Supply Chain

Spare parts: when the long tail runs the business

Mohit Saini · 14 Jul 2019 · 4 min read

Spare parts: when the long tail runs the business

For a company that sells industrial equipment (forklifts, compressors, generators, construction machines), nothing damages its reputation faster than a customer’s machine standing idle for want of a spare part. Customers remember a long breakdown far longer than they remember a good price.

So the company’s service arm commits to support every machine in the field. That creates a familiar tension:

  • Uninterrupted service means stocking even parts that fail rarely.
  • Stocking thousands of rarely used parts ties up a lot of cash.

And every new model adds hundreds more parts to support. For equipment makers with modest numbers of machines in the field, most parts sell only occasionally. This is the long tail, and it dominates the spares business.

The usual answer: shorten lead times

Most service organisations try to resolve this by cutting replenishment lead times: faster suppliers, quicker transport, more stocking points closer to customers. The logic seems sound. The stock needed at a location (its norm) covers peak demand during the replenishment time. Shorter replenishment time, smaller norm, less stock, so more range can be stocked for the same money.

Yet many companies find that even after they succeed in cutting lead times, stock doesn’t fall much. Why?

How norms really behave for slow parts

Recall how a norm is usually set: look at past sales in rolling windows equal to the replenishment time, and take the highest total in any window. For a 15-day replenishment time, compare total sales in every 15-day window over the past year and take the maximum.

For a fast-moving part that sells every day, this works as expected. Halve the replenishment time, and the peak window total roughly halves. The norm falls steadily as lead time falls.

For a slow-moving part, something different happens. Let us take a simple example. A part sells 4 units at a time, about once every couple of weeks, and the shortest gap ever seen between two sales is 10 days.

  • With a replenishment time of 10 days or less, any window can contain at most one sale. The peak window total is 4 units.
  • With 11 to about 20 days, a window can catch two sales: 8 units.
  • And so on.

So for slow parts, the norm doesn’t fall smoothly with lead time. It moves in steps. Within a step, cutting lead time changes nothing. If the replenishment time is already shorter than the minimum gap between sales, cutting it further can’t reduce the norm at all. The part is already at its lowest possible stock: one sale’s worth.

Let us call the shortest gap between two sales of a part, at a given location, its minimum time between sales. A part whose replenishment time is already below that gap is at its base norm.

A counter-intuitive consequence

If a part is at its base norm at a location, then lengthening its replenishment time (up to the minimum time between sales) won’t raise its norm either.

Suppose a part sells at a dealer at most once every 30 days, and today it is replenished from a regional warehouse 7 days away. It could just as well be replenished from the central warehouse, 15 days away, without the dealer needing any more stock. Since 15 days is still below 30, the dealer’s norm stays at one sale’s worth.

But now the regional warehouse doesn’t need to stock that part at all.

Simplify the network

Parts usually move more slowly the further they are from the centre. A part that moves daily at the central warehouse may move every ten days at a regional warehouse and once a month at a dealer. In long-tail businesses, many parts become base-norm items at the outer locations.

Most companies, by long habit, replenish every part from the immediately preceding node: central to regional to dealer. Slow parts then sit at every level, turning very slowly.

A better rule:

  • Identify base-norm parts at each location.
  • Supply them directly from the most aggregated location where they move faster than their delivery time, often the central warehouse, as long as that delivery time stays below the part’s minimum time between sales at the destination.
  • Remove them from intermediate stocking points.

Customer service at the point of use doesn’t change. Stock across the network falls. The cash released can be used to stock more of the long tail, improving service further.

The money involved is not small. Picture a network with six regional warehouses, each stocking about 2,000 parts that turn out to be at their base norm at the dealers it serves. If each such part holds, on average, ₹2,000 of stock at a regional warehouse, that is ₹4 million per warehouse, ₹24 million across the network, sitting still. Supply those parts directly from the central warehouse, and most of that cash is released without a single dealer holding more stock or waiting longer than the part’s usual gap between sales.

For slow parts, faster isn’t always better. Sometimes the best move is to skip a warehouse altogether.

It is a good example of what Goldratt called inherent simplicity: a network with more nodes looks more sophisticated, but often only adds stock and complexity. Reviewing the assumptions behind every stocking decision regularly can free up a surprising amount of money.

Related reading

Mohit Saini

Strategy, supply chain & sales transformation consultant. About · Book a discussion