This article describes the various steps that General Motors (GM) has taken to improve forecasts for better decision making. These include reducing waste, using consistent methods and data to see what is working and what is not, employing more statistical tools and less judgment, and encouraging employees to share their learning with others. The forecasting team was also encouraged to constructively challenge analysis rather than accept anything at face value. Above all, the initiatives created a culture where senior leaders became champions of our forecasting efforts.

General Motors’ spiral into bankruptcy in 2009 and post-bankruptcy reorganization reduced the company’s Long-Term Forecasting group to no more than a group of administrators, or bookkeepers. The department suffered from a lack of engagement with its employees and lack of trust from management. The company’s long-term business plan continued to show the famous “hockey stick”, built more on hopes and dreams and less on reality. As GM continued to miss targets, management overrides of forecasts increased and investment in forecasting tools decreased.

The Challenge

I took over the Long-Term Forecasting group in 2016 with a simple goal of answering the following question: “How to turn the group into a Valued Business Partner? ” Reflecting on my own personal toolbox, I sought to reduce the problem to my core beliefs. I am an engineer, not an economist. I have a passion for manufacturing systems, as well as for human decision making. In challenging the team to improve, my first step was to pull back the curtain and remove the mystery from forecasting. Utilizing my manufacturing background and following the Toyota Production System (taken from Decoding the DNA of the Toyota Production System by Steven Spear and H. Kent Bowen, Harvard Business Review, September 1999), I began the journey of adopting lean principles into the knowledge work of forecasting. For that, I followed six steps:

  1. Continue to root out all waste
  2. Strive to make tacit knowledge explicit
  3. Specify how workers communicate with one another
  4. Use scientific methods to solve problems
  5. Plan for an incremental journey
  6. Leaders must blaze the trail

Continue To Root Out All Waste

Study after study points to analysts’ failure to make the most of information available to them. In our Long-term Forecasting group, we eliminate work which does not require judgement or expertise, which eats up a huge amount of analysts’ time. Knowledge workers underestimate the amount of inefficiency in their jobs, and we took time to teach everyone “the five whys” to help us get to the root cause of every activity performed. Additionally, the leadership team regularly reviews the structure and content of every job. One of the most significant changes is the implementation of a data center which gets data in the hands of our customers and alleviates the non-stop “10 minute” data pulls requested by analysts. As a result, the analysts have the time to gather valuable inputs for the task at hand and at the same time avoid doing non-value added work.

Strive To Make Tacit Knowledge Explicit

As we started on our lean journey, one key issue that quickly surfaced was that the group used inconsistent methods and approaches. The tacit knowledge held by forecasters was neither transferred to new employees nor held up as best practices. To gain credibility with leadership, we had to use the same data consistently. Over the course of a couple of weeks, we asked the team to document exactly how it performs each task—specifically, how it uses different forecasting inputs like pricing, research data, customer survey data, portfolio assessments, past performance, brand health, vehicle specifications, and competitor intelligence. Only after we specified them did we start to take steps to improve them. The implementation of forecasting standards had to eliminate noise in our analysis and help to correct true errors.

Specify How Workers Communicate With One Another

GM has over 200,000 employees, making communication across departments and divisions difficult. We challenged our employees to understand their networks and understand how to communicate with one another.

Additionally, the forecasting group set out to challenge one another. We hold weekly constructive confrontation meeting to stress test our analysis and push the team to seriously consider alternative possibilities. Forecast analysts ought to review not only the results of their analysis but also the methods and assumptions underlying them. The goal is to root out noise in our forecasts and push the forecasters to think about many plausible alternatives. Increased awareness and utilization of employee networks is evidenced by a reduced number of “fire drills” and reduction in re-work. Constructive confrontation has removed “blind spots” in our analysis and contributed to more robust analysis.

Use Scientific Method

People tend to revise their beliefs by over-weighting newer information and under-weighting prior and longer-term information. Predictions and decisions generated by simple statistical algorithms are more accurate than those made by experts, even when experts have more information than the formulas used. A key advantage for algorithms is that they are noise-free and always give the same output for the same input. We anchor our forecasts by employing statistical forecast tools, and frame forecast risk not only in an individual vehicle but also in its family. Our analysis goes through the history of forecast errors to determine where we can add forecast value for continuous improvement. Additionally, we are looking outside the auto industry to understand technology adoption and applying technology diffusion models for disruptive trends.

Plan For Incremental Journey

The Long-Term Forecasting team codifies lessons learned and drives a continuous improvement culture. We have established a vision and mission for the department:

  • Vision: Deliver robust, data driven market analysis and sales forecasts.
  • Mission: Support GM business through the development of long-term forecasts which reflect the current state of business, competitive environment, pricing strategy, and the future state of the industry.

Most importantly, we take time every other Friday to review our work on continuous improvement items. Through a small, focused list for continuous improvement items, we are taking the incremental steps necessary to reach our vision.

Leaders Must Blaze The Trail

Most important to driving change, leaders must blaze the trail. The primary role of every supervisor, manager, executive and CEO is to link people to their purpose. Senior leaders are expected to be long-term champions, and managers are expected to motivate their teams. We are focusing on the human side of the equation as well to drive results. When people see their work as important, they are willing to give their discretionary energy. Our journey is not over; we will continue to grow everyday into a valued business partner of General Motors.

This article originally appeared in the Fall 2018 issue of the Journal of Business Forecasting