Skip to content
GetProfit

← All terms

Demand forecast

A demand forecast is an estimate of how much a store will sell in the coming months, built from the shape of its season and its current level of sales.

How it works

Monthly sales have two parts: the season, which repeats each year, and the level, which moves with growth, budget and assortment. Each month’s seasonal index comes from past years, with growth taken out by detrending. The level is the latest months’ sales divided by the sum of their indices.

The season’s shape needs a second year that repeats it: one year cannot tell a season from a one-off event. Ad revenue follows the budget, so its forecast holds only at the current spend. A store with no history can borrow its category’s shape from Google Trends, which shows search interest on a 0–100 scale, not volumes. The method step by step: demand forecasting for an online store.

Formula

Forecast for a month = current level × that month’s seasonal index

Where you see it

  • Google Ads: the Insights page, under “Predicted trends”, lists products and services whose search interest is forecast to start rising within 180 days.
  • GetProfit portal: Assortment and seasonality takes 24 completed months and checks whether the second year repeated the first. If the years look nothing alike, it builds no seasonal plan.

Example

Example store, not client data.

The tableware shop’s indices are 0.75 for June, 0.75 for July and 0.8 for August: 2.3 together. This June to August brought 496,800, so the level is 496,800 ÷ 2.3 = 216,000 per average month. November’s index is 1.6, so the forecast is 216,000 × 1.6 = 345,600.

Not to be confused with

  • Performance Planner — forecasts what campaigns get at a given budget, from auctions simulated over the last 7 to 10 days. It plans spend; a demand forecast estimates sales.

Right and wrong readings

  • Wrong: “Last November brought 288,000, so we plan 288,000 again.” Right: that repeats last year’s level of 180,000 a month. At this year’s 216,000, the same index of 1.6 gives 345,600.
  • Wrong: “September came in 40% above August, so the level rose by 40%.” Right: with August at 0.8 and September at 1.12, a 40% rise is all season: the level has not moved.

Sources