Pharma Forecasting Tips

Pharmaceutical Marketing Coach:

Forecasting is one of the most difficut exercise of a Marketer,but also one of the most excting Being under sales target by even just 1% meant millions of € at risk. Being over sales target wasn’t much better, as this exposed the brand to a potential backorder situation, and backorders lead to a ton of other problems. Some forecasting situations are much easier than others, and some are very complex. Imagine being the forecaster of a newly launched product that is in a new therapeutic area, or a brand that is breaking into a therapeutic area where there have been several recent recalls. There are so many assumptions to consider for your forecast, and finding the right analog as a base for your forecast is not always easy.

Since forecasting is so critical in ensuring a brand’s success, Here are some forecasting tips with my website visitors. See Rob`s comments in the following pages to learn more about pharmaceutical forecasting:

Click here for an easy-view printer-friendly copy of the "Pharma Forecasting Tips":

Rob MacLennan, MBA, MA, has experience working for several multinational pharmaceutical companies and a leading international market research company. His decision to start PharmAnalytics was based on a clear need for an experienced firm to provide more accurate modeling solutions for pharmaceutical clients. Over the years he has led over 100 assignments from a diverse client group and thoroughly enjoys the challenge each new project brings.

You can reach Rob at rob@pharmanalyticslive.com .

Rob MacLennan (RM): Without a doubt, the two biggest reasons why pharma company professionals need help with their forecasting is because they lack either the time or the skills needed to build a robust sales forecast. From a time perspective, we estimate that preparing a forecast model from scratch can take anywhere from 10 to 20 days. With increasing demands on work schedules, finding the time to prepare the forecast becomes a big challenge.

Even if time is not an issue, marketers are often faced with a skillset challenge when it comes to forecasting. Many are presented with the task of developing a brand forecast with little or no background or training in this area. It is often said that forecasting is part art and part science. The science part includes some experience with statistical forecasting methodologies, while the art part involves such things as knowing which analogues to select, how to design a robust and user-friendly forecast model. The forecasting challenge is particularly acute when facing a situation with a great deal of market uncertainty, such as trying to forecast a new product or trying to model the impact of new competitors or new indications.

RM: Many companies have different corporate structures. Whatever the corporate structure, it is useful to do some preliminary work to set out the plan of attack. This essentially scopes the forecasting project, providing details on what the final ‘product’ should look like and who in the company that will need to be consulted before the forecasting project is completed.

The first step in forecasting often involves a detailed analysis of the historical market data. Ideally, you will want to go back at least 10 years and examine monthly data and try to develop a good understanding of the market dynamics. This is useful when developing analogs for future events. However, to gauge the appropriateness of these analogs, it is useful to speak to someone in the company that has some detailed insights into the market dynamics.

Following the data analysis exercise, and once you have a solid understanding of historical market fluctuations, speak to various people within the company who can speak to future market events. This includes identifying which events are expected and then trying to model each event individually. Basically, you are seeking some insights into such future market events as new competitor products, changes in reimbursement status, changes in product formulation or new indications. These events become your forecast assumptions.

The entire forecasting process works best when company professionals develop a detailed plan which clearly sets out the key deliverables and identifies on a task-by-task basis the key steps needed to achieve these deliverables. The list of potential contacts may be long, but you should consult employees within the Sales, Marketing, Market Research/Market Intelligence, Government Affairs, Market Reimbursement, Medical and Finance departments. The more departments and personnel are involved in the development of future event assumptions for the brand’s forecast, the greater the chance to gain buy-in for your forecast.

RM: Key forecasting assumptions revolve around anticipating future market events and modeling the impact of these events on the marketer’s brand. Typically, you can look at about six different kinds of events that can have an impact on the brand’s sales. The forecasters need to identify which events are likely to happen in the foreseeable future. They don’t need to be able to quantify the impact of these events, but simply to identify them.

The list of potential market events include:

• New competitors entering the market;
• New product indication;
• New product formulation;
• Change in reimbursement status;
• Generic entry; or
• Change in medical guidelines/publication of major study.

Once the list of events are identified for the market and brand, a lot of work needs to be done on modeling these events. At PharmAnalytics, we try to build a time element into our models which allows the user to change the timing of the event. This can save a lot of time since the expectation surrounding the timing of market events tend to change over time.

RM: The decision on which type of model to use depends largely on the availability and richness of the data at hand. If there is a long history of prescription volume data and many historical analogs to draw upon, then it might be best to go with a prescription-based model. If, on the other hand, there is little data – such as in the case of a hospital-based product – or where historical analogs are lacking, you might want to pursue the patient-based approach.

For the most part, it is preferable to build prescription-based models which rely on historical analogs. The reasons for this are twofold: prescription models provide a level of statistical accuracy that cannot be attained using patient models; and secondly, you can -often look back in history using prescription modeling to provide reasonably good estimates of future events.

The downside of prescription-based models is that they lack the ability to provide insights on patients. For example, a prescription-based model might indicate that a brand’s prescription volumes has increased by 10% over last year, however, the model does not indicate where this new business came from. Was it sourced from new patients entering the market or from patients switching from a competitor brand? Similarly, patient models can provide such insights as patient diagnosis rates or drug compliance that is not available using a prescription-based approach. So, depending on the forecasting objectives, the forecaster may need to sacrifice statistical accuracy for a model which provides valuable patient insights.

RM: Sales forecasting provides the basis for planning the future of the company. Once all the brand forecasts are developed, senior management can determine how to apportion the marketing and sales budgets. Being over or under budget means an improper allocation of resources which can lead to lost sales opportunities.

A forecast that is too low or too high can also have consequences for sales representatives who rely on sales performance bonuses for a significant portion of their total compensation. For example, a sales rep might decide to leave a company if he/she feels that the sales bonus was inadequate.