Rate-Of-Return (RoR) is the ultimate metric for Supply Chain
In my book Introduction to Supply Chain, I argued that the rate of return – RoR – should be regarded as the ultimate metric for governing supply chain decisions. I presented it there as one theme among many. Here, I want to focus on this idea alone: to explain it in plain terms, to show how it differs from the mainstream view, and to suggest what changes once you take RoR seriously.
I will keep the discussion self-contained. You do not need to have read the book - although you should!
What are we really trying to optimize?
If you open a standard textbook or slide deck on supply chain management, you will find a surprisingly consistent definition. Supply chain management is usually described as the integration of suppliers, factories, warehouses and stores so that goods are produced and distributed in the right quantities, to the right locations, at the right time, in order to minimize total system cost while satisfying service requirements.1
Frameworks like the SCOR model turn this vision into a large catalogue of metrics. SCOR organizes performance around five attributes – reliability, responsiveness, agility, cost and asset management – and then defines hundreds of associated indicators.2 Balanced Scorecard approaches do something similar from a different angle: they recommend monitoring the supply chain through four perspectives – financial, customer, internal process, learning and growth – each with its own list of KPIs.3
The result is familiar: dashboards filled with service levels, OTIF, forecast accuracy, inventory days, cash-to-cash cycle, transportation cost per unit, warehouse productivity, CO₂ per shipment, and so on.4
There is nothing wrong with these metrics as such. They describe the health of the system from many angles. The problem is that they do not, taken together, tell you what you should actually do with the next unit of scarce resource – the next dollar of capital, the next pallet position, the next truck, the next hour of production capacity.
Should you put that dollar into extra stock of Product A or Product B? Should that last pallet go to Store X or Store Y? Should you pay for air freight or wait for the boat? Should you accept a low-margin promotion that ties up capacity for several weeks?
A dashboard full of metrics will describe the consequences of the choice once it has been made. It rarely tells you, in a single number, which option is better before the fact. That is the role I assign to the rate of return.
A simple definition of rate of return
Let me strip the idea down to its essentials.
Whenever we commit resources in the supply chain – when we buy inventory, book capacity, send a truck, open a warehouse, grant a credit line to a customer – we are effectively making a small investment. We commit an amount of capital now and we hope to get it back later, with a surplus.
The rate of return is simply:
“Net gain” is the profit generated by that decision in coins (whatever your currency). “Capital committed” is the amount of capital tied up because of this decision – in inventory, capacity, receivables, etc. “Time” is how long that capital remains tied up.
The units are telling. RoR is measured in “per unit of time”: per year, per month, per day. In other words, RoR is a speed. It captures how fast a given decision turns capital into more capital.
A positive RoR means that the decision creates value. A negative RoR means that it destroys value. Between two feasible decisions, the one with the higher RoR is better, economically, than the one with the lower RoR, once you adjust for risk.
This is a very old idea in finance, but it is surprisingly underused in supply chain, where we often look at margins and costs without fully accounting for time.
Why the frequency of returns dominates everything else
Consider a very simple example.
Suppose you have two SKUs, A and B, with similar prices and similar margins per unit. If you only look at margin percentage, they look equally attractive.
Now add time. A rotates every month; B rotates once a year.
Even if the margins are equal, SKU A gives you twelve opportunities per year to earn that margin on the same pool of capital, while B gives you only one. Once you compound the rotations, A will generate a much higher rate of return than B.
Retailers already glimpse this through metrics like GMROI (Gross Margin Return on Inventory Investment): they look at how many units of gross margin they get per unit of inventory investment over a year. But GMROI is usually computed at an aggregate level, and typically without integrating the broader supply-chain context.
The key point is that differences in frequency are huge. Across a large catalog, you routinely find flows whose effective annual rate of return differs by one or two orders of magnitude. Some items barely cover the cost of capital; others, if you could feed them enough capacity and shelf space, would generate enormous returns.
Once you see this, many debates in supply chain look different. Shaving a few percentage points off transport costs may be worth less than increasing the rotation speed of the right items by a modest amount. Shortening a lead time may pay off not because “customers like it” in the abstract, but because it lets you recycle capital two or three times more per year on the same assortment.
Time enters the picture not as a KPI to “reduce lead times”, but as the denominator of the fundamental metric.
RoR as the single objective
From this perspective, the objective of supply chain becomes very simple to state:
Allocate scarce resources – capital, capacity, time, attention – to the admissible decisions that offer the highest expected rate of return, once risk has been taken into account.
Everything else is secondary.
Service levels matter because stockouts damage revenue, margins, and reputation; when this damage is translated into money, it lowers the rate of return of understocked decisions. Inventory days matter because they change how long capital is tied up; longer cycles reduce RoR. Transport cost matters because it reduces net gain; but if faster or more expensive transport increases the speed and reliability of profitable rotations, it can actually increase RoR.
This is what I mean when I say that RoR is the ultimate metric. It does not mean that we stop measuring anything else. It means that all other metrics are subordinate: they matter only to the extent that they improve or degrade our long-run rate of return on constrained resources.
Implicitly, this is already how markets judge us. Over the long run, investors look at how efficiently a company converts capital into profit, year after year. I am simply advocating that supply chain practitioners adopt the same lens inside the firm.
How this confronts the mainstream view
The mainstream view, as I said earlier, tends to organize itself around a wide family of KPIs: cost, service, reliability, agility, asset utilization, ESG, and so forth. SCOR and Balanced Scorecard frameworks give this approach a respectable scaffolding.
What they do not do is elevate one metric above the others as the unique objective function. Instead, they invite trade-offs: increase service while keeping cost “under control”; improve agility while maintaining asset efficiency; balance financial and customer perspectives.
In practice, these trade-offs are handled through politics, habit and intuition rather than a strict economic calculus. The transport department fights for lower freight cost per unit. Sales pushes for higher service levels and more promotions. Finance looks at working capital. Sustainability wants lower emissions. Each has its own metrics and often its own local optimum.
RoR cuts through this tangle.
Take service level. In a RoR-centric view, you do not optimize service level “as high as possible”. You estimate, as honestly as you can, the economic impact of a stockout: lost gross margin, future lost sales, brand damage, penalties. You express this as a cost in coins per unit of shortage. Once you do that, service level becomes a consequence of an optimization problem whose objective is RoR.
The same applies to sustainability or resilience. If a disruption threatens catastrophic losses, investing in redundancy or buffering is perfectly rational – but it should be treated as an option that has a cost today and a probability-weighted benefit over time. Again, RoR gives you one coherent yardstick for comparing such investments with more mundane uses of capital.
The mainstream KPI view is therefore descriptive and plural. The RoR view is prescriptive and singular.
What changes in day‑to‑day decisions
You may reasonably ask: does this change anything concrete, or is it just a philosophical preference?
Let me illustrate with a few recurring patterns.
Imagine you have scarce inbound capacity – say, a limited number of containers you can afford to bring in this month, given your cash constraints. You have a long list of candidate SKUs and quantities you would like to import.
The usual approach is to fill the containers against a mix of rules: minimum stock cover, forecasted demand, service constraints, perhaps some category targets negotiated with buyers. When you are done, you measure the expected service levels and costs resulting from the plan.
A RoR-centric approach is different. For each candidate unit you might import, you estimate its expected net gain over the horizon that matters (sales margin minus expected obsolescence, discounting, handling, etc.), and the amount of capital and time that unit will tie up. You then fill the containers starting from the units with the highest expected rate of return, moving down from there until you hit your capacity or cash limits.
Quite often, you will find that some “strategic” items that look good in terms of margin or brand positioning have miserable RoR once you account for their slow rotation and high risk of markdown, while modest items that quietly fly off the shelves are starved of capital.
Or consider the choice between regular transport and a more expensive but faster mode. In a KPI world, you might see this as “cost up, service up” and weigh it qualitatively. In a RoR world, you ask a more precise question: does the faster mode, after deducting its extra cost, increase the rate at which this flow converts capital into profit? If the answer is yes by a comfortable margin, you should choose it even if it makes your transport cost KPI look worse.
The same reasoning applies all the way down to the smallest unit of decision. The last unit of free shelf space, the last slot in a delivery route, the last hour of overtime: each should go where the expected RoR is highest.
Uncertainty is not an afterthought
Supply chain lives under uncertainty: demand fluctuates, suppliers slip, lead times stretch, promotions misfire, competitors react.
RoR does not magically remove uncertainty, but it forces you to deal with it honestly. The “net gain” and “time” in the RoR formula are not fixed numbers; they are random variables. You never know exactly how fast a product will sell or what discounting you might need later.
Instead of chasing a single “best forecast” and then optimizing as if it were certain, a RoR-centric mindset encourages you to think in terms of distributions: a range of possible futures, each with its own probability and financial outcome. What matters is the expected rate of return, adjusted for the risk of unpleasant tails.
This has two practical consequences.
First, it refocuses attention away from abstract forecast accuracy and toward economic impact. Improving forecast accuracy by 5% on a slow-moving, low-margin product may be mathematically satisfying but economically negligible. Improving it slightly on a fast, high-margin flow with tight capacity constraints may be worth a lot. RoR gives you a way to rank such efforts.
Second, it makes “waiting” a legitimate decision. Sometimes, the best thing you can do with your capital is nothing yet: wait for more information, let demand reveal itself, avoid committing to inventory whose RoR is currently ambiguous. In many organizations, “do nothing” is not perceived as a decision at all; under a RoR lens, it is often the benchmark against which all other options should be compared.
RoR is demanding – and that is a feature
There is a reason why RoR is not the dominant language of supply chain today: it is demanding.
To apply it seriously, you need a reasonably coherent view of how your decisions translate into economic outcomes. You must decide, explicitly, how much a stockout costs, how to value obsolescence, how to account for the long-term benefits of better service or greener operations. You must face the fact that different departments may have conflicting preferences and that not all of them can be reconciled without putting numbers on the trade-offs.
Dashboards with many KPIs are attractive partly because they avoid these painful conversations. Everyone gets “their” metric and can claim success locally. RoR, by contrast, is ruthless: it exposes local optimizations that destroy value globally.
Yet this is precisely why I believe RoR deserves to be put at the top of the hierarchy. Supply chain is, at its core, applied economics. We decide, thousands of times per day, how to deploy scarce resources in a world of uncertainty. Pretending that this activity does not ultimately answer to a single economic yardstick does not make that yardstick disappear; it only obscures it.
Bringing RoR into your own practice
I am not suggesting that you throw away your dashboards and replace every KPI overnight with a RoR gauge. That would be unrealistic and unwise.
What I am suggesting is a shift in mental posture.
When you look at your metrics, try to see them as instruments on a cockpit, not as goals in themselves. Ask, for each: if I push this metric up or down, how does it affect the long-run rate of return on our scarce resources? When you design a new process, a new planning rule, or a new system, ask: does it help us systematically allocate capacity, capital, and time toward the options with the highest RoR?
Over time, you can move more and more decisions – starting with the ones with the largest financial impact – under this way of thinking. You do not need to compute a perfectly precise RoR for each micro-decision; even an approximate, order-of-magnitude comparison is often enough to reveal which flows are visibly superior and which ones are quietly destroying value.
The mainstream view has given us a rich vocabulary for describing supply chains. What I am arguing for is a sharper principle for governing them. In a competitive market, the real scoreboard is coins per unit of capital per unit of time. We might as well make that explicit and let it guide the way we design, measure, and automate our supply chains.
