How to measure the accuracy of artificial intelligence sales forecast?


One of the most important worries when using artificial intelligence techniques to make a sales forecast is the accuracy of the prediction, but this accuracy should not be measured only by how closest is the predicted to real value, because the "game" does not about predict the business failure, for example, the real purpose is empower the business to change, adapt and prepare to situations and the whole time grow. Let's think in a scenario, a company gets six months predictions, for March 2020 to August 2020, for June and July the algorithm says that sales will be decreased by 35% and 43% respectively. The company has at least three months to prepare, deploy new marketing strategies, call customers, launch promotions and more. If these preparations help the company to neutralize the situation, the company increases the sales and the negative situation never happens, but the value of prediction will be 35% far to the real value. The opportunity to prepare the business for positive or negative scenarios is the real value of the sales forecast. The accuracy should be measured by the final business result, not only the value of the prediction. #ArtificialIntelligence #Business #AnniQ

©2020 by AnniQ.