Safety stock calculation: bu ne anlama geliyor
Safety stock is the buffer held against demand and supply uncertainty. The problem is that most companies set it by habit rather than by formula: “let us hold about two weeks of sales”.
The standard formula and what it assumes
The most common form:
Safety stock = Z x sigma_demand x sqrt(lead time)
Z is the service level factor (1.65 for 95%), sigma_demand is the standard deviation of demand, and lead time is in days.
This formula assumes two things: that demand is normally distributed, and that lead time is constant. The second is almost never true.
When lead time varies too
In real operations a supplier delivers in 12 days sometimes and 25 days other times. Then:
Safety stock = Z x sqrt(lead_time x sigma²_demand + mean_demand² x sigma²_lead_time)
The second term is often larger than the first. Which means the real reason you run out of stock is not demand volatility but supplier unpredictability — and if you are only using the first formula, you will never see it.
Service level is a cost decision
The gap between 95% and 99% sounds small: Z moves from 1.65 to 2.33, so safety stock rises by 41%. That is the price of those four points.
The right question is not “how high do we want it” but “does a stock-out cost us more than carrying the stock that prevents it”. The answer differs per product: 98% may be right for A-class items and 85% for C-class.
One number does not fit every product
Every SKU has its own demand volatility, its own lead time and its own margin. A single “two weeks of stock” rule causes stock-outs on fast movers and dead stock on slow movers — simultaneously.
Using it with ABC classification
Safety stock decisions belong per product, but analysing every product individually is impractical. ABC classification bridges the two:
A-class (70-80% of turnover, 10-20% of items): high service level, frequent review, lead time variability measured individually.
B-class: medium service level, quarterly review, group averages are sufficient.
C-class (5% of turnover, half the items): simple rule, annual review. Here the cost of analysis exceeds the gain.
The quiet cost of dead stock
Safety stock is insurance, and like any insurance it has a premium. Carrying cost is usually 20-30% of stock value per year: capital, warehousing, insurance, shrinkage, obsolescence.
Excess safety stock produces that cost silently — no report has a line called “unnecessary safety stock”. Stock-outs, by contrast, are visible and generate complaints. That asymmetry pushes most companies into systematically over-stocking.
When supplier performance improves
The formula takes supplier variability as an input. Working with a supplier to improve delivery consistency reduces safety stock directly.
This is usually the cheapest route to lower inventory: it needs a supplier conversation, not software. But it is only visible if you measure the lead time distribution.
Service level is a decision, not a formula
The most important input to a safety stock calculation is not statistical: it is the service level you are targeting. The difference between 95% and 99% is a substantial increase in stock, and whether it is warranted varies by product.
A critical component may justify 99%. A slow-moving spare is well served at 85%. Applying one level across the catalogue is the most common way to hold too much and too little at the same time.
Not ABC, but four quadrants
Classic ABC ranks products by revenue. A safety stock decision needs two axes: value and predictability.
High-value predictable items can run lean. High-value volatile items are what safety stock exists for. Low-value volatile items are usually cheapest to simply cover. Low-value predictable items were never the problem.