Client

Our client is a global manufacturer and provider of essential, personal care products. They employ 800 people across Australia and New Zealand, all working together to deliver ‘Better Care for a Better World’.

Purpose

Our client required a re-haul of their manual process for analysing and understand promotions across all their 9 product categories, 4 retailers and more than 500 products. The final objective of the project is to be able to give the internal team the tools to be able to better understand the impact of the different promotions being run together (as well as competitor promotions).

The key requirements included:

• An automated pre-processing pipeline that would fetch each raw data file and perform all the data transformations and reshaping to produce a modelling dataset for each model. It was required that the process be flexible to accommodate for which brands/products to model on, which unit to use (SKU/segment) as well as how to define the promotion groupings.
• A modelling pipeline to run linear regressions across all the different modelling units whilst optimising for the best fit possible to the data.
• A promotion calendar template across each retailer and product group to enable the user to run different simulations (own brands and competitors) to understand the financial impact on the year P&L.
• Upskilling of the internal team in the use of the tool to be able to run the process in-house.

Approach

Forecast worked closely with the client to understand the nuances and complexities of the data and types of promotions existing across different group promotions. Leaning on Python and Jupyter notebooks as the tools to develop the end-to-end process, Forecast developed an initial solution for the product category with the highest level of complexity (more products and segments). Building on this solution, Forecast expanded and tested the process across the remainder of the product range.

Outcome

Forecast delivered a full end-to-end data modelling process that is flexible and easy to manipulate for the inhouse team. This automation has substantially reduced the manual work and complexity of the task while delivery high value in enabling a holistic understanding of the impact of promotions across different products. This enabled our client’s team to use the models to forecast different financial outcomes based on business assumptions on which promotions to run.