Can we forecast new product and service developments? And how should it be done?

20 May 2016

Professor Robert Fildes and his doctoral student Oliver Schaer attended the 18th IIF Workshop on Forecasting New Products and Services, held in Milan May 2016. This international workshop series, organised by the International Institute for Forecasters (IIF) with Dr. Mohsen Hamoudia (Orange), and supported by the Politecnico di Milano, aims to bridge the gap between theory and practice by bringing together leading academics and practitioners.

Forecasting the adoption of new products is a challenging task, even more so when it includes the development task itself. The workshop covered a large variety of topics within the area including a talk from Dr. Larry Vanston from Technology Futures about the novel fundamental driving forces of future technologies such as robotics, machine-to-machine interaction and personalisation,. A particularly interesting talk by Professor Paul Goodwin, an associate of the Forecasting Centre, highlighted the role of judgement in the forecasting process: This is important to the field of new product forecasts, as there is necessarily a limited amount of historical data present, usually zero! In a later session of the workshop a simple example using the Delphi method was carried out where the audience estimated the adaption of self-driving cars by 2025 and 2035 in Italy. The anonymous setting of the survey makes this decision tool more robust against judgmental biases. The range of estimates went from zero (no Italian male is complete without his car apparently) to 90%.

One way to overcome the dependency of judgemental decision is to use explanatory variables. Oliver Schaer, presented findings from a recent experiment in which Google Trends information was used for estimating the market size within product generations. In the context of physical video games sales, market size determined by Google Trends have found to perform better, compared to benchmark models using analogy information from past generations values only. The results also hold with a longer lead time that takes into account real world setting in the supply chain.

Further presentation included topics such as investigating the drivers for success of new Kick-Starter projects and highlighting evolutionary trajectories of technology in the medical device industry. Other presentations focused more on the forecastability of the product development process and elaborated on the role of forecasting within the setting of lean development. Here the on-line collection of consumer data is critical so forecasts and even the design of the product/ service can be updated. The final round table chaired by Professor Robert Fildes highlighted that a practical impact can only be achieved with a clear forecast objective and the concluding remark that the area of judgment and new product forecasting remains under-researched topic in general.

Download Oliver's presentation:

Further practitioner oriented presentations can be found here.