A recent study has shown that many sales projections for new pharma products are glaringly inaccurate. The study of 1,700 forecasts for 260 drugs found actual peak sales differing by 71 percent from predictions a year before launch; and that many forecasts overstated projections by more than 160 percent. The data also showed that even six years after launch, forecasts were 45 percent off from actual results.
While it is difficult to predict initial demand for a product – especially in a completely new therapeutic area – the inaccuracies found in post-launch forecasts prove that we need to do better. Fortunately, the current generation of forecasting technology makes it possible to dramatically increase accuracy – but only if pharma companies are also willing to change their workflow. That includes re-examining fundamental assumptions around the forecasting processes and the metrics ‘success’ is measured by.
Every new technology implementation requires a complementary culture change to make it work. Here are five things great forecasting teams can do to get the most out these new investments:
1. Focus on patients, not dollars. Many life sciences forecasters have had to simplify their assumptions and calculations in order to get to a dollar estimate. These forecasts calculate how many patients will take a drug overall, then convert that to volume and revenue. But key pieces of the patient journey with high downstream impact may be left out:
- When do patients start using a drug?
- How long they are likely to stay on a therapy?
- What would make them quit or miss doses?
- How do demographics and comorbidities affect attrition?
- How do these trends vary across countries, regions or populations?
New forecasting platforms provide better handling of real-time patient level data as well as the often-complex patient journey through therapy. This makes it possible to understand not just what patients are doing but why. Commercial forecasts that incorporate these detailed journeys deliver more accurate and detailed forecasts.
2. Think about the market, not just your product. Historically, forecasts are developed by researching and making assumptions for your own product, plus potentially for the current standard of care. But with broader access to vast amounts of data and more powerful forecasting tools, it is becoming increasingly useful to view a product in the context of the entire marketplace. This requires a paradigm shift for forecasters, requiring them to look more outward at the growth and share trade-offs between competitors, not just inward at the research on their own product.
To create this market forecast, forecasters start by defining all the global patients for the target market, all of the treatments currently on the market, and what market share each competitor owns. They then factor in how the market is likely to evolve based on factors such as patient population growth, key market events, innovations in development, emerging competitor products, and changing payer positions and standards of care. And finally, they assess how external events, such as government policy, are likely to affect the market dynamics.
This approach better places the individual product in the context of the entire market and allows forecasters to calibrate their own assumptions by modeling the impact on other players.
3. Reduce transcription and focus on data quality. More than 80 percent of the forecast teams we work with today use Excel to manually copy and paste data from a wide variety of sources into their models. Some are even hand-transcribed from websites and PDFs – a slow and cumbersome process that’s prone to mistakes. That means forecasters spend a lot of time making sure the data they use isn’t full of errors and empty fields.
But highly curated global data sources are now readily available and can be ‘plugged in’ to forecast models to eliminate this time-consuming work. The best forecasts today leverage these data sets and even compare results from different datasets simultaneously to generate consensus forecasts. This eliminates transcription and provides better quality data to drive more accurate forecasts.
4. Review your data flow. The flow of data, which comes at different times from different sources, can feel overwhelming if you don’t know how to use it. To accommodate the sudden ability to analyze huge disparate data sets, forecasters need to rethink their workflow, including how data flows into their system, how it is verified, and how the system will grab the right data at the right time to integrate it into the analysis process.
Quarterly updates on epidemiology, capacity forecasts from supply chain, and the latest demand assumptions from market research can all be automatically imported into the forecast so that the forecaster can focus on the implications. This used to require a complex integration effort costing tens or hundreds of thousands of dollars and taking months to complete. But thanks to advances in connected forecasting systems, it’s become a simple effort that can be done in a matter of days.
5. Build momentum through incremental success. Any time you introduce a new process or technology into an organization, it requires behavior change. If you don’t establish a change management strategy to accompany the implementation, it is much more likely to fail. To ensure the new system is embraced:
- Build a roadmap for rolling-out the technology that includes training, marketing and measures of success
- Identify senior- and mid- level champions to actively promote the value it creates
- Pilot the platform with a select group of users to prove it works, fix the kinks, and get people excited about what it can do
- Identify early adopters to communicate the value of the platform and train their peers
- Share early success stories with anyone who will listen. The more people are talking about what this technology can do the better
- Don’t over-sell it. Business leaders are weary of the hype around new technology, so be honest about what you can accomplish, and what value you expect to achieve
Remember, the latest generation of forecasting technology can provide incredible insights, but you must be willing to adapt your workflow, and way of thinking, to make it work for you.