Stock-outs vs. idle cash: setting inventory where the money is

Ask most businesses what service level they target and you'll get one number, often something like "we aim never to run out." It sounds responsible. In practice, one target applied everywhere is one of the most expensive policies a business can have. It treats a slow, cheap item in a quiet location the same as a fast, high-value item in your busiest one, and it allocates cash accordingly.

We learned this running a currency supply chain across many locations, where every note held is idle cash and every shortage is a lost customer. The same trade-off applies in retail, distribution and any business with stock spread across sites. The goal isn't to never run out. It's to run out in the places where it costs least, and almost never where it costs most.

Why one number doesn't work

The cost of holding more stock rises steeply as you approach full availability. Moving from good availability to very good is usually affordable. Moving from very good to near-perfect can need far more stock for very little extra service. Apply a near-perfect target across the whole range and most of your inventory ends up protecting items and locations where an occasional shortage would barely be noticed.

Meanwhile the cost of a stock-out isn't the same everywhere. It depends on:

• Value and margin: some items earn far more per sale than others.
• Customer behaviour: will they wait, switch to an alternative, or go to a competitor?
• Knock-on effects: does a missing item lose the whole basket, transaction or relationship?
• Location role: a flagship site, airport or key account often matters more than its sales alone suggest.
• Replenishment speed: a shortage that's fixed tomorrow costs less than one that lasts a week.

Segment before you set targets

A practical starting point is to group products and locations by what matters, then set service levels by group. This builds on the long-standing ABC classification approach and on Marshall Fisher's argument, in What Is the Right Supply Chain for Your Product?, that predictable and unpredictable products need different treatment. A simple version looks at three things.

1. Value: how much revenue or margin the item carries.
2. Volume and predictability: how much it sells and how stable that demand is.
3. Criticality: what a stock-out really costs, including the customer impact.

High-value, high-criticality items in important locations get the highest targets. Low-value items that customers will happily substitute get lower ones. Erratic, slow items may not be worth stocking everywhere at all, and can be held centrally and moved on request.

An illustrative example

This example is illustrative. Picture a network with a uniform availability target. Analysis shows that a small share of products and locations generate most of the margin, while a long tail sells slowly and is often substituted. Rebalancing means a higher target for the top group, a moderate one for the middle and a lower one for the tail, with some tail items centralised. In a case like this, total inventory can fall while availability on the items that matter rises. The service level customers notice gets better, and the cash tied up gets smaller.

The point isn't the specific numbers. It's that cash moves from where it's idle to where it earns.

Location matters as much as product

The same item can deserve different targets in different places. A busy location with fast replenishment may need less buffer than a remote one with infrequent deliveries, even at a similar level of demand. In currency, a store in a major airport and a quiet regional branch need very different holdings of the same currency. Setting targets by product alone misses this. Setting them by product and location together is where most of the gain is. Hau Lee's The Triple-A Supply Chain makes a related case for supply chains that adapt to local conditions rather than running one uniform policy.

Making it stick

Differentiated service levels can feel uncomfortable. Someone will point out that an item ran out, and the instinct is to raise its target. That's why the policy has to be agreed in advance by commercial, operations and finance, with clear reasons for each segment. Review it regularly as products, locations and demand change, rather than one complaint at a time. Report availability by segment too, so a planned, low-cost shortage in the tail isn't treated like a failure on a key item. These are core planning disciplines in ASCM's CPIM certification.

It also depends on a forecast you can trust. Safety stock covers uncertainty, so better forecasts at location level reduce the stock needed for every target you set. Getting there starts with measuring forecast accuracy properly.

Examples marked illustrative are hypothetical. Third-party figures are from the sources linked.