Watch this Ask the Expert session to learn some of the tools that collect, visualize and interpret data so you can boost sales and profits even while enhancing customer engagement.
Watch the Webinar
You will learn about:
- End-to-end promotion analytics. Explore the entire life cycle of promotions, from evaluating past promotions to optimizing and forecasting current and future ones.
- Recommended SKU list. Using promotion optimization, you’ll discover how to obtain the optimal SKU list to enhance your promotions.
- Cannibalization/halo effect. Better understand promotion cannibalization and how to use data to help prevent it.
The questions from the Q&A segment held at the end of the webinar are listed below and the slides from the webinar are attached.
Q&A
Which models do you use for forecasting the promo?
It depends upon the time series which we are looking forward to. From my personal experience, if you have seasonal time series, you can use a seasonal ARIMA model with the input causal factors. One such causal factor could be promo price & type that can be defined for the past & future. ARIMA models will then pick the causal event and give you a beautiful forecast for the future. But in case your time series are not that seasonal and most of them are highly volatile, long short-term memory models (LSTM) works very well for such cases by capturing the impact of the past promotions and projecting it again in the future.
How do you calculate the baseline sales & separate the uplift from the Actuals?
So, for calculating the baseline, it's like a step-by-step procedure what we apply (refer to slide 21). In the first step, we determine the list of SKUs, which influence a certain promo SKU. Basically, we calculate the cross elasticities between SKUs. Second, we group the SKUs based on the customer decision trees, the products in the same branch are substitute products. Finally, we calculate the baseline sales using the formula shown in Slide 21, where ‘S’ stands for seasonality, ‘T’ stands for the trend and ‘C’ stands for the coefficients of cross elasticity. Uplifts are calculated by subtracting baseline sales from Actual sales.
How do you calculate cannibalization & cross-elasticities between products?
Cannibalization = (Sales of existing product before promotion - Sales of existing product after promotion) / Sales of existing product before promotion
Cross-Elasticity = (Change in Quantity Demanded of Product A / Average Quantity Demanded of Product A) / (Change in Price of Product B / Average Price of Product B)
Recommended Resources
Enhance Forecasting Accuracy with Time-Series Segmentation-and Machine Learning
Demand Classification - Documentation
Working with segmented pipelines - Documentation
Neural Network-based forecasting strategies in SAS Viya
Please see additional resources in the attached slide deck.
Want more tips? Be sure to subscribe to the Ask the Expert board to receive follow up Q&A, slides and recordings from other SAS Ask the Expert webinars.
Click image to register for webinar