Decisions are Made in Two Ways

Behavioral Economics is a game-changing idea if you are in the game of predicting decisions. By now most of us know about Thinking, Fast and Slow. Kahneman and Tversky’s groundbreaking idea (supported by lots of creative and compelling social science) is that most decisions are made instinctually or based on habit. Kahneman calls this “fast thinking”, which is easy, automatic, and intuitive.

There is also “slow thinking”, which is hard, conscious, and deliberative. It is easy to think of fast and slow thinking as separate systems, that you tackle a problem using either fast or slow thinking. However, the most powerful implications of Behavioral Economics, particularly for marketers, stem from the interplay of these two systems.  This is where behavior change starts and finishes. Specifically, to change ‘fast thinking’ in any way we must engage ‘slow thinking’.

Understanding the root causes to help define an initial step to disrupt behavior is definitively ‘slow thinking’. Any hope to change behavior requires this kind of conscious effort. The good news is sustained conscious effort (some say it takes 7 attempts) leads to a new habit, which lets you get back to ‘fast thinking.’

Implications for Changing Behavior with Marketing

Behavioral Economics clearly shows us that we need to understand how both ‘fast’ and ‘slow’ thinking relate to any specific behavior.  For example, to change a physician’s prescribing decision, we need to recognize the ‘fast’ and ‘slow’ thinking underpinning the physician’s original prescribing choice. With this insight, we can develop strategies to change this behavior.

To help this effort, we leverage data integration and advanced analytics to understand both fast and slow thinking, particularly with respect to primary and secondary data. In essence, primary data captures what customers say, which is our best view into ‘slow’ thinking. Secondary research, on the other hand, captures what customers do and is our best view into ‘fast’ thinking.

Therefore, the need to understand the relationship between ‘fast’ and ‘slow’ thinking prompts the need to combine primary and secondary data. We leverage state-of-the-art (but often underutilized) analytical tools that overcome these problems and provide a single set of cohesive insights from multiple different data sources. Our approach is analogous to combining results from clinical studies into meta-analyses.

Our Advanced Analytic Tools

One robust example of these kinds of tools is Bayesian Belief Networks (BBNs). With the computational resources that are now commercially available, BBNs can simultaneously evaluate literally thousands of interdependent probabilities to sort out the most important factors and relationships in individual or across multiple data sets. More specifically BBNs allow us to create links between networks, thereby allowing us to connect data that cannot be connected otherwise. In addition, a BBN can be directed to predict a particular decision or behavior. And in some cases, a BBN can also help parse out cause and effect.

To give you a concrete example, it is often helpful to evaluate the extent to which a ‘soft’ measure like quality-of-life (QOL) impacts a ‘hard’ measure like prescribing. Primary research is our only way to measure something like QOL, reflecting what HCPs ‘say’. And secondary data is our best measure of prescribing reflecting what HCPs ‘do’. Therefore, to best answer that question we must combine primary and secondary data.

Another example is trying to identify which HCPs have the greatest number of patients in need of treatment change or, ideally, have candidates for a new treatment. The common approach is to leverage either claims (i.e., secondary data) or chart data (i.e., primary data) to identify which HCPs have the biggest unmet need. The issue is these two data sources each provides a different perspective, where the best answer requires combining primary and secondary data.  By creating a BBN for each data source and then finding connection points to evaluate specific relationships between the data sources. The result is an integrated solution that allows us to simultaneously evaluate data that reflects both what HCPs ‘say’ and what HCPs ‘do’.

The Bottom Line

Leveraging behavioral economics using BBNs to combine primary and secondary data is one distinctive way we help our clients establish better answer mission-critical business questions. And while we are committed to expanding the use of advanced analytics, in the end, it is not about using fancy tools.  It’s about using the best tools available to generate better answers to our three critical business questions: What is our opportunity? Where do we focus? How do we win?

Want to drive better business decisions with advanced analytics and Behavioral Economics thinking? Please contact us and someone will be in touch with you right away.

Pharmaceutical companies have long used advanced technology to increase both the speed and likelihood of the success of their new product development. Think about how the mRNA platform is revolutionizing vaccine development. If pharmaceutical companies make use of sophisticated science to create the product, they can also use the available advanced analytics to sell the product.

Creating Actionable Segmentation Studies

Selling is fundamentally about deciding where to focus and that requires targeting the best prospective customers. The best targeting requires an actionable segmentation which:

1. connects what customers think towhat they do
2. produces identifiable segments

This requires the integration of primary data (what customers think) and secondary data (what customers do) and assigning each physician on a call list to a specific segmentation.

Why Segmentation Studies Fail

Segmentation studies often fail. In fact, most of our segmentation projects start with the question “how can we avoid the mistakes we made last time?”. Many segmentation studies fail at the implementation phase because implementing the results requires sales reps to administer a typing tool to assign HCPs on the call list to a particular segment. This is awkward, error-prone, and wastes precious time.  

Brabeuo Five-Step Process

Our five-step process combines primary and secondary data to assign HCPs on the call list to specific segments. Throughout the process we utilize our armamentarium of contemporary analytics to integrate and analyze data. Additionally, we apply our practical experience from implementing segmentation studies within a pharmaceutical company to think ahead and deliver truly actionable segmentation studies for its clients.

Our process includes:

1. Leveraging existing secondary data to HCP call list. The goal is to generate as much secondary data as possible for each HCP in the existing Call list.
   
2. Creating primary and secondary research connection points. We recruit respondents for the primary research exclusively from a client provided ‘call list’, focusing on those HCPs with sufficient secondary data. This ensures that every HCP taking the survey will have both primary and secondary data.
   
3. Generating segments exclusively with the primary data using contemporary ensemble method. This ensures that we are maximizing our ability to identify segments with fundamentally differing approaches to treatment, makes sure we have the most robust mathematically distinct segments. In the process, we explore a wide range of hypothesis drive solutions to ensure we identify the most useful segmentation scenario.
   
4. Developing an algorithm to assign HCPs to the different segments. We leverage the secondary data (match to the primary sample) in a Random Forest Model to increase the power, reliability, and predictive accuracy of segment assignments. This eliminates the need for a typing tool.  Instead, since the model is based exclusively on the secondary data, we can apply it to the entire call list to assign every HCPs to a segment.
   
5. Validating the results with people who have first-hand knowledge of the customers.  This can take the form of simply showing a sample of HCP segment assignments to the client’s leadership. Or, if more rigorous validation is required, we can invite HCPs, by segment, into additional interviews to see if they match their assignment and then test segment-specific tactics and messages.

The Bottom Line

After all the cutting-edge science, time, and money that gets poured into developing a new product, your product deserves the best chance at commercial success. Our extensive experiences, proven process, and contemporary analytics methods can help you make better decisions and drive your commercial success.

Do you want to learn more about how we can improve your targeting? Please contact us and someone will be in touch with you right away.

Winning in sports or business requires knowing and measuring the drivers of success. This is the lesson of Michael Lewis’s best-selling book, Moneyball. Moneyball tells the tale of how the Oakland A’s used advanced analytics to isolate the metric that could predict a baseball player’s potential to score runs.

For years, baseball teams unquestioningly relied on batting average to determine which players to hire. But using advanced analytics, the Oakland A’s determined that the On Base Percentage (OBP) is a better predictor of a player’s potential to score runs.

Once the A’s started basing hiring decisions on this metric, they began winning, most notably the American League division title. The A’s won because they knew what factors caused them to win.

Using Advanced Analytics to Determine Causation

Business results tend to be characterized into three broad buckets:

• Profit-How much money is the business making after costs
• Share-How well is the business doing relative to the market and its competitors?
• Customer Acquisition and Retention-Are customers staying or leaving?

These business outcomes are caused by commercial activities that generate specific metrics such as number of calls, customer satisfaction, customer churn, revenue, and costs. For example, increasing the number of calls on a particular medical specialty may drive higher market share, or not. Higher customer satisfaction metrics may or may not result in improved customer retention.

Contemporary advanced analytics can separate out the meaningful versus the interesting metrics. We use Bayesian Belief Networks (BBNs) to examine a wide range of interacting variables that may be impacting your business results. With BBNs, we can estimate the probability that a particular metric is actually causing your business results. Specifically, we leverage AWS Data Lakes tools to have the computational power to examine millions of data points from disparate sources like claims data, medical records, prescription data, and company-generated data to establish which activities are actually producing your business results.

The Brabeuo Process

We have a robust process to work with our clients to answer four questions regarding their Key Business Metrics (KBMs):

1. Are they simple to define and understand? Complexity often kills company-wide metrics plans because it makes alignment and implementation difficult.

• Can the KBMs be easily quantified? If you struggle to create a KBM, it will fall by the wayside.

• Can the KBMs be influenced? Not all metrics can be influenced by company initiatives. In some categories, like hemophilia, an inherited and easy to diagnose condition, the number of patients can’t be increased by a disease awareness campaign.

• Do your KBMS drive a key business objective? Have you gone beyond intuition to determine which metrics actually drive business results?

The Bottom Line

Armed with the knowledge of what actually drives their business (and equally important, what doesn’t), we inform tactical plans and strategies to produce better business results. We apply our unique suite of advanced analytic tools and decades of experience to determine the metrics that are responsible for moving your business. Measuring what matters directly informs how to win.

Are you interested in discussing what drives the success of your business? Please contact us and someone will be in touch with you right away.

Accurately estimating demand, either for a new product or new product features, is essential for several critical business decisions. It drives Go/No Go decisions, the design of clinical trials, leveraging key and differentiating demand drivers into marketing and communication materials, and sales force sizing and organizations. Therefore, estimating demand is the most fundamental element of establishing the scope and size of an opportunity.

We have developed new product demand prediction for most therapeutic categories, in both rare disease and primary care. We can report with confidence that there are two keys to more accurately predicting new product demand.

One: Test a Full Range of Variables and Scenarios

We strongly advocate testing a full range of variables. Markets are complex and can’t easily be described using a few variables. To accurately predict new product demand, you need to replicate the market. This requires testing all relevant marketplace attributes in your research.

The good news is contemporary estimation methods do not limit what can be tested. So many of our clients believe methods confine or limit what can be tested in research on new product demand. The convention, they think, is they can only test 8 to 10 product attributes in a limited number of alternative product profiles. 20 years ago, this was the case. Things have changed! With the contemporary analytic methods, it is feasible to test the full range of variables that could impact product demand. In fact, now we concern ourselves with the real threat to a weak forecast. If we constrain the elements we test, we may fail to account for a factor that could dramatically alter your forecast.

For example, clients often want to just test the details of their product and avoid testing the potential competitive responses to keep things simple. But isn’t likely that an existing competitor will have an extensive counter-detailing initiative? What will happen when a competitive product goes generic? Or what if a competitive product receives an additional indication? Beyond competition, other marketplace factors will continue to evolve such as the standard of care and Payor practices.

However, just because we can test an almost unlimited number of variables, it doesn’t mean we should. At the end of the day, there is a real human being, a doctor, or a patient, at the receiving end of any survey we design. At Brabeuo, we maximize the number of variables we can test with robust contemporary methods and intuitive survey design. We employ graphic designers and world-class computer programmers to construct surveys that are easy to navigate, modeling our approach on familiar e-commerce sites. This way we minimize respondent fatigue and maximize the reliability of the responses.

Two: Focus on Specific Patient Profiles

To replicate what happens in real life, we focus on specific patient profiles, not broad patient types. Marketers like to think in broad patient types, whether they be age-related (Pediatric/Adult/Seniors) or disease-based (mild/moderate/severe). But that’s not how doctors think. In real life, doctors evaluate product trade-offs (i.e., efficacy, safety parameters, dosing regimens, etc.) for specific, individual patients. In new product demand forecasting, it is important to assess HCP perceptions on how a specific attribute would impact individual patients and/or caregivers. For example, what impact does an HCP think a monthly injection would have on patient acceptance versus an oral taken TID? To really answer that question, they have to think about an individual patient. With the advanced analytics Brabeuo employs, it is easy to systematically vary the details about a patient, much like we do about a product, to produce a highly detailed patient profile.

It’s also important to consider how patient preferences evolve over time. By the time your new drug is launched, will patients be thinking differently about efficacy, safety, side-effect trade-offs? Markets evolve. HIV patients were initially focused on the lifesaving properties of new drugs. However, as their life expectancies have grown, they are increasingly focused on long-term safety and convenience.

Lifecycle Simulation Tool

An added benefit to the more detailed and comprehensive Brabeuo approach is that it can produce a simulation tool that can account for the new events as they evolve over time and after launch. There is no need to do new research or create a new tool. Instead, you can just adjust the inputs into the existing Brabeuo model. The more comprehensive Brabeuo approach ends up saving clients time and money over the long run.

The Bottom Line

Using contemporary advanced analytics and testing comprehensive and realistic market scenarios allow us to model out a wide range of future market scenarios. If you can imagine it, we can model it. With our contemporary methods, we can integrate a wide variety of data resources to produce better insights. The result is a better forecast for more informed decision-making in the future.

Want to improve your new product demand forecasting? Please contact us and someone will be in touch with you right away.