AI-driven sustainable advantage

Challenge
- • Accurately predict very low volumes of sales because in 95% of cases the number of copies sold is between 0 and 3
- • Lower the rate of unsold copies
- • Reduce the carbon impact
Our approach
- • Data collection leveraging 13M observations, covering 20,000 points of sale over 5 years
- • A chain of modeling that includes classification, regression and optimization algorithms
- • A direct data flow connected to the distributor data lake
- • Input from market experts to improve model performance
- • A user-centered web app to support decision-making
Outcome
- • An end-to-end solution that’s fully adopted by the business and embedded into decision processes
- • Robust modeling that addresses the challenge of predicting very small volumes of magazines and better manages impulse purchases
- • A fully industrialized solution that can improve profits and reduce carbon emissions by reducing volumes of unsold magazines

Results

+10 margin points

and +6 points on annual rate of sales

-25%

Reduction of paper used

2,300 tons

of CO2 saved per year

Challenge

Given that unsold items represent a loss of 7,000 tons of CO₂ per year, how can the number of copies sent to points of sale be optimized to avoid this unsold stock? Ekimetrics has been focusing on this question in order to meet two crucial challenges: significantly reduce carbon emissions; and maintain or even improve margins.

Approach

Ekimetrics has developed a sales prediction solution called AthenIA Press. This technology comprises several algorithms to meet the challenge of optimizing paper sales across stores. In concrete terms, the first algorithm, known as “classification,” categorizes points of sale according to sales potential. The “regression” algorithm determines the volume of magazines to send to each high-potential point of sale, while a third algorithm determines the optimal volumes of magazines to distribute in order to maximize profit.

The publisher’s teams have also been provided with a dedicated web application enabling them to parameterize, plan and visualize predictions, as well as access financial reports to optimize their decision-making. At the same time, the app features a business insights module that enables operational teams to enter product and market information to feed the AI with their business intelligence.

Outcome

Based on a sustainable data architecture structured to scale, Ekimetrics’ algorithms have helped the client to better manage demand forecasts and embed daily predictions into its daily business operations. By introducing not only financial but environmental KPIs—such as tons of CO₂—this project has had both an economical and a very strong environmental impact.