Skip to content
Shyenova

Inventory

Why people systematically under-order high-margin products

The mathematics of a single-period stocking decision is a hundred years old and unambiguous. Experienced buyers still get it wrong in the same direction, every time.
Published
11 December 2025
Reading time
6 minutes
Topic
Inventory
Author
Shyenova

The newsvendor problem is the cleanest decision in inventory theory. You order once. Demand is uncertain. Order too little and you forgo margin; order too much and you carry the cost of the excess. There is no replenishment and no second chance.

It applies to considerably more than newspapers: fashion seasons, perishable food, event merchandise, spare parts for an end-of-life product, vaccine batches, festival stock. Anything where the ordering decision is made once and the demand arrives afterwards.

The answer, and why it surprises people

The optimum is determined by the critical fractile, the ratio of the cost of under-stocking to the sum of the costs of under- and over-stocking. Order up to the quantity at which the probability of meeting demand equals that ratio.

The consequence is worth stating plainly, because it is the part that cuts against intuition. For a product with a high margin and low disposal cost, the critical fractile is high (often above ninety per cent) which means the optimal order quantity sits far above expected demand. Deliberately, and by a lot.

MarginSalvage recoveryCritical fractileOptimal order
70%30%0.921,422
50%30%0.831,286
30%30%0.681,140
30%0%0.30843
15%0%0.15689
Optimal order quantity for demand normally distributed with mean 1,000 and standard deviation 300.

A seventy-per-cent-margin item with reasonable salvage value should be ordered at over forty per cent above expected demand. Very few buyers do this. Most order somewhere close to the forecast, and a substantial minority order below it, reasoning that excess stock is a visible error while a lost sale is not.

The pull-to-centre effect

This bias has been studied extensively in behavioural operations research, and it is remarkably robust. Given a newsvendor decision, people order between the mean demand and the mathematical optimum, pulled toward the centre of the demand distribution regardless of which direction the optimum lies in.

For high-margin products, where the optimum is well above the mean, people under-order. For low-margin products, where the optimum is below the mean, people over-order. The bias is not conservatism; it is an attraction to the expected value, which feels like the safe answer and is not.

The error is not random and it does not average out. It is directional, it is persistent, and it is largest precisely where the stakes are highest.

It survives experience. It survives training in the formula. It survives feedback across many rounds. When the decision is run with practitioner groups, players who can state the critical fractile correctly still order below it, and continue to do so after seeing the results of the previous eleven rounds.

Why organisations amplify the bias

Individual psychology is only part of it. Most organisations add incentive structures that push the same way.

  • Excess stock is measured, reported and attributed. Lost margin from a stockout is invisible unless someone deliberately measures it, and almost nobody does.
  • Write-offs require explanation to finance. Under-ordering requires no explanation to anyone.
  • Inventory turn targets are set at portfolio level and applied uniformly, penalising exactly the high-margin items that should be deliberately over-stocked.
  • The buyer who over-orders and takes a markdown is criticised. The buyer who under-orders and sells out is congratulated on the sell-through rate.

That last asymmetry is the most damaging and the most common. A ninety-eight per cent sell-through on a high-margin seasonal line is usually presented as an excellent result. It is strong evidence that the line was under-ordered and that demand was left unserved at full price.

What actually corrects it

  1. 01Measure lost sales. Even a rough estimate (days out of stock multiplied by average rate of sale multiplied by margin) changes the conversation, because it makes the invisible cost visible and comparable to the write-off.
  2. 02Set service targets by margin segment rather than uniformly. High-margin, high-salvage items warrant a deliberately high target. Low-margin items with no salvage value warrant a low one. A single portfolio-wide number is wrong for both.
  3. 03Compute the critical fractile explicitly and put it in front of the buyer as a decision input. Simply displaying the recommended quantity alongside the forecast substantially reduces the pull-to-centre deviation.
  4. 04Review outcomes on both tails. A post-season review that examines only markdowns will reinforce the bias. Reviewing sell-outs with equal seriousness is what breaks it.

The broader lesson generalises well beyond single-period stocking. Where a decision has asymmetric and unequally visible costs, the organisation will systematically drift toward whichever error is harder to see. Making the invisible error countable is usually a cheaper intervention than any amount of training in the correct formula.

Start a conversation

Tell us what is not working.

Most useful first conversations are thirty minutes and involve no proposal. Describe the problem you are looking at and we will tell you honestly whether it is one we can help with.

Or write directly to hello@shyenova.com