How Stockouts Distort Your Demand Forecast
Here is a pattern many retailers know: a product sells out, the reorder suggestion comes back smaller, it sells out again sooner, and the next suggestion is smaller still. The product isn’t getting less popular. The forecast is reading the stockout as falling demand.
Why it happens
Most simple forecasts average sales over a period. When you’re out of stock, sales are zero, not because nobody wanted the product, but because there was nothing to sell. Those zero days drag the average down.
Example: a product sells 10 a day. It is out of stock for 12 of the last 30 days. Sales over the month are 18 × 10 = 180, an average of 6 a day. The forecast orders for 6 a day, the stock runs out even faster, and the cycle repeats.
Measure demand on in-stock days only
The fix is simple to state: calculate the average using only the days when the product was in stock. In the example, 180 ÷ 18 = 10 a day, which is the real demand.
Other things that hide demand
- Low stock on display: one last unit on a shelf sells slower than a full facing.
- Variant gaps: if the popular size is gone, sales of the product drop even though other sizes are in stock.
- Promotions: one-off spikes can inflate a forecast if they aren’t recognised.
A second example: the variant trap
A T-shirt sells in five sizes. Medium and large make up most of the sales. When medium sells out, the product page still shows the shirt as available, but total sales drop by a third. A forecast at product level reads that as falling demand for the shirt. At variant level, the picture is clear: medium had no stock, and the other sizes sold as normal. Forecasting per variant, on in-stock days, avoids cutting the whole order because one size ran out.
What to do about it in a spreadsheet
If you forecast by hand, you can correct for stockouts with three columns per product:
- Units sold each day.
- In stock? Yes if the product had stock at the start of the day.
- Average = total units sold on in-stock days ÷ number of in-stock days.
It works, but it is slow to maintain for more than a few dozen products, and you need a daily record of stock levels, which Shopify’s standard reports do not keep for you.
The cost of getting it wrong
The stockout spiral is expensive in two ways. First, every day out of stock is lost sales, and some of those customers buy elsewhere and do not come back. Second, small reorders often mean paying more per unit or more for shipping, because you miss price breaks and order more often. On a bestseller, a forecast that runs 30% low can cost more than all your slow-moving stock combined.
Promotions and one-off spikes
The opposite problem happens after a big sale or a mention by an influencer. A one-week spike can inflate the average and lead to over-ordering. Mark promotion periods, and either exclude them or treat them separately, so normal demand is not overstated. Our guide to finding overstock shows how to spot the result.
Put it together with your reorder point
Corrected demand feeds directly into your reorder point and safety stock. If demand is understated, both come out too low. See how to calculate a reorder point and safety stock explained for the formulas.
How to check your own forecast
- Pick five bestsellers that sold out in the last three months.
- Compare their average daily sales with and without stockout days.
- If the gap is large, your reorder quantities are too small.
How Restocked helps
Restocked measures demand only on days you actually had stock to sell, so a stockout never shrinks your next order. Each forecast shows a confidence badge, and every recommendation shows its math: sales velocity, lead time, safety stock and what is already on order.
Read next: Stocky Alternatives Compared: Purchase Order Apps for Shopify · Shopify Stocky Alternatives: What to Use Now Stocky Is Retired
FAQ
How many in-stock days do I need for a good forecast?
Several weeks of in-stock days is a reasonable minimum for steady products. With fewer, treat the forecast as a starting point and review it often.
Should I forecast at product or variant level?
Variant level. Sizes and colours sell at different rates and run out at different times.