It is widely acknowledged that the right demand plan can balance inventory levels with costs and lead to positive cash flow and higher customer satisfaction. Therefore, more and more companies are focusing their resourcing on achieving demand planning excellence in order for their business to be successful. However, the art and science of forecasting demand is often misunderstood and lacks the attention it requires. 

Achieving Demand Planning Excellence In Medical Devices

This is no different for the medical device industry, which is one of the biggest industries in healthcare, and is driven by innovation and new technologies. There has been an unprecedented growth in innovative and improved technologies in the last decade, resulting in dramatic advancements in medical devices. My company, Stryker, a Fortune 500 medical implant and equipment company based in Michigan, has been keen to capitalize on these changes through better forecasting and demand planning. The changes we have made have reaped significant rewards in our medical devices industry, but are applicable to all industries.

Medical Device Industry Has Shifted, And So Must We

According to recent analysis of the medical device industry, there has been a shift in the manufacturer – buyer dynamics, because physicians now tend to choose to be employed by big hospitals rather than owning their own practices. The result is that hospitals and insurance groups now have influence over medical device buying decisions, forcing a change in how we plan for demand. If we as demand planners and forecasters want to take advantage of this buoyant market sparked by new product innovation, we must navigate the following developments:

  1. Patients becoming more informed consumers
  2. Growth of structured quality measures
  3. Revenue-driving consolidation
  4. New and alternative provider payment models
  5. Information technology innovations driving inter-stakeholder communications

All of these have had a catalytic impact on demand and has made Stryker turn its eye to benefits of centralized forecasting and the crucial role this plays in decision-making, and how accurate forecasts can bring important benefits to the organization.

How Stryker Leverages Data Analysis Profiling

Data Analysis Profiling is one of the most important components of our centralized demand planning process as it enhances the collaboration between the forecasting and demand planning teams. It enables both sides to focus on combinations that increase forecast accuracy.

Data Analysis Saves Time, Makes Demand Planners More Productive

We have experienced that implementing focused groups for data analysis, specifically for the statistical forecast in combination with strong forecasting tools, brings great results as it helps improve the demand signal that goes to the demand planners for additional inputs. The statistical-focused groups also support the data profiling process to help demand planners identify items that are highly forecastable, thus reducing time spent reviewing them. By segmenting, demand planners do not need to monitor each item every time. They can also identify items that need more attention due to high variability, intermittency or order gap.

The Advantages of Using Data Analysis Profiling Are Clear


  • Clear visibility on the combinations which cannot be statistically forecastable and need business intelligence overrides.


  • Easy to use both in the current system and in an offline report which can be sliced and diced at any planning level required by the demand planners.


  • Has significantly decreased the total statistical and demand final errors.


  • Has increased the trust in the statistical forecast.

How We Overcame Resistance To Centralized Forecasting

This data analysis profiling is part of a wider shift to centralized forecasting. There has been resistance to this from demand planners as their traditional ways of forecasting were trusted, and to be fair they were delivering satisfactory (albeit uninspiring) results. They did not have much faith in the new centralized statistical forecast. Since we embarked on this journey, however, we have seen significant improvement in the statistical forecast and subsequently gained acceptance from demand planners and the business.

Gaining cross-functional acceptance and buy-in means using inputs from different functions, including judgemental forecasts from Sales and Marketing. These judgemental forecasts are still being applied on top of our forecasts as before. Whilst they may reduce forecast error under certain circumstances, they are inherently biased given the nature of human behavior. We strongly believe in using statistical forecasting to drive continuous improvement, adding ‘balance’ to the judgemental forecast whilst still benefiting from this insight.

Performance is The Best Driver of Acceptance

Furthermore, we’ve driven acceptance of centralized forecasting and new methodology by improving forecast accuracy and allowing demand planners to deliver serious value. Using data profiling to identify growth opportunities and cost saving measures speaks for itself. As forecasters and planners, we know that one of the critical elements underpinning a company’s growth, and indeed a mature demand panning environment, is a focus on high value add activities. Stryker’s data profiling methodology helps forecast modelers and demand planners to identify and analyze the highly forecastable combinations in relation to volatility, intermittency and order gap variability for SKU’s with high absolute forecast error. Adopting this methodology has made demand planners more effective, helping us to overcome any resistance to change.

Practical Tips For Those Looking To Implement Data Analysis Profiling

  1. Start with the easiest component of forecastability which is demand volatility or variability
  2. We have chosen a 40% threshold which we consider forecastable
  3. Demand planners should be instructed to review only those combinations with variability above 40%, high volume and forecast error higher than their targets. For the rest of the SKU’s the system should be generating a satisfactory statistical forecast
  4. Business intelligence should be applied where needed.
  5. Explore the value add strategy where you compare the value of the statistical forecast compared to a naïve forecast, and the value that demand overrides bring compared to statistical forecast. Then assess which of the 3 demand forecasts is more accurate.