How to Predict Stockouts Before They Happen

Most stores find out they're about to stock out from a badge, a report, or a customer message, and all three arrive after the trend that caused it. By then you're choosing between a rush order and a few missed sales, when a week earlier the same decision would have cost five minutes.

Prediction is a habit, not a formula you calculate once. The individual signs to watch are covered in a separate checklist — ten warning signs you're about to run out. This post is the process behind noticing them before the SKU is already in trouble; for the full prevention picture beyond prediction alone, see how to prevent stockouts on Shopify.

Why prediction beats reaction

A reorder point tells you when to act. It doesn't tell you the number behind it is going stale. Reacting to a crossed threshold is still useful, but it's built on an assumption someone made when they last set it — usually a snapshot of sales velocity and lead time.

Prediction means checking that assumption on a schedule, not just waiting for the alarm that trusts it's still correct. That's a different habit than setting up alerts once and moving on — and it's the one that catches a stockout while the reorder point still says everything's fine.

Days of inventory vs. lead time

The comparison worth making regularly is simple: put your days of inventory next to your supplier's lead time. If days of inventory is comfortably above lead time, an order placed today would land with room to spare. If it's fallen below lead time, an order placed today would still arrive after the shelf is empty, no matter what the reorder point says.

Here's why that gap can open up even on a SKU with a correctly-built reorder point. Take Cedar & Fig, 250g, from the reorder point formula: 5 units a day, a 12-day lead time, 30 units of safety stock, for a reorder point of 90 units. That number was right the day it was set.

Three months later, say sales on this candle have quietly doubled to 10 units a day, a seasonal push or a slow build nobody flagged. Nobody recalculated the reorder point, so it still sits at 90. Stock crosses 90 and the alert fires exactly as designed. But at the current velocity, 90 units is only nine days of inventory, three days short of the 12-day lead time, before the order has even been sent.

The reorder point wasn't wrong when it was built. Comparing days of inventory to lead time directly is what exposes that it's now too low, the same stale-velocity problem the reorder point guide's own "three mistakes" section names, caught earlier by comparing two numbers instead of trusting one threshold.

Watching the velocity trend

The days-of-inventory check above catches a velocity change once it's already large enough to move the number. Watching the trend catches it earlier: put a recent window (last 30 days) next to a longer one (last 90 days) for your higher-volume SKUs and look for the recent number pulling away from the longer one.

A trailing-30 average running well above the trailing-90 average is the same signal as the worked example above, spotted a few weeks sooner, while there's still slack to recalculate the reorder point rather than rush an order. Our guide on improving forecast accuracy covers recalculating velocity properly once you've spotted the shift.

The stockout feedback loop

Prediction matters more than it might seem, because a stockout doesn't just cost the days a SKU sits at zero — it can distort the data you'd use to predict the next one. Sales history from a stockout window shows near-zero demand, not because customers stopped wanting the product but because there was nothing to buy. A forecast that treats those flat days as real demand orders a little less next cycle, and runs out again a little sooner.

This effect, known as censored demand, is explained in full with a diagram in our post on common inventory forecasting mistakes — worth reading once, not re-argued here. The takeaway for a prediction habit: catching a SKU before it actually hits zero keeps this loop from starting on that SKU at all.

Building a review habit

A workable routine for most small catalogs:

• Weekly: check days of inventory against lead time for your top-selling and highest-risk SKUs.
• Monthly: recalculate velocity for anything that's trending away from its recent average, and update the reorder point to match.
• After any spike: a viral moment, a promotion, a seasonal jump — recheck the SKUs involved immediately rather than waiting for the next scheduled review.

Doing this by hand past a few dozen SKUs is where most stores stop keeping it up. StockCue recalculates demand forecasting from live sales data nightly, on every plan including Free, so the comparison above runs on its own instead of on a spreadsheet you remember to open.

Frequently Asked Questions

How can you predict a stockout before it happens?

Compare your days of inventory remaining against your supplier's lead time for each SKU, rather than watching the raw stock count alone. When days remaining drop below lead time, an order placed today would still arrive after the shelf is empty — that's the moment worth acting on, days or weeks before a formal reorder point or low-stock badge fires.

What's the relationship between days of inventory and lead time as a warning sign?

Days of inventory tells you how long your current stock will last at its recent selling pace; lead time tells you how long a new order takes to arrive. When days of inventory sits comfortably above lead time, you have room. When it drops below lead time, the math no longer works — and a reorder point calculated on an older, slower sales pace can miss this even when it looks fine on paper.

Why does one stockout make the next one more likely?

A stockout stops sales for however long the shelf is empty, and that flat period gets recorded as low demand rather than lost demand. A forecast built on that history reads the gap as real, orders a little less next cycle, and runs out again a little sooner — a pattern called censored demand, covered in full in our guide to inventory forecasting mistakes.