Demand planning is the work of deciding, weeks or months ahead, how much of each product a company will sell, and then getting the right stock to the right place at the right price. It is also called demand and supply planning. The software that does it forecasts sales, plans what to make or buy, splits the stock across stores and sets the prices. Its buyers are manufacturers, retailers and the brands that sell through them.
Every Diwali, people in India buy far more TVs, clothes and sweets than in an ordinary month. A TV maker must decide in June how many TVs to make for October, because the parts take weeks to arrive. A clothing chain must decide how many kurtas to order, for each store and in each size. After Diwali, it must decide when to cut the price of the kurtas that are left over.
Here is a typical case. A clothing chain has 150 stores across western India. In May, its merchandise planning software decides which kurtas the chain will carry for the festive season, and how many of each. It starts from a demand forecast, a prediction of how many of each item will sell, where and when. The forecast reads last year's daily sales, which are time series data: numbers recorded one day after another.
Next, supply-chain planning software turns those quantities into a plan. It works out how many kurtas each garment factory must make, how much fabric to buy, and how much stock to hold in each warehouse. The plan goes to the chain's enterprise resource planning (ERP) system, where the purchase orders are raised.
From September, competitor price monitoring software checks what rivals charge for similar kurtas on marketplaces and in their own apps, several times a day. Finally, price and promotion optimisation software picks the price in each store. It also picks the offers, such as "buy two, get one free", and the markdowns. A markdown is a price cut that clears unsold stock. Prices that models change as demand and rivals' prices move are called dynamic pricing.
Why does this matter? If the chain orders too few kurtas, customers leave empty-handed in the busiest week of the year. If it orders too many, it must sell the rest at a loss in November. Every extra kurta also uses money that the chain could have spent on something else.
The newest planning software does more than present a plan. AI agents now replan as sales and shipments change, and they act on the new plan themselves. Human planners approve only the unusual decisions. AI that acts on real stock, trucks and machines in this way is described in physical AI.