Pharma forecasting: How to fine tune market research
Todd Johnson of Kantar Health outlines how market research can be honed into reliable data to improve forecastsBy Jan 19, 2012 on
There’s a curious fact about primary market research: Typically, it’s wrong.
It does, of course, provide some bearing on how physicians perceive your product and plan to prescribe it once it’s launched. But as a measure of future market share, time and again primary market research paints an inaccurate picture. Indeed, the discrepancy between preference share (the market share that primary market research estimates) and actual market share is often 50 percent or higher.
There are a number of reasons for this discrepancy, according to Todd Johnson, senior director of forecasting at Kantar Health. For one, during primary market research, physicians are forecasting their future plans based on bullet points in a survey. “It’s very difficult to meet the reality of what’s going to happen two or three years down the line, or maybe even a year from now,” Johnson said at eyeforpharma’s Pharma Forecasting Excellence conference in Berlin.
Secondly, primary market research generally assumes equal access, whereas in the real world different tiers, agencies, and organizations can restrict access, and affordability weighs more heavily on prescribing behavior.
Furthermore, primary market research overlooks feedback mechanisms; if physicians start to prescribe a drug and receive negative feedback, that will impact their future prescribing. It also ignores counter detailing; once a product launches, sales reps from pharma competitors will present arguments on behalf of existing drugs. “As good as market research is, no matter how large your sample size or how rigorous your approach, it’s still a contrived exercise,” said Johnson. “It’s not reality.”
Johnson stops short of arguing that companies should cease primary market research altogether. As he points out, market share could theoretically be anywhere from zero to 100 percent. That’s a wide range, and primary market research helps narrow the gap.
Still, forecasters must take additional steps to hone this research into reliable data entries for their forecasts. “If we are going to spend the time and effort on the market research, we should spend an equal amount of time and effort on making the adjustments,” Johnson said.
The standard approach to adjust primary market research is to use a rule of thumb, wherein forecasters dock a standard amount like 33.3 or 50 percent from preference share so that it more closely resembles market share. Johnson argues against this approach because forecasters who use a rule of thumb almost always end up changing it based on a given situation. “They’ll look at it and say, ‘That’s too much or too little,’ which then makes it just as subjective as if they hadn’t had a standard rule of thumb to begin with,” Johnson said. And this subjectivity can have big implications for forecasted revenue.
Let’s say a preference share of 20 percent yields estimated revenue of $600 million. If a company’s rule of thumb is to decrease that preference share by 33.3 percent, that means market share will yield revenue of $400 million. If a forecaster decides to increase that rule of thumb to 50 percent based on the specific situation, revenue drops to $300 million. That’s a $100 million difference in projected revenue based upon a subjective adjustment.
“I would argue that unless you have some really sound support behind a rule of thumb [like a normative database that provides historical trends], it may not be a good idea to always apply a set percentage,” Johnson said.
If companies insist on sticking to a rule of thumb, Johnson recommends that they at least standardize the way they adjust it, with middle range preference shares regularly getting higher adjustment rates than high or low preference shares. “Physicians are good at picking out the stars,” Johnson said. “If it’s a good product, they will use a lot of it. On the other hand, if it’s a dog, a dog is a dog and they will probably not use a whole lot of it.”
In the middle ground, however, they have a more difficult time making decisions due to the parity among drugs, and follow-through rates will be consistently lower than anticipated.
Verbal probability scales
A good alternative to a rule of thumb is a scale that correlates verbal assertions to internal commitments, like the Juster scale. In the wake of World War II, the US Federal Reserve began polling the public’s intent and likeliness to buy durable goods like TVs and cars. The results were used to predict what the financial status of the US economy would be in the coming years.
Gradually, a correlation began to emerge between verbal commitments and an individual’s internal commitment to follow through on that intent. In 1966, an academic named F. Thomas Juster conducted further research on the subject and developed a comprehensive scale to adjust for the disparity between words and actions, intent and follow-through.
The Juster Scale is an 11-point scale with specific follow-through percentages linked to each point. For instance, individuals who give a score of eight or more have a better than 50 percent chance of following through on their intent. “This makes a lot of sense because my ‘very probable’ may be very different than your ‘very probable’,” said Johnson. “But if we assign a number to it, we can relate those things, so people have a better chance of being accurate with their responses.”
Granted, durable goods like cars and TVs are a far cry from drugs, but Johnson argues that forecasters can use the essence of the scale to create tailored algorithms that serve as standardized proxies for prescribing. Due to the elasticity of these algorithms, the adjustment for different preference shares changes in a standardized way, but not in a linear standardized way as a rule of thumb does. “When we use something like the Juster scale, we weed out those physicians who are not going to prescribe at all by using these probabilities” to arrive at market share, Johnson said. “We not only look at the depth of interest but the breadth of interest.”
Awareness, trial and repeat purchase models
Another scale that forecasters can use as a template for preference share adjustment is the Fourt-Woodlock equation. This scale emerged from research on consumer packaged goods. It estimates the total volume of purchases per year based on individuals who make trial purchases and repeat purchases of a product within the first year.
Consumer packaged goods are closer proxies for the pharmaceutical industry because consumers buy goods like cereals and pastas more regularly than they buy cars or TVs, just as physicians interact with prescriptions on a regular basis. The difference is that consumer packaged goods incorporate more TV and print advertising, sampling, and couponing, whereas pharma marketing primarily involves sales rep detailing.
Therefore, Johnson says, forecasters still need to create a pharma-specific model from the standard equation that predicts physician awareness. They also need to take into account which country they’re measuring, because the impact of promotion is higher in some countries than in others. “This isn’t the whole adjustment, but it gives you a good read on what a company’s efforts will do to impact follow-through,” Johnson said. “If you can adjust your promotion depending on how many sales details, how much journal advertising, that’s going to affect the awareness and potentially the follow through from the physicians.”
If forecasters want to account for the overstatement inherent in primary market research and run different scenarios based on their promotional assumptions, this is the ideal way to calculate it, according to Johnson.
For an overview of eyeforpharma’s forecasting coverage, see Highlights from eyeforpharma’s Forecasting coverage.