The retail scene in Portugal is highly competitive and distinguished by a strong promotional activity, where promotional sales contribute to almost 50% of overall sales. Consequently, consumer goods companies and retailers face significant challenges regarding promotional monitoring, forecasting, and planning.
Our client Unilever FIMA, that manages Unilever’s presence in the Portuguese Market, was determined to review its promotional planning activities to increase overall efficiency, not only to keep competitive but also to assure its long-term sustainability and success. To achieve this goal, 3 main questions needed to be targeted:
The aforementioned challenges were targeted with a new approach to promotional planning switching from a manual process, poorly leveraged by data insight to a systematic, data-driven and decision supported process, based on four axis whose interdependency fuel a holistic solution.
The first step was the development of an agile and robust data gathering process, to assure the centralization of the promotional information without dependence on sparse excel files and specific stakeholders’ knowledge to access the information.
The access to information in a centralized database allowed the development of a dashboard to foster promotional planning effectiveness and efficiency monitoring with access to relevant business metrics. Decision-makers with augmented access to information and effective KPIs are more prepared for better and faster decision-making.
To assess the impact of different promotional activities in several performance indicators (gross sales, margin, among others), a predictive model using machine learning tools was developed by gathering historical sales data, products attributes, time horizon and promotional typification, such as discount and communication method. Additionally, external data was used, (e.g., customers’ sales data, usually known as sell out), in order to bring incremental value to the developed machine learning algorithms and ultimately improve forecasts accuracy.
The predictive model allows not only a more reliable promotional sales forecasting but also access to what-if scenarios to understand the impact of alternative promotional offers to retailers.
Finally, an optimization module was developed with the goal of proposing optimal promotional plans across a planning horizon, choosing the right offer, to the right customer at the right moment, out of a portfolio of previously approved promotional offers. This module is able to maximize different business objectives such as sales volume, turnover, or gross profit.
By ripening decision-making from an empiric and manual to a standardized data-driven process, Unilever FIMA can monitor and improve the effectiveness of different promotional approaches, but also question already settled promotional guidelines. Results include excellence in planning processes, increased forecast accuracy and optimized promotional plans, with a direct impact on both sales and gross profit. As future work, being able to support the change of promotional guidelines is the ultimate step towards trade promotions planning excellence.
We transformed Unilever FIMA’s trade promotions planning by systematizing the activities and enabling access to data to all key accounts, and by leveraging advanced data analytics to achieve superior plans