Skip to content
lesslater

Guides Forecasting

Seasonality in the inventory forecast: seasonal index step by step

How to calculate a seasonal index per month, use it in the forecast and handle products with a short history.

Updated 8 November 2026 · 3 min read

A seasonal product sells differently from month to month, but in the same way every year. An average over the year is wrong at both ends: too low in the peak season and too high for the rest of the year. A seasonal index corrects this.

Step 1: sales per month

Take sales per month for the last 12 or 24 months, cleaned of stockout days and promotions. Two years are better than one, because a single year can have random fluctuations.

Step 2: index per month

Divide each month’s sales by the average for all months. If a product has sold 1,200 units in a year, the average is 100 units per month:

Month Sales Index
September 90 0.90
October 120 1.20
November 150 1.50
December 189 1.89

An index of 1.89 means December sells 89% above an average month. If you have two years, take the average index for each month, and make sure the 12 indices add up to 12, so the level does not shift.

Step 3: find the level

The level is the product’s current sales without seasonality. You find it by dividing the most recent month’s sales by that month’s index. If the product sold 120 units in October and October’s index is 1.20, the level is 100 units. If the product has sold 15% more this year than last year, you can see it in the level.

It is probably more stable to use an average of the last two or three months, each divided by its own index.

Step 4: forecast

Forecast for a month = level × the month's index

December: 100 × 1.89 = 189 units. January with an index of 0.70: 70 units.

When the history is short

With less than a year of sales, you do not know the product’s own seasonality. There are then two options:

  • Borrow the seasonality from the category. A new Christmas star will probably sell like the other Christmas stars, even though it has no history yet.
  • Borrow the forecast from a similar product. If one product replaces another, the new one can use the old one’s history.

This is generally better than calculating a new product from its first three months. If those three months are September to November, the product looks like a product in growth, and you end up ordering far too much for January.

Seasonality and lead time

Seasonality also has to be used when ordering. With a long lead time, Christmas products have to be ordered while sales are still low, and then looking at the last few weeks’ sales does not help. Read more in the guide to reorder point.

Seasonal profiles in lesslater

lesslater calculates a seasonal profile for each product with at least a year of history. Products with a shorter history borrow the seasonality from their category, and new products can borrow the forecast from a similar product or be given an expected sales figure per month.

See your own products in lesslater

Create an account and connect your webshop. The first 14 days are free, and no credit card is needed. If you would rather go through it on your own data with us, book a demo.

No credit card · No commitment · Data in the EU