Data-driven sales forecasting in consumer goods production

Challenge

Companies are becoming more aware of the advantages of a data-driven decision-making policy, even though a large part of them are still learning how to effectively extract information and insights from data.

Consumer goods producers may largely benefit from advanced analytics methods in supply chain planning and operations.

Production planning and procurement tasks rely on accurate sales forecasting, which is typically done by sales and marketing managers using information about future orders and customer behavior. Sales plans are also used by a variety of teams within different departments of a company, emphasizing the importance of having a cohesive master plan that serves as a foundation for all processes within the organization.

Besides the clear challenge of creating a better method of generating a sales forecast that could leverage past sales data and analytical models, which are able to encode market dynamics and seasonal trends, there is also the challenge of including all relevant stakeholders in the process and integrating the valuable information they can provide into a cohesive master sales plan.

Solution

In order to include all stakeholders and improve accuracy and communication between all company departments in regard to the sales forecast, the forecasting solution we developed for a beverage producer followed three key structures:

1. Forecast generation: Leveraging historical data from sales and orders with advanced analytical forecasting methods. Sales plan generation was divided into two distinct processes, with different goals: long-term and short-term forecast generation. The long-term forecast (1-3 years horizon) feeds both the budgeting and procurement processes. On the other hand, the short-term forecast (12-15 months) is mainly used as input for production planning and purchasing. The forecasting model uses past sales data to extract patterns and learn the best method for each individual product family and market. It runs several tests on different levels of data aggregation to decide which level of granularity enables the most accurate prediction;
2. Validation by commercial team: After the short-term generation process, the plans are validated and adjusted by each market’s manager, according to the information they have on future orders or customer insights. The plans are accessed and adjusted on a centralized platform that is connected to a unified plan, guaranteeing cross-visibility. This process of validation was also redesigned to ensure a more frequent (monthly) review of the generated forecasts;
3. Monitorization: Every change and the new process should be tracked through KPI analysis. Considering this, a monitoring dashboard was developed, where the decision-makers can track the main indicators, such as the sales plan’s accuracy, bias, or even the quality of the market’s managers adjustments.

Result

By centralizing and standardizing sales planning in one single place, accessible to the whole organization, and supporting empirical adjustments with a data-driven methodology, the organization was able to effectively improve internal processes that depend on planning and actively monitor the KPIs that are core to the business.

This analytical approach to the planning process will support the path to excellence in future processes.

Increasing the accuracy of sales forecasting can lead to a number of financial benefits for a company, such as increased efficiency by preventing over or underproduction, improved financial planning, higher customer satisfaction by serving more clients and missing fewer sales or even increased competitiveness enabled by a more reactive posture.

Preliminary results showed that this project not only made the planning process more robust, but also added quantifiable benefits, namely an increase in forecasting accuracy of 6 p.p. overall and of 10+ p.p. in the high volume/low variability products. According to benchmarks, this accuracy increase typically represents a financial gain of 1 to 2% of sales volume.