Is Your Forecasting Methodology Letting You Down?

Five ideas to improve forecasting accuracy.



Much of marketing is based on setting expectations (forecasts) and meeting these (sales achieved). Taking that a step further, the stock price is driven up or down based on whether the forecasts (when translated into revenue and profit expectations) are met and exceeded, or not met. Therefore, it is critically important to get it right, not only for our brand, but also for our company value.

Unfortunately, many large companies do not have a strong handle on this and the accuracy of forecasts range from poor to fair. Why is this? You would think it should not be too difficult. You make a reasonable estimation based on the evidence, and then all your teams focus should be on making this forecast happen within the time frame set. You could break it down into smaller time increments and then add them together. Why is it all going horribly wrong?

The reality is that it is a complex issue to develop a near accurate forecast as there are many variables to consider, especially in the pharmaceutical environment, in which we have seen more structural and environmental changes in the past 20 years than in any other industry. We have the impact of mergers and acquisitions on an ongoing basis; we have patent expiries; we have generic competition; we have competitors changing their actions; we have unexpected product withdrawals; we have lawsuits to settle; we have decreases in impact of our activities; we have a changing mix of activities; we have changes in the key message drivers depending on what clinical studies have come out; we have changing influences depending on where the product is in its lifecycle; we have changes in the key stakeholders and customer buying cycles to name but a few! Every one of these will have an impact on how much revenue our brand/s will create, and when.

In a nutshell, we have a very complex environment in which to forecast - and this is often only one of the priorities of the team tasked with the job.

In the past, it was usually the sales director who was heavily involved in the process, in addition to having many other issues to focus on training the sales team, ensuring the sales team have what they need to sell, managing the sales team, ensuring the sales team meet their numbers, and growing revenue. Often they asked the sales team to provide them with revenue they are going to commit to and then factor in a variety of other issues (often gut feel is part of this) and a forecast basis is born.

These days, most pharma companies have analytics departments to do the above. These teams collate a whole host of information and analyze the data and make forecasts. Unfortunately, the quality of these forecasts depends on the quality of the data they have to work with, which, on the whole, is not great. The majority of it is historical and by its very nature, given all the changes constantly going on in the environment, flawed to begin with. Despite having mathematical geniuses to work on the data, companies cannot understand why they cannot deliver the goods. GIGO (garbage in, garbage out). No matter how good they are, if they have poor data to work with, how can they be expected to come up with meaningful figures?

Okay so I am not making this look positive, am I? Well, it can be, but there are some very real reasons why companies struggle with creating accurate forecasts, especially if they are pointing fingers internally. Here are some ideas to improve your forecasting accuracy.

  1. Keep in mind how critical this job is. CEOs have lost their jobs due to this area being poorly managed and share prices plummet when expectations are not met. This needs the appropriate resources and revenue put behind it to make sure it gets done properly. It is a critical job that sometimes does not get the resources and focus it deserves.
  2. Get the right data. Historical data is interesting but not of great relevance in the pharma environment (e.g. Vioxx was strong and then was suddenly withdrawn). To really have relevant data to input into this process, it must be taken from the market at the time the forecast is created. Even then, not everything can be predicted, but at least the data will be much closer to being accurate than using irrelevant archaic data!
  3. Ensure that the same process is used for all brands. There must be a process employed for forecasting for all brands that follow the same principles. If one brand uses one method and another a different method, and then these are meshed together to create the overall forecast to give to the market analysts, is it any wonder companies do not meet these? Implemented correctly, once the process is consistent across all brands, trends can be factored in and probabilities examined.
  4. Implement continual measurement and accountability. Dont reward sloppy forecasting habits (e.g. massing the forecast to meet market expectations, forecasting what you think your boss wants to hear, being overly optimistic, being overly pessimistic, etc). Make sure you have consistency and then do continual measurement to ensure that you are accountable to that forecast; ensure each level of management that has input into the forecast is held accountable for their component.
  5. Invest in forecasting data and methodologies. In order to get more accurate forecasting, the right data must be collected and invested in. Dont just use the data you have if it is not enough. Get the right data!

The accuracy of the forecast is a critical issue which many senior level positions depend on being as close to accurate as possible. Given the level of importance of this job, it is critical to get the right data, the right resources and the right processes to ensure that you are at least giving this process as much importance as it deserves.

For any information on anything mentioned in this article, please contact Dr Andree K. Bates at Eularis, www.eularis.com.