With customers increasingly aware of the hazards of smoking, the global demand for reduced-risk tobacco products augmented massively. However, country-level differences in culture, product availability, and awareness make demand projections for each product a difficult challenge. Moreover, the existence of conflicting (historical) market size information means compiling reliable historical data is a tough task.
Our goal was to build a market model that would compile all information sources into reliable historical demand data, which could then be leveraged into long-term country-level demand forecasts to help worldwide strategic planning.
The first step of the developed methodology is a confidence data-mining algorithm that identifies the most reliable data sources for each country to compile product-level demand for historical years.
Distinct sources of information, such as surveys, third-party forecasts, demographic data, and similar-country adoption rates are then leveraged to forecast the evolution of reduced-risk products worldwide, including countries in which these products are yet to be launched. To do so, (bass) diffusion curves are applied and fitted through machine learning techniques.
The implemented model provides the organization with unique most-reliable estimates for both historical and future demand for reduced-risk products.
By compiling these figures within an intuitive web interface, different stakeholders can validate the forecasts, provide additional inputs whenever they become available, and simulate the impact of market drivers, such as inflation, regulation, and taxation, on the expected demand of each product.
The project enabled the company to feed each market with easy-to-use analytical demand planning capabilities while gathering in-depth market knowledge into a single company-wide platform.