Safety Stock vs Reorder Level: How Indian Factories Decide When to Buy

Walk into the stores of almost any fabrication or pump shop in Gujarat and ask the storekeeper when he raises an indent. The honest answer is usually some version of “when the rack starts looking empty” — and the surprising thing is that this works reasonably well for the twenty items he handles personally, and not at all for the other four hundred.

The gut-feel method fails quietly. Nobody writes down the ten lakh sitting in slow-moving bar stock, because it is not a loss, it is an asset on the balance sheet. Nobody costs the Tuesday morning when the assembly line waited six hours for a seal kit, because the men were shifted to another job. Both are real money, and both come from the same cause: one number being asked to do three jobs.

Two questions, not one

Safety stock and reorder level get used as if they mean the same thing. They do not. They answer different questions, and one sits inside the other:

• Safety stock answers how much cushion do I keep? It is the stock you hold purely for the weeks when something goes wrong — the foundry runs late, or the line runs hot. In a normal month you never touch it.
• Reorder level answers when do I buy? It is the stock figure that tells you to raise the purchase order today. It has to cover everything the line will eat while you wait for delivery, and still leave the cushion sitting underneath.

That distinction is the most common single fault we find during onboarding. A factory sets “minimum 200 pieces” meaning “never let stock go below 200”, the software raises an alert at 200, the buyer places the order that day — and the material lands twelve days later, by which time the line has eaten straight through the 200 and stood idle for three days. The number was fine. It was in the wrong field.

How big should the cushion be?

You do not need statistics for this. You need your own records and about twenty minutes.

Take a real item: an SG iron pump body at a Rajkot pump assembler. The line uses 40 a day. It comes from a local foundry. Landed cost is ₹1,450 a piece.

Ask two questions.

1. How late does the supplier actually get?

Pull the last twenty purchase orders for this item and count the days from PO to the material clearing inspection. Suppose the normal delivery is 12 days, and the bad ones took 16.

So a bad delivery costs you 4 extra days of waiting. The line does not stop for those four days — it keeps eating 40 a day. Covering them needs 4 × 40 = 160 pieces.

2. How much busier does the line get?

A normal day is 40. A busy day is 65 — that is 25 extra. Allow for a couple of busy days in a cycle and you need another 50 pieces.

Now look at those two numbers again, because this is the point of the whole article. Covering your supplier took 160 pieces. Covering your customers took 50. Roughly three-quarters of your cushion exists because deliveries slip, not because demand jumps.

That is not a quirk of this item. It is arithmetic that applies almost everywhere: a delivery that is four days late costs you four full days of consumption, one after another, with nothing to offset them. A busy day, on the other hand, is usually followed by a quiet one, and over a two-week cycle they largely cancel out. Lateness accumulates. Demand wobble mostly does not.

The number nobody measures

Since three-quarters of the cushion is there for the supplier, delivery time deserves more attention than a field in the item master. Three things go wrong with it in practice.

The quoted delivery time is not the real one. The foundry says fifteen days. Your own records say the normal is twelve and the bad ones are sixteen. Use your records — and count to the date material actually cleared incoming inspection, not the date the truck reached the gate. A consignment that failed QC and is waiting for replacement has not been received, whatever the security register says.

Delivery times are seasonal and your reorder level is not. Every Indian factory knows this and almost none of them write it down. Foundries and forging shops shut for Diwali. Steel mills work to rolling schedules that push a non-standard section out by weeks. The monsoon adds days to road freight. March is a wall for anything moving on a supplier’s books. A reorder level set on the annual average is too low for about six weeks of the year and too high for the rest.

Your own follow-up is part of the delivery time. A good share of what gets recorded as “supplier delay” is really an order that sat unacknowledged for four days because nobody chased it. That part is free to fix, and fixing it shrinks the cushion — fewer late days means fewer pieces held. This is the practical argument for a vendor portal over a WhatsApp group: acknowledgement and dispatch dates are data, and data lets you carry less stock.

The number that says “buy now”

With the cushion settled, the reorder level is one line of arithmetic:

When stock touches 700, the purchase order goes out. The 480 carries the line through the twelve-day wait, and the 210 underneath is there for the week the foundry runs late. Round up, never down.

One more figure is worth writing down: your floor, which is just the cushion itself — about 220 pieces here. Stock below that means something has already gone wrong and somebody needs to know today, not at month-end. The reorder level calculator works all of this out, along with the classic maximum and average stock levels, if you would rather not do it by hand.

How much to order

Knowing when to order says nothing about how much. Order too little and you pay to raise, chase, receive and inspect an order forty times a year. Order too much and you finance a mountain.

There is a standard way to balance those two costs against each other — the economic order quantity — and the EOQ calculator will do it for you. For this casting it works out at about 582 pieces.

You will not order 582. The foundry pours a furnace heat, so the lot is 600. A steel supplier sells full bundles; a bearing distributor has a minimum order quantity; a transporter charges the same for half a truck as a full one. Treat EOQ as a sanity check, not an instruction. Its real job is to tell you whether the 2,000-piece lot the supplier is pushing is three times more than you should be financing.

At 600 a time, this item is bought twenty times a year — roughly every fifteen working days. That is the sawtooth in the diagram at the top.

What the cushion costs

A cushion is insurance, and insurance has a premium. Holding stock costs you capital, space, insurance and the risk that the part is obsolete before it is used — for most Indian SME factories, somewhere between 18% and 25% of its value a year. Call it 22%. You can build your own rate in the carrying cost calculator.

Now compare two ways of setting the same cushion:

The second row is genuinely safer. It also costs ₹1.95 lakh a year more, on one item, to protect against a coincidence — your busiest week landing exactly on your slowest delivery — that might happen once in three years. A pump shop carries perhaps 300 purchased items. You do not need to fix all of them; the top thirty by value will be most of the money.

One quick sanity check closes the loop. This policy leaves you holding about 510 pieces on average, or ₹7.4 lakh, against ₹1.74 crore of the item consumed in a year — so you turn this stock about 23 times a year, roughly 16 days of stock. That is healthy. If your numbers work out to 4 times and 90 days, the arithmetic is telling you something the stores register never will. The inventory turnover calculator checks this in both directions.

Do not do this for all 400 items

Nobody is going to sit down with twenty POs for every line item, and nobody should. Sort the list by annual value first and spend the effort where it pays:

There is a second cut worth making. An item the line eats steadily every week suits a reorder level. An item that moves only when one particular customer’s order lands does not — you should be planning that against the order book instead. That is what MRP is for: it reads your production plan, explodes the bill of materials, and tells you what to buy for those specific jobs. Reorder levels are for the steady stuff underneath — fasteners, consumables, standard sections, common castings.

Five ways this goes wrong

Where these numbers live in OEMup

Everything above is one item worked out on a notepad. That part never changes — the arithmetic is yours, because only you know how late your foundry runs. What a system should do is hold the answer and act on it without anybody having to remember.

In OEMup, each item carries a minimum quantity on its master. Set it and the system works the reorder point out from there, then flags the item as soon as stock runs down to that level. From that flag, a purchase indent can be raised for approval rather than waiting for somebody to walk past an empty rack — which is the difference between reordering on a number and reordering on a memory.

Two other things in the same system matter for the mistakes listed above. Stock reservations hold material against the job it is committed to, so quantity already promised to a work order does not read as available. And the MRP run nets what you need against stock in hand, open purchase orders and those reservations before it suggests anything — so the plan and the reorder trigger are reading the same figures rather than arguing with each other.

That last point is worth restating, because it is mistake number two. A reorder alert on its own is only as good as the stock figure behind it. Once open purchase orders are part of the picture, the item stops re-flagging every morning between the day you ordered and the day the truck arrives.

The bottom line

Three questions, three numbers, and they are not interchangeable. How much cushion — count the late days and the busy days, and add them. When to buy — what the line eats while you wait, plus that cushion. How much to buy — whatever the supplier actually sells, checked against EOQ so you know when you are being oversold.

If you change one thing after reading this, make it the delivery time. Pull the last twenty purchase orders for your top ten items and count the days from order to goods receipt — the normal figure and the bad one. On most shop floors that single half-hour moves the cushion more than any refinement to the method, because the thing you are really buffering against was never your customers.

Related reading: The Hidden Costs of Manual Inventory for what holding stock really adds up to, MRP Explained for the planning side of the same question, Procurement Management for Indian Manufacturers for the indent-to-payment chain a reorder trigger feeds, and Capacity Planning for Small Factories for when the constraint is machines rather than material.

FAQ

What is the difference between safety stock and reorder level?

Safety stock is the cushion you keep for when things go wrong. The reorder level is the stock figure that tells you to buy now, and it is made up of two parts: everything the line will use while you wait for delivery, plus that cushion sitting underneath. Safety stock is one part of the reorder level, not an alternative to it. Set your minimum level equal to the cushion alone and you are ordering about one delivery cycle too late, every time.

How do I work out safety stock without doing statistics?

Count days. First, find how many days late your bad deliveries run, and multiply by daily usage. Second, work out how much extra a busy day eats, and allow for a couple of them. Add the two. For a casting used at 40 a day, where the foundry is normally 12 days and sometimes 16: 4 extra days × 40 = 160, plus 25 extra on each of two busy days = 50, giving a cushion of 210 pieces.

How do I calculate the reorder level?

Work out what the line uses while you wait, then add the cushion. At 40 a day with a 12-day delivery, the wait costs 480 pieces. Add a 210-piece cushion and the reorder level is 690, rounded up to 700. When stock touches 700, the order goes out.

Is safety stock protecting me from my customers or my suppliers?

In most Indian SME factories, your suppliers. In the example above, covering four late days needs 160 pieces while covering busy days needs only 50 — about three-quarters of the cushion exists because deliveries slip. A late delivery costs you a full day of consumption for every day it is late, whereas a busy day and a quiet day largely cancel out over a two-week cycle.

How much safety stock is too much?

Price it. A cushion costs roughly 20–25% of its value every year to hold. In the example above, a 210-piece cushion costs about ₹67,000 a year; covering the absolute worst case would need 820 pieces and cost ₹2.62 lakh. That ₹1.95 lakh gap is on one item. If the cushion costs more per year than a stock-out of that item would, it is too big.

What is the difference between reorder level and EOQ?

The reorder level decides when to order; the economic order quantity decides how much to order in one go. EOQ balances what it costs to raise and receive an order against what it costs to hold the stock that arrives. In practice it is a sanity check rather than an instruction, because you round it to whatever the supplier actually sells — a furnace heat, a standard bundle, a minimum order quantity or a full truck load.

How often should I update these numbers?

Monthly for your highest-value items, quarterly for the middle group, twice a year for the cheap ones — and immediately whenever a supplier changes or a delivery time shifts. A reorder level typed in when the software was installed and never revisited is the single most common inventory fault we see during onboarding.

Does the MSME 45-day payment rule affect when I should order?

Indirectly, yes. Under Section 43B(h) of the Income-tax Act, an amount payable to a supplier registered as Micro or Small is deductible only in the year it is actually paid where payment falls outside the 45-day window (15 days without a written agreement). Buying earlier than you need does not only add carrying cost — it starts that payment clock early, against stock still lying in your stores.