The Opportunity
- The sales executives of this consumer packaged goods company relies on sales and promotion data to help them plan for profitable promotions. Data from syndicated data suppliers is combined with internal sales and promotion data to estimate the incremental volume generated by each promotional or price discounting event.
- In the past, estimates of base (non-promoted) volume and incremental volume attributed to the promotion was heavily dependent on metrics as provided by the syndicated data supplier. But the estimates were unreliable and varied unreasonably across events, products, and geographies.
- The company was determined to develop their own promotional response models and baseline algorithms.
Our Approach
- The company had implemented a preliminary system many years prior, but it had never been consistently and fully adopted.
- We did an assessment of the methodology, algorithms, and models, as well as of the programs written to support the system.
- The key recommendations that we implemented resulted in improvements of the models’ ability to quantify promotional lift and generate more reasonable baselines, while maintaining acceptable levels of overall sales prediction accuracy.
The Impact
- The previous barriers to adoption where largely due to the lack of credibility of the syndicated baselines, and internal models built to estimate promotional lifts. The new baselining algorithm and enhancements to the models demonstrated more believable results, and fostered adoption.