How Much Safety Stock Is Enough (Without Tying Up Cash)?
Too little safety stock means stockouts; too much drains cash. How to size safety stock per SKU using your real demand variability and lead times.
Safety stock is the buffer that keeps you in stock when demand spikes or a shipment turns up late. Hold too little and you stock out at the worst moment. Hold too much and you have turned working capital into a pile of boxes. The whole skill is sizing it right, and sizing it differently for each product.
Most advice hands you a formula and stops. The harder, more useful part is deciding how much you actually need per SKU, and what it costs you to get it wrong in either direction.
What safety stock is actually for
Safety stock is insurance against two specific uncertainties during your supplier’s lead time: demand coming in higher than expected, and the shipment arriving later than expected. If both demand and lead time were perfectly predictable, you would need none. They never are, so you hold a buffer to cover the gap between the expected case and a realistic bad case.
That framing matters, because it tells you what drives the right amount: how unpredictable each of those two things is for each product.
Why a flat rule fails
“Keep two weeks of everything” feels safe and is the most expensive mistake in the book.
A steady bestseller with a reliable supplier barely needs a buffer, because little can surprise you. A volatile product with a flaky supplier needs a much bigger one. Apply the same rule to both and you simultaneously under-protect the risky product and over-invest in the safe one. Multiply that across a catalogue and the over-buffering is where a serious amount of cash silently disappears.
What really decides how much you need
Four things, weighed per SKU:
- Demand variability. How spiky are this product’s sales? Steady demand needs little buffer; lumpy, promotion-driven demand needs more.
- Lead-time variability. How reliable is the supplier? A supplier who always delivers in 14 days needs far less buffer than one who delivers in anywhere from 14 to 40.
- Service level. How confident do you want to be that you will not stock out? This is a dial, not a fixed answer, and it has sharply diminishing returns. Going from 90% to 95% confidence is cheap; going from 95% to 99% can cost a lot of extra stock for a small gain.
- Product importance. Protect your A items, the ones driving revenue, more than your long-tail C items. (See ABC analysis.)
Do not skip the cost side
Every unit of safety stock is carrying cost and tied-up capital, the same silent drain we covered in how much cash is trapped on your shelves. The aim is not maximum availability. It is the right availability for each product’s value and volatility. A 99% service level on a cheap, slow C item is money badly spent. The same on a high-margin A item is money well spent.
A practical way to set it
You do not need a statistics degree to do this well.
- Segment your catalogue by ABC and by how volatile each product is.
- Set higher buffers on A items, volatile sellers, and long lead-time products.
- Set minimal buffers on steady, cheap, low-revenue items.
- Review as velocity and lead times change. Safety stock set once and forgotten drifts out of date the same way a static reorder point does.
How Stockful helps
Doing this by hand across a real catalogue is exactly the kind of per-SKU tuning that does not happen because there is never time. Stockful’s AI threshold suggestions propose a safety stock level for each SKU, alongside lead time, restock-to days, and low-stock threshold, with the reasoning shown so you can sanity-check it before applying. You can tune one SKU in its detail view or many at once in bulk, and those values feed straight into your reorder recommendations and days-of-cover alerts.
Safety stock is not about holding more. It is about holding the right buffer in the right places, so you are protected where it counts and not bleeding cash where it does not.
Get started free at stockful.app. Stockful suggests safety stock per SKU from your real demand variability and lead times, with the reasoning shown.
More from Stockful: