Miss Historical Data? ML can help

November 13, 2022

Promo campaigns can reach the maximum level of effectiveness only when you approach their planning with a deliberate strategy and, what’s important in modern trade, state-of-the-art technology. The best way to ensure accurate and fast promo execution is to automate each step of planning, including baseline and uplift predictions.

One of the worldwide-recognized beverage brands used the TPx Tool from SoftServe Business Systems. Yet, the promo team calculated uplift manually in a third-party system, resulting in:

• Time-consuming work
• Low data accuracy
• Inadequate labor costs

To eliminate these issues and ensure better promo execution, the company set the task to automate uplift predictions for trade promo planning and optimization using AI/ML.

Challenge

Clean and accurate promo data is crucial to use AI/ML for promo predictions. In this case, the company had sales data available. However, we faced two challenges with historical data required for the development of network learning and prediction models:

• Promo details data was not harmonized.
• Historical baseline data was absent

Solution

SoftServe Business Systems used AI/ML to reconstruct the historical baseline and predict planned promo uplift. The ML models ideally fit product category specifics, and predictions were made based on promo type, multiple internal and external factors, and cannibalization effect.

Results

Once AI/ML had all the necessary data in place, baseline and uplift calculations were fully automated, and the team could make quick and accurate promo predictions. The results were outstanding:

• +10% accuracy improvement
• x100 times faster
• 100% automated calculation

Conclusion

AI/ML is the key to automated promo predictions that leads to accurately targeted promo execution. The better your business automates promo planning, the more effective your team will put all data to work for impeccable customer engagement and, as a result, high profits.