Product Innovation using Stage-Gate: An exploratory study of Machine Learning on product innovation.

Abstract

With the rapid pace of development of technology and the increase demand for new products across industries and markets, there has been a growing focus on New Product Development, models that will improve the process, reduce the level of uncertainty when it comes to making decisions on whether to move forward with a new product or not and speed up development time. One of the most well-known and used models for New Product Development is the Stage-Gate Model, proposed by Cooper (1990). This model has seen a number of iterations over the years, leading to many adjustments to the model to improve the processes, feedback from customers and reduce failures. This model has been extensively used within manufacturing, and it is well documented that it is used within the automobile industry. As technology has developed, new tools have come to the fore and there has been a movement from Big Data to Machine Learning and more recently the growth of AI through the development of Large Language Models. Although there have been some recent proposals for how to use AI within NPD, there is not a lot of newly published research on how to use Machine Learning or AI within Stage-Gate, meaning no clear framework or well tested model. The exploratory study focused on the use of Machine Learning and how it was being used. The Electric Vehicle industry was selected for the study as this is a rapidly developing industry with constant demands for new products or features. There is also a focus on new technology and a race to develop vehicles that provide greater safety features, longer driving ranges and better self-driving experiences. In the study, five organizations provided a total of fifteen people involved in NPD to be interviewed as well as Senior Managers manning the Gates. The findings showed that Machine Learning is being used at some stages, that the companies are at an early phase of use of Machine Learning or AI and that there is mistrust of the suggestions from Machine Learning due to concerns around hallucinations of data. The study proposes a new model, based on Stage-Gate, for the incorporation of AI, to ensure that the right data is ingested in the Stages and the AI provides suggestions at the Gates based on the data and development in the Stage. This new model would allow for more flexible and well-rounded use of AI within Stage-Gate and allow Senior Management to build more trust in the suggestions. This model would ensure that the suggested decisions from the AI are based on the correct data and should reduce the likelihood of data hallucinations. The study suggests further research into the model by working with a number of organizations to test the model and results.