Forecasting

The importance of quality forecasts cannot be understated. Accurate forecasts are crucial in pharmaceutical marketing for four primary reasons.

The importance of quality forecasts cannot be understated. Accurate forecasts are crucial in pharmaceutical marketing for four primary reasons. First, they are necessary to ensure that product is in stock and available to fill prescriptions without risk of committing too much value to stock that may ultimately run out of date. Second, the purpose of the marketing plan is to ask for money in order to carry out the specified tasks. Ultimately, a product managers or marketing managers credibility is dependent upon delivering the projected sales on time and within budget.

Third, from a corporate perspective, each country is required to submit market share forecasts and the reputation of the country manager and his team rests on the reliability of their projections. Finally, accurate forecasts of sales, timelines and profit form the basis on which investment managers value your companys shares. Forecasts, therefore, are critical to effective and efficient marketing planning.

So what can go wrong? Generally, failure is due to one or more of the following; (1) the data used for the forecast or the assumptions underpinning it may be wrong, (2) the statistical method used to arrive at the forecast may give spurious results or (3) the results may be right, but are interpreted wrongly.

Awash with data, devoid of information

In comparison to many others, the pharmaceutical industry is awash with data but devoid of information. For instance, pharma gathers vast quantities of information within its CRM systems, yet frequently fails to synthesize this data efficiently.  It also relies too heavily on external data sources. The first of these issues is extremely important when developing forecasts because companies need to choose the right data and to be aware of how to structure and manage it in order to analyze it effectively. The second issue often hinges on targeting.

Most companies use target lists obtained from an external vendor. These are used to assign potential and prioritize target customers for sales and marketing activities. But therein lies the problem. Because common lists and methods are used to derive targets, most of the industry spends its time chasing the same people. This has important implications for sales forecasting, because frequently our target doctors are subjected to phenomenal call pressure with the result that the vast majority of these calls are non-effective. In fact, given this scenario, non-target doctors represent much higher potential.

New research throws an even more interesting light on this targeting issue. Recent study results indicate that doctors can be accurately categorized into two key groups. Doctors are either dynamic in their prescribing, in which case they regularly take account of new products and respond positively to detailing and other promotional activities, or they are static. Static doctors stick to a very restricted choice of drugs and rarely respond positively to sales calls.

The proportion of static doctors is a critical statistic for both effective marketing and accurate forecasting. Current ongoing research indicates the static doctors constitute 60% of the GP population, which may account for a great deal of the waste that occurs with sales teams. Understanding this dynamic is critical to achieving sales targets and building a sustainable competitive advantage.

Choosing the right model

The second problem is choosing the right model for the data. Without doubt, this is a major issue in pharmaceuticals. The most common method is to use an Excel-based forecasting system utilizing linear regression, multiple regression or time series regression. All of these approaches are reliant on the data meeting three critical conditions. First, the residual value for each of the data points must be independent. Second, the variance (which is the spread of points on either side of the line) must be equal. And finally, the data has to have been largely drawn from the normal distribution and the relationship must be truly linear.

Experience shows that most pharmaceutical data violates at least two and often all four of these critical assumptions, which means that the results derived are largely meaningless.  To produce accurate and reliable forecasts from pharmaceutical data, simple methods just will not do and the tools needed are not generally even discussed until the doctoral level in most university curricula.

Poor analysis methods are a major reason why forecasts often fail to predict demand and marketing activities directed by return on investment analysis fail to produce the expected result.

Interpretation raises further issues. First, it needs to be clear at what level the forecast is being made. Bear in mind that national forecasts are based on the country average. So although reasonable as an estimate for total sales, the actual figures when broken down generally bear little resemblance to the correct figures at a local level, taking into account environmental and local marketing factors. This is increasingly important when looking at countries like Germany or the UK where local variation and local practice can markedly affect results.

The second issue is attempting to project beyond the capabilities of the data. As a rule of thumb, do not try to project any further than 30% beyond the current data. Commonly, the data will not fit a linear relationship and in these circumstances projecting as far as 30% can be extremely unwise. Generally, examining the 95% confidence limits surrounding a predicted forecast will reveal that towards the end of the projection, these limits may become very wide indeed. The analogy is a shot gun, where shot leaves the barrel tightly configured but rapidly spreads out.

In conclusion, accurate forecasts are critical for successful marketing and the achievement of commercial goals. But the strength of forecasts rely upon using the right methods and applying them correctly to the data.  New research, however, indicates that some of the key pivotal assumptions on which forecasts are often based should be revised.

Dr. Graham Leask is a member of the Economics and Strategy group at Aston University, where he specializes in the measurement and prediction of marketing activities.