interest and shrinkage assumed at VND 4 million, is set against 60 million of contribution that could be lost. The two lines share one roaster. Chopra calls this tailoring, as Zara makes trendy items in Europe and basics in Asia (Chopra, 2019, p. 44); the post on Uniqlo and Zara looks at the two firms in more depth. If you outsource the fast-moving part, add the buffer stock that slower delivery forces, as in the invented example in the post on Uniqlo and outsourced factories. If the boxes sell through several channels, implied demand uncertainty rises with the number of channels (Chopra, 2019, p. 38), so a single stock figure comes first.

This suits you if you can read pre-orders early and have room for green beans. It does not suit you if the box is the same design every year and always sells 1,500: it is then a functional product and the buffer is dead money. Fisher also warns that products drift from functional to innovative without managers noticing (Fisher, 1997, pp. 108-109).

Are lean and agile the same framework?

Close, not identical. The site's summary of lean versus agile supply chains puts staples, with predictable demand, in the lean chain, and fashion goods, with volatile demand, in the agile chain, citing Mason-Jones, Naylor and Towill (2000) and the Slack et al. (2022) textbook. I read only the 2000 paper's abstract: it confirms the main idea, but I could not check the site's detailed table (Mason-Jones et al., 2000).

The authors add a hybrid they call leagile: a different paradigm on each side of the decoupling point in the material flow (Mason-Jones et al., 2000). According to the library, that is where real orders enter the flow; upstream of it runs on forecasts. In the example, green beans and plain boxes sit upstream; the printed sleeves and the roast-to-order batches sit downstream. That mapping is mine.

Where is the framework weak?

Two types is a simplification. Fisher says products are "primarily functional or primarily innovative" (Fisher, 1997, p. 106), so he knew some are mixed. Chopra replaces the two types with a spectrum of uncertainty and adds supply uncertainty: coffee has steady demand but uncertain supply because of weather, so it sits mid-spectrum (Chopra, 2019, p. 39). The example ignored the supply side.

Fisher's table of figures rests on experience: the article names no sample or method, and its cases are selected ones.

Uncertainty is partly your own promise. Chopra separates demand uncertainty from "implied demand uncertainty", the part the chain must absorb given how you serve customers: "raising the service level increases the implied demand uncertainty even though the product's underlying demand uncertainty does not change" (Chopra, 2019, p. 37). Promising same-day delivery of coffee turns a steady seller into a hard one to serve.

Products change type, and the chain has to follow. In Chopra's pharmaceutical case, new drugs use costlier flexible capacity and mature ones move to dedicated capacity (Chopra, 2019, p. 45); a two-by-two grid cannot draw that movement. He also gives the example of two suppliers for one product: a cheap one for the predictable part, a fast one as back-up (Chopra, 2019, p. 49).

Finally, a mismatch may be plain waste. Chopra draws a cost-responsiveness frontier: a firm off the frontier can improve both at once, and only a firm on it must trade one for the other (Chopra, 2019, p. 40).

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

For Fisher (1997) I read the full Harvard Business Review reprint (reprint 97205), pages 105 to 116, from a PDF on the course-materials server of an online university in Mexico, because the HBR page cited in the library gives only a summary. The site's figure flips the vertical axis of the print version: Fisher puts the efficient chain on the top row, so his "upper right-hand cell" is bottom right in the site's figure.

For Chopra (2019) I read chapter 2 (pages 33 to 50) of the seventh Global Edition, in the preview Pearson publishes; I have not read other chapters.

For Mason-Jones and colleagues (2000) I read only the abstract on the Cardiff Metropolitan University research page.