Leveraging Machine Learning to Optimise your Trade Spend

Trade spend is typically the second highest expense item in the P&L statement of a consumer goods company, after cost of goods sold (COGS). Hence, the planning, execution and analysis of trade spend strategies is highly significant to the company's overall marketing strategy. Consumer goods companies adopt a variety of tactics for their promotions as part of their trade spending, few common examples include the shopper’s delights like BOGO (Buy One Get One Free), coupons or traditional events like short-term price discounts.

While it is crucial to have an experienced demand planner in formulating the annual promotional plan, it is also important to have a system which can use the history for analysis and predictive planning of future promotions. The system should ideally be capable of making predictions with very little human intervention.

Event modeling to better understand your promotion strategy

Any prediction system is as good as the quality of data made available in training the system. Poor quality data results in subpar predictions which lead to retailers stocking an amount that hardly meets the customers demand, resulting in lost opportunities.

Event modeling is an art and is enriched by the inputs and hypothesis testing induced through data discovery, backed with insights from experts. The more causals and history that could be included for the model development, the more we will be able to avoid black swan or outlier scenarios.

A simplistic representation of an event modeler (machine learning framework) is as below:

Leveraging Machine Learning to Optimise your Trade Spend

The model training step forms the heart of the system where the causals are treated as predictors and is assembled in mathematical form to explain the outcome. One should handle the training data with utmost care and ensure to incorporate methods to detect data outliers. Algorithms should be enabled for modeling the derived causals for training like seasonality, which could decide when the event should be positioned.

Enabling Machine Learning for accurate event prediction

There is a whole lot of buzz around the AI framework called machine learning and often is perceived as a very complex system to be put together. Applying machine learning specifically to trade promotion planning yields a lot of valuable insights, for example allowing the identification of ideal promotion tactics not visible in simple financial analysis, such as which promotional event to plan just after the summer vacation (unlike the traditional approach of planning a promotion event only during the summer vacation, this framework also helps in planning non-standard promotion design).

There are 2 key facets in implementing machine learning based event detection system –

a) Learning algorithm

b) Scalable solution

For trade promotion modeling, a simplistic multi-variation regressive model with a good variable screening system should yield excellent results and is a more easy & scalable solution to implement. The Promax Optimize solution takes your promotion returns to a new growth trajectory with machine learning framework complemented by insights from skilled CPG domain experts and data scientists.

Promax TPO has predictive planning capability – providing users the ability to quickly and efficiently conduct ‘what – if’ analysis, hypothesis based testing to model different promotion scenarios, and tailor the best promotional calendar based on the demography and company needs.

Our data science and professional services teams are led by experienced domain experts to give you a competitive advantage through best practices in trade promotion optimization.

For more information, visit our website http://promax.wipro.com For queries, write to us at WPAS-Promax@wipro.com

About the Author

Saju has 19+ years of experience in the areas of Data Science, Modeling & Product Development working with leading product organizations like General Electric, Honeywell. He presently works as a subject matter expert (Analytics) on Trade Promotion Optimization and Econometric modeling. He graduated from the flagship GE Technical Leadership Program - "Edison Engineering Development Program". His prior roles include Engineering Manager for sensor development - leading a team for modelling and data science, Developing Analytics based Energy Management Solution for Hospitality Segment.
Saju has filed 4 Patents in the area of analytics & sensor design, also has a publication "Applied Analytics for Hospitality Energy Efficiency and Associated Core Operations Transformation" published in the Cornell Hospitality Report’2015. He holds a Master’s degree in Mechanical System Design from BITS, Pilani & Bachelors in Industrial Engineering from Mysore University.

Service Offerings

Service Offerings

Client Themes

Client Themes

Leveraging Machine Learning to

Optimise your Trade Spend

About the Author

Service Offerings

Industries

Cloud

Cybersecurity

Digital

EngineeringNXT

Holmes

Contact Us

About Wipro

Careers

Location

Leadership

Investors

Innovation

News and Events

Insights

Analyst speak

Products & Platforms

Partner Ecosystem

Sustainability

Contact Wipro