By Faruk Abdullah of Applied Predictive Technologies
Pharmaceutical marketing is changing. With increased industry spend, new marketing channels, a more informed consumer, and a shifting healthcare landscape, it is clear that marketers need to be smarter than ever with their investment decisions. It is also clear that DTC marketing is no longer just about pumping money into national TV ad buys (though they continue to receive substantial investment). It’s now also about targeted messaging. It’s about digital. It’s about patient engagement. In this environment, it is more complex than ever to understand how to get the right message in front of the right consumer at the right time. Marketers will need to innovate.
As organizations try new marketing strategies, they will inevitably realize that not all of their ideas will achieve the desired outcome. In fact, it is incredibly risky to dive head-first into new ideas without empirically validating them first.
The question is how can marketers get the most accurate, data-driven recommendations about which actions work with each customer segment? And, how can they do this without first risking significant budget or opportunity cost of not rolling out the most effective ideas sooner? The traditional method of allocating marketing spend has been to use promotion response models, which rely on historical data to attribute the impact of a given marketing action (e.g., print ads) on KPIs (e.g., NRx). Promotion response models are important tools for the industry and can serve as great sources of hypotheses about what the impact of a given action might be. However, traditional regression-based approaches are unable to uncover what would have happened if the action had not been taken.
It is time for life sciences marketers to move towards a process that leading retailers, restaurants, manufacturers, and banks have been driving for 15 years: rapidly testing their ideas to figure out what works and where they work best before making significant investments. As pioneers of controlled experiments for clinical trials, life sciences companies know that test versus control experimentation is the gold standard in analytics. Trying an initiative with a test group (e.g., in just some markets or with some physicians) and comparing those results with a highly similar control group (not experiencing a change) is the only way to truly understand the incremental effectiveness of each investment. Leading commercial organizations are now also beginning to use this experimental methodology to optimize their marketing and sales programs.
Experimentation is most valuable where the outcome of the experiment will directly impact a decision and create a new learning for the organization. With the consumerization of healthcare, consolidating health systems, new competitors, and a shifting reimbursement paradigm, the outcome of a new marketing action today is highly uncertain. We suggest that organizations should test all shifts in marketing strategy before making any changes.
There are three critical reasons why marketers should begin to incorporate credible, empirical data into their marketing decision-making process.
Correlation is not causation. Promotion response models seek to identify relationships between marketing actions and KPIs. With rich data and complex equations, it’s easy to conclude that this approach leads to optimal recommendations. Unfortunately, because promotion response models are generally not based on test vs. control analytics, they are unable to isolate cause-and-effect relationships between changing a given marketing lever (e.g., increasing digital spend) and a change in KPIs (e.g., TRx).
The future is different than the past. Companies construct promotion response models on the basis that historical relationships between marketing actions and changes in KPIs will continue to hold true in the future. As fast as healthcare is changing today, relying on data from the past to make decisions about a volatile future is a mistake.
It can’t be modeled if it hasn’t been done. Promotion response models rely on measuring the relationship between past marketing investments and KPIs. To state the obvious, there is no real world data from an action that has never been put in market, and thus no way to build a model on that action. For these situations, marketers may rely on market research or analysis of similar campaigns before rolling out a new tactic. However, these approaches may not generate the most actionable, accurate insights. Alternatively, by embracing rapid, statistically credible, field experiments, marketers can know with confidence what will happen before they take the risk of rolling it out more broadly.
In today’s changing healthcare and marketing environment, it is critical that marketers truly know which ideas work and which do not. Organizations should rapidly tests their ideas, discard the unsuccessful ones, and understand how to refine their strategies to dial up ROI.