Update from Crowd Predictions From the Frontline

We are now four months from accomplishing the three-year empirical study “Crowd Predictions from the Frontline”, which investigates the ability of frontline employees to predict critical performance measures of multinational organizations to improve decision-making. In Lego Group we study product quality and in Radiometer Medical we study retention and attraction of employee talent. It’s been a great and inspiring research process.

During the research process, we have hosted two successful conferences on the topic of Crowd Predictions, with guest speakers from Israel and the UK specialized in AI, Big Data and prediction markets, and we have experienced an increased interest in these events over the years.   

Since the project began back in August 2017 we have structured the research to support a deep dive into the unique conditions that affect each respective partner company in order to tease out the elements that affect the relationship between frontline employees and the study area. To obtain research validity, the project follows a rigid methodology and validation over time, structured as follows:

Strategic discussions with partner companies reveal the respective area of interest, so that the prediction study is developed as a unique case study. Based on the company’s respective areas of interest, frontline employees were interviewed for a first collection of data that was later developed in subsequent phases: ideation, filtering, and predictions.

In the ideation phase, software is used to solicit input from a broader range of frontline employees in order to exhaust the breadth of characteristics associated with the interest area, e.g. perceived product quality. In the filtering phase, the goal is to narrow down the most relevant factors affecting the areas of interest for the respective organizations. In order to do so, participants were given a certain amount of virtual coins that they had to allocate among the factors developed in the previous phases. Participants could place different emphasis on each factor by investing a different amount of coins in each of them.

In the prediction phase, CIU is testing the judgmental accuracy of frontline employees over time  as they are asked to make predictions about their companies’ performance on critical issues. The predictions will then be compared to the actual performance metrics achieved by their company, which will illustrate how well employees can make forecasts about strategic issues using their knowledge from the frontlines.

Things to come

The following months will see the culmination of the research with the two Danish multinational companies. In that period, case studies and guidelines will be made available with our sponsoring partner, The Danish Industry Foundation, which may be used by industry decision-makers to understand the case-specific characteristics of the partner companies and the requirements necessary for implementation of crowd prediction methods in their own organizations. We will also be hosting our first virtual workshop with research partners and SMEs, where we will address the issue of scaling prediction platform design for SMEs. Lastly, as conditions would have it, our 3rd annual Crowd Predictions Conference has been postponed until Autumn ’20, which we hope to announce further developments on in the coming months.

Authors

Brian Christopher Pollstergaard
      Wladimir Santana Fernandes

 

Share this:

Leave a Reply

Your email address will not be published.