Insights
Trials and error: new New Product Planning...
Written by Mike Rea — 2022-07-28.Most of our new product planning processes are built on the idea that each is, to an extent, predictable. However, none of us, for a moment, really considers that we have any degree of accuracy in our predictions (does anyone believe that the 'expected' NPV is expected?). It’s rare to find even confidence intervals properly incorporated into assumptions at this phase.
I suspect that we all believe that biology and clinical outcomes are the least predictable. It’s why we do the trials at all, of course - to pick up safety signals that would be impossible to capture from either chin scratching or small trials. However, as they are the basis for the other predictions, any error there is amplified through the system. And that is a huge error. (A simple glance at the failure rate from phase II onwards shows the effect of that prediction error.)
If we believed that the best AI, the best translational science minds, and the best Key External Experts could predict a molecule's effect in a disease with any validity, we would head straight to phase III to generate the requisite evidence. But we don't believe that. Our efforts to improve early phase are, at best, incremental improvements in probability, from astonishingly low to astonishingly low plus a few percent.
We believe in (and practice) learning by doing, by laying out hypotheses and testing them. So, asymmetric learning applies - if you learn faster than your competition, you may establish opportunity. As the ways to improve prediction of biology and clinical outcome in early phase improve, we will have better hypothesis generation, but not conclusions. Separating possibility from probability is essential to a healthy planning process.
So, a New Product Planning function that embraces Asymmetric Learning will generate market hypotheses, not forecasts. It is significantly more helpful to quickly and roughly indicate the attractiveness of a range of market positions than to pretend to accurately forecast the value of one. 95% of what is needed to be known about an opportunity should be at hand - a rating of attractiveness can be almost immediate given a handful of variables (addressable market, treatment days, price, market share). Treating those variables as variables is the point of New Product Planning, when it is a strategic activity, rather than an operational one. Generating the Plan To Learn includes what can only be learned in the Clinic, versus what can only be learned by market research.
It is far better for early phase strategy to be helpful while generating alternative market positions than to try to generate a number with any decimal places from a fixed point. New market positions, alternative label claims, more meaningful clinical endpoints, as well as more interesting indications. A better plan would be to plan to execute quickly in the presence of new, unexpected, information that shines a light on an opportunity (Viagra, Herceptin, Ocrevus...). A best plan would include the opportunity to generate interesting information on purpose - phase I and phase II don’t have to be MVPs (minimum viable protocols…).
If we stop pretending that the TPP is accurate, and accept that the probabilities that surround it are of low validity, a New Product Planning process has to embrace uncertainty as an opportunity, not as a problem. Unfortunately, this is the exact opposite of how it is implemented in 99% of companies, where supreme overconfidence is more likely to get programmes moved forward. Those companies that understand this first will benefit from the asymmetry.
IDEA Pharma
We work with clients early in lifecycle, crafting a compelling product story and building a best-in-class strategy that helps every molecule reach its potential. It’s what we do best. And there’s nobody that does it quite like us IDEA Pharma