Measure and predict and improve retail promotion success

AIR-Pro helps retailers grow revenue and margins by optimizing promotions. Promotion effects are complex and multidimensional, which makes them notoriously hard to measure and predict. Most retailers deploy simplified methods which leave money on the table. Any solution that aims to automate promotion optimization has to provide a standardized and generic approach that works well in very specific situations. AIR-Pro solves this puzzle with three unique innovations which have helped European food retailers improve promotion margins by 20-40%.

1. Post-ROI based on 11 Customer Lifetime Value effects

Most retailers measure promotion success by looking at sales uplifts. This means comparing sales during the promotion period to regular sales. Although this gives a useful first rough estimate, there are many second and third order effects which should also be considered. These effects can make seemingly profitable promotions unprofitable, and vice versa.

AIR-Pro measures promotion effects by calculating 11 drivers which impact the customer lifetime value.

  • The first three drivers measure the cannibalization, substitution and hoarding effects on product level. This means compensating for margin loss for customers who would have bought the product anyway, foregone sales of substitution products and sales drops after the promotion due to hoarding.
  • The other eight drivers go a step further and account for individual customer behavior. They calculate for instance the value creation from newly attracted customers.

All these effects combined form the post-ROI. Retailers can use this advanced post-ROI to evaluate promotions, segment their promotions into different success categories and shift their promotion portfolio from loss making to winning promotions.

2. Self-learning Pre-ROI optimization

AIR-Pro is not only able to measure the ROI of past promotions, but also predict the ROI of new promotions, even if these have not been done before. The application uses various machine learning algorithms to extract predicting patterns from past promotions. It uses up to 15 different parameters to predict the ROI of a new promotion.

Category managers, promotion managers and supply chain managers can use this functionality to:

  • Determine their optimal promotion mix
  • Design winning promotions by maximizing ROI
  • Determine stock levels and purchase quantities


AIR-Pro can be easily integrated with existing systems via APIs, such as promotion and supply chain management software.

3. Special kind of AI: Augmented Intelligence

AIR-Pro is unique in its ability to create synergy with users. It deploys a couple of features which creates a 2-way feedback, allowing both the algorithms and users to continue improving over time.

  • AIR-Pro gives a reliability score for any new prediction. 80% of new promotions typically score green, meaning that their predictions will be quite accurate and can be trusted without much further consideration. Amber and Red predictions deserve extra attention from the user, and benefit from subject matter input.
  • In particular when predictions have low reliability, users can override expected sales. After the promotion period, both human input and algorithmic predictions can be compared to actual results. This tells the users when they are right to apply their judgment, and learn when to trust the algorithms.
  • Model Management functionality provides insights into the prediction performance of the algorithms. This is monitored over time, and input for maintaining and upgrading the algorithms. This can be compared to the dashboard lights in a car, indicating that oil needs to be refilled or batteries need replacement.

Using AI to increase retail promotion profitability by 20-40%.

“Buy one, get one free… Don’t miss this special offer: click here for 2 months premium membership.” Promotions. Retail is hard to imagine without them. Spending on promotions in Europe has doubled over the past 10 years, with nearly 30% of purchases being on promotion.