Marketing attribution modelling

As one of the largest food producers in Central and Eastern Europe, Maspex was looking for a solution to simulate promotional activities and measure their effectiveness, in order to select the optimal promotion scenario. It had employed various promotional campaigns based solely on experts’ knowledge. The company suspected that some of those campaigns had not generated the effects management expected, though it could not confirm this analytically.

Maspex performs multiple promotional activities up and down the supply chain, covering 1000 SKUs. This made it extremely difficult to determine which particular promotion actually affected sales. The company assessed each of its promotional campaigns separately and based on different data, which prevented it from carrying out direct comparisons. The biggest challenge was therefore to combine all available internal and external data sources and to establish a uniform set of variables that would have the greatest impact on predicting the effectiveness of promotional activities. An additional confounding factor was the SARS-CoV-2 pandemic, which significantly influenced the consistency of historical data.

Maspex turned to deepsense.ai for a solution to their marketing needs. Our job was to consolidate and integrate multiple sources of data, including transaction data, data on promotions and their coexistence, data on competition activities and market trends, and other external factors such as sales seasonality, weather conditions and variable restrictions related to the pandemic. We also optimized the integration time and created a pipeline for regular data updates. Based on the dataset, we started training predictive models to assess the performance of five different algorithms, which we whittled down to the best performing one. In addition to those for predicting effective promotional campaigns, we created auxiliary algorithms to analyze the periods immediately before and after promotions and evaluate the interactions between different promotion types. The solution enables Maspex to define their promotion, product and target audiences, which the system then analyzes taking into account the list of promotions that have been analyzed, uplift 12 weeks ahead in weekly granulation and the impact on the company’s market share.

The solution supported customer marketing activities by:

• Optimizing the allocation of marketing budget funds by marketing attribution modelling
• Providing recommendations on the features of campaigns that had impacted its effectiveness
• Making it possible to test hypotheses on promotional scenarios
• Reducing workload by automating the analytics

deepsense.ai also helped Maspex put together and kick-start its internal Data Science team, which built a strong foundation for the company’s further Data Science initiatives.