How to Forecast Demand by Size

To forecast demand by size, start with your past sales by size, correct them for stockouts and returns, then turn the result into a size curve. You apply that curve to the total units you plan to buy.

The method needs only a spreadsheet and your order history. It also pairs well with AI size recommendation for fashion stores, because accurate size advice makes the sizes shoppers keep match what they actually need.

Why does demand by size matter more than total demand?

Total units can look healthy while the wrong sizes sit on the shelf. A full stock of small and extra large with nothing in the middle loses sales and ties up cash.

Size mix also changes what you reorder. A forecast that only counts totals tells you how much to buy, but not what to buy.

What data do you need?

Gather these from your store admin and your returns log:

• Units sold by size for each product or product type.
• Days each size was out of stock, so you know which sales were lost.
• Returns by size, with the reason where you have it.
• Your size chart, so you can compare sizes across styles.

If you don't track returns by size yet, a simple log is enough. Our spreadsheet returns tracker shows one layout.

How do you build a size curve step by step?

A size curve is the share of units each size should take. Here is the method.

1. Group similar products. Pool styles with similar fit and customers, such as all women's tops. Don't mix tops with trousers.
2. Count net units by size. Take units sold and subtract units returned.
3. Fix the stockout gap. If a size was sold out for a long stretch, raise its count to estimate what it would have sold.
4. Convert counts to shares. Divide each size's units by the group total.
5. Smooth the edges. Small sizes at the ends have few sales, so one odd order can swing the shares. Use more history for them.
6. Apply the curve to your next buy. Multiply your planned units by each size's share.

What does a worked example look like?

These numbers are hypothetical, built only to show the arithmetic. Replace them with your own data.

Size	Net units sold	Estimated lost sales	Adjusted units
S	20	0	20
M	50	10	60
L	40	0	40
XL	20	10	30
Total	130	20	150

M and XL sold out for part of the period, so their adjusted units are higher. Dividing by the total gives M two fifths of units, L just over a quarter, XL one fifth and S just over a tenth.

If you plan to buy 300 units, the curve gives about 40 S, 120 M, 80 L and 60 XL, rounded to match your supplier's pack sizes. Check the rounding adds back to your total.

How do returns change the forecast?

A size that sells well but comes back often is not really in demand. If many XL orders return because the cut runs small, shoppers wanted a different size.

Count returns against the size that was returned, then look at the size they exchanged into. That second number shows true demand. A size that gains units through exchanges should get more stock next time.

How virtual try-on reduces clothing returns explains how better size advice shifts the mix of what comes back.

What mistakes make a size forecast wrong?

Most errors come from the inputs, not the math. Watch for these:

• Counting gross sales. Units that came back are not demand, so subtract them.
• Ignoring stockouts. A size that sold out early looks less popular than it was.
• Mixing product groups. Tops and trousers have different size curves.
• Trusting tiny samples. A handful of orders at the ends of the range can mislead you.
• Forgetting promotions. A discount on one size can skew the mix for that month.

Write down your assumptions next to the numbers. When you reorder, you can see which assumptions held.

How often should you update the curve?

Review it each time you reorder, and after any change that affects who buys. Common triggers are a new size chart, a new supplier or a new marketing channel.

Keep a separate curve for each product group. A new fit or a change in the customer base can make last season's curve wrong.

Store your stock in a way that matches the curve. If bins and labels are unclear, counts will drift, so see labeling sizes clearly in your stockroom.

Forecasts also depend on garments being made to spec. If sizes run inconsistent from the factory, your sales by size are noisy. Holding factories to spec covers that side.

FAQs

How much sales history do I need?

Use as much as you have for the same product group, and more for sizes at the ends of the range. If a product is new, borrow the curve from the closest similar group.

Should I forecast by product or by product group?

Use groups with similar fit and customers. Single products often sell too few units per size for a stable curve.

What if I sell made-to-order items?

Then you hold less stock by size and forecast fabric and capacity instead. The curve still helps you plan what to cut.

Do I need software for this?

No. A spreadsheet with units, stockout days and returns by size is enough to start.