Home › Writing › Too much stock, still stocking out

Why you can hold too much stock and still run out

Basel Hegazi · 6 minute read

It is the complaint I hear most often, and it sounds like a contradiction. Inventory has grown for three years, finance is asking why working capital is tied up in a warehouse, and the sales team is still calling every week about the lines they cannot get. Both things are true at once, and they usually have the same cause.

They are not two problems

The instinct is to treat them separately: a cost problem to be solved by buying less, and a service problem to be solved by buying more. Run both at once and you get what most businesses in this position already have — a range where the wrong items are overstocked and the right ones are not.

The common cause is almost always the same. One replenishment rule is being applied across a range that is not uniform. Whether it is written down as a policy or lives in a buyer's head as "keep about two months," the effect is identical: items with completely different behaviour are treated as though they behave the same way.

They do not. A staple that sells steadily every week and a slow item that moves three times a year need different treatment. Applying one rule to both guarantees you carry too much of one and too little of the other, and the arithmetic is unforgiving: the slow items are usually the majority of the range, so they absorb the majority of the cash, while the fast items are the majority of the revenue and the ones customers notice.

Segment before you do anything else

Before touching a single order quantity, split the range on two axes. How much it sells, and how predictably it sells. Volume and variability are different questions and both matter.

Variability is the one people skip, and it is the one that decides how much safety stock an item actually needs. The usual measure is the coefficient of variation — the standard deviation of demand divided by the mean. Below about 0.5, demand is stable enough to forecast. Above about 1.0, you are not forecasting, you are guessing, and the policy should reflect that.

Four segments come out of it, and each needs a different answer:

Segment	What it looks like	What to do
High volume, stable	The staples. Most of your revenue, predictable week to week.	High service target. Frequent small replenishment rather than large infrequent buys. This is where tight control pays.
High volume, erratic	Promotional lines, project-driven items, anything tied to a campaign.	Do not solve with safety stock. Solve with information — get the promotional calendar into the plan before the buy is committed.
Low volume, stable	The long tail that ticks over quietly.	Low cover, simple reorder point, minimal management attention. Automate and stop thinking about it.
Low volume, erratic	Spares, specials, one-off requests.	The honest question is whether to stock them at all. Make to order, or accept a longer quoted lead time, or delist.

That last row is where most of the trapped cash sits. Nobody ever decides to stock a slow erratic item — it accumulates, one exception at a time, and then stays because removing it feels like a decision while keeping it does not.

Safety stock is about variability, not importance

Safety stock exists to cover one thing: uncertainty during the replenishment lead time. Not how important the item is, not how much you sell of it, and certainly not how loudly the account manager argues for it.

Four inputs drive it, and only four:

• Demand variability — how much actual demand swings around the average.
• Lead time variability — how much the replenishment lead time itself swings. Usually the largest of the four, and usually the one nobody measures.
• Review period — how often you look. Reviewing monthly rather than weekly means carrying weeks of extra cover for no other reason.
• Target service level — the probability you are willing to accept of not stocking out during the lead time.

The relationship in that last one is worth internalising, because it is not linear. Moving from a 95% to a 99% service level does not cost you four percent more safety stock — it costs roughly forty percent more. That is a real trade-off, and it is a commercial decision rather than a supply chain one. Which is precisely why it should be set per segment, deliberately, rather than applied at 98% across the whole range because that number sounded responsible in a meeting.

The lead time in your system is probably wrong

This is where I find the most recoverable value, and it takes a day to check.

Most ERPs hold one number per item or supplier, entered when the record was created and rarely revisited. It is usually the lead time the supplier quoted, which is what they achieve on a good run, not what they achieve on average — and it is a single number where reality is a distribution.

Pull the last two years of purchase orders and compare promised date to actual receipt date. What you want is not the average but the spread. If the average is 45 days and the worst decile is 80, your planning parameters need to reflect the 80, because that is the case safety stock exists to survive.

For anything imported into Saudi Arabia this matters more than it does in a domestic European network. The chain runs supplier ready date, booking, sailing, arrival at Jeddah or Dammam, discharge, customs clearance, conformity documentation, and inland transport. Each has its own variability and they compound. A quoted 60 days can arrive anywhere from 55 to 110, and the planning system that assumes 60 will stock out roughly as often as reality exceeds it — which, on that distribution, is most of the time.

Fixing the lead time data alone frequently resolves a stockout pattern that had been blamed on forecasting for years.

Where to start on Monday

• Pull twelve to twenty-four months of demand history by item and calculate mean and standard deviation. A spreadsheet is enough. You do not need a planning system for this step and buying one first is a mistake.
• Plot the range on volume against coefficient of variation and mark the four segments.
• Compare promised against actual lead times by supplier. Record the spread, not the average.
• Set a service target per segment and have the commercial team agree it, because they are the ones who will feel it.
• Recalculate cover for the top segment only. Prove it works there before rolling out.
• Take the low-volume erratic segment to a delisting decision. This is where the cash is.

What it looks like when it is working

Total inventory falls while service on the lines that matter improves. Both move in the direction you want, which is the tell that the original problem really was allocation rather than quantity. The cash comes out of the tail. The service comes from putting cover where demand is actually unpredictable rather than spreading it evenly across a range that does not need it.

It is not a systems project and it does not need new software. It needs someone to look honestly at how the range behaves and be willing to make a decision about the items nobody wants to discuss.