Stefan Stremersch wins 2012 IJRM Best Article Award
It has important implications for marketing managers
Professor Stefan Stremersch and his co-authors have developed a dynamic model of new product adoption with remarkable explanatory power. The research, which has important implications for marketing managers, was chosen by the Editorial Board of the International Journal of Research in Marketing (IJRM) as one of the two best articles published in the journal in 2012.
Country segmentation can be a powerful instrument for prelaunch forecasts of new products, argue Stremersch and his co-authors former ERIM Member Aurélie Lemmens (Tilburg University) and Christophe Croux (Catholic University Leuven) in the award-winning article. Before the first launch, firms can study how similar products have fared in countries that resemble the target market.
In such international segmentation studies, similar countries are identified only once. Yet the authors argue that product adoption is a dynamic process: which countries make for suitable comparisons, changes over the lifecycle of a product. Indeed, "[F]rom a managerial viewpoint, ignoring dynamics in international segments is likely to lead to suboptimal marketing strategies," stresses the research team.
In order to better understand product adoption and optimise marketing strategies, Stremersch, Lemmens and Croux have developed a dynamic model – a "semiparametric hidden Markov model". Hidden Markov models "allow segment membership to dynamically vary over time," explain the authors. In this case, they allow for changes in country segments. Semiparametric modelling in turn solves problems in parametric modelling when the observed time window is short or when data contain repeat purchases.
The new model beats existing models when tried with historical data. Using data on the adoption of six new ICT product segments such as DVD players, internet subscriptions and mobile phones in 79 countries between 1977 and 2009, its forecasts proved remarkably accurate. "For [DVD players], we find that our forecasts deviate by only .64 units (i.e., 0.64% of the households in the country) from the actual penetration levels 5 years after introduction," the authors write. "For Internet subscriptions, the dynamic segmentation yields forecasts that deviate by 2.22 units (i.e., 2.2% of the households in the country) on average from the actual penetration level 5 years after introduction."
"These results are particularly encouraging given that the forecast of the first five years after launch is probably the most crucial for managers when they plan to launch a new product on the market," Stemersch and his co-authors argue. They hope their research will encourage multinational corporations to adopt dynamic segmentation methods rather than static methods.