Is Forecasting Futile?

Overestimate demand for your new drug and you could spend millions on a plant that’s grossly underutilized; underestimate, and your lack of inventory could result in delays and lost sales. Demand forecasting is crucial when it comes to making good manufacturing decisions. Given all that’s at stake, how effective are pharma companies at forecasting demand for new products? Well, according to a survey from ORC International: not very effective at all (1).

Drug forecasts combine scientific, clinical, regulatory and commercial data, but conversion of that data into useful information is difficult. In a survey of 50 pharma industry senior managers, researchers found that over 60 percent of drugs forecasted are either over or underestimated by more than 40 percent – and a substantial number of companies are overestimating peak revenues by 160 percent or more. Nearly all survey respondents claimed that unused or underutilized facilities existed in their network, although underutilization was generally below 25 percent.

Why? According to the report, overestimated demand was typically caused by unexpected market volatility or simple optimism. In the case of underestimation, lack of background data to support forecasting information was usual suspect.

“Demand forecasting is hard because companies need to start pursuing capacity for manufacturing a new drug 3 or 4 years before it is actually needed,” says Joe Principe, Vice President of Strategic Partnerships at Patheon, who sponsored the study. “There’s no crystal ball good enough to understand the future and companies, understandably, can get it wrong. It’s not really their fault – most are using forecasting models to the best of their abilities.”

The survey found that many companies react to the problem by investing more in forecasting tools or by outsourcing, as opposed to building in-house capacity based on potentially incorrect demand forecasts. Is there a better way?

According to the report, around 400 new products are likely to be launched in the next three years – 60 percent of which will require unique manufacturing processes. Increased complexity coupled with inaccurate demand forecasts sounds like a potentially bad headache. And if forecasts cannot be improved, coping more efficiently with demand variability is the only pill to swallow. Flexible manufacturing options exist that could mitigate the need to build a new plant – so do these uncertain times represent a tipping point for their uptake?