Fast-Fail

Failures are not something to be avoided. You want to have them happen as quickly as you can so you can make progress rapidly. 

 - Gordon Moore, Intel Corporation

Futility Theory Meets The Law Of Unintended Consequences

Popularized by the sociologist Robert K. Merton, Unintended Consequences are defined as outcomes that are not the outcomes intended by a purposeful action. They can be roughly grouped into three types: positive, negative and perverse.

Positive consequences yield additional benefits to an action. Negative consequences are unforseen detrimental effects. And perverse consequences are those which are contrary to the intended action or policy.

The Streisand Effect is a good example of perverse consequences. 

Citing privacy violations, the actress and singer Barbara Streisand attempted to have an aerial photograph of her Californian mansion withdrawn from a publicly available collection of photographs commissioned as part of the California Coastal Records Project. Her unsuccessful law suit generated signficant public interest driving 420,000 visitors to the photograph in the first four weeks.

The Law of Unintended Consequences states:

When a simple system tries to regulate a complex system there will be unintended consequences.

So what has this got to do with pharmaceuticals?  And what has this got to do with Futility Theory?

In pharmaceuticals, the management system is simple, it operates with limited information (rational ignorance), short time horizons, poor feedback, and poor and misaligned incentives. Pharmaceutical R&D though is a complex, evolving, high-feedback, incentive-driven system. And when a simple system tries to regulate a complex system you often get unintended consequences.

In the 1990s the pharmaceutical industry embraced business process re-engineering. In focusing on minimizing time to market, the industry hoped to maximize returns on R&D investment. Reducing the development time of successful molecules has been achieved by pushing more and more development activities into parallel. Sadly for the industry the majority of molecules are unsuccesful. We have optimized a process around the tiny minority of molecules that make it to market.

Strategies directed at increasing development speed and reducing the development time for successful compounds can actually increase the expected time to market. 

   - The Development Speed Paradox

By addressing development speed without tackling the high attrition rates in pharmaceutical development, business process re-engineering had the opposite effect on R&D productivity.  By treating a stochastic problem without regard for the probabilities of failure we have sub-optimized the development process.

Maximising development speed has clogged the development pipeline with marginal or failing medicines (see More Haste, Less Speed).

Twenty years later the industry is wrestling with the consequences.

Copyright Dennis Lendrem 2011

Fast-Fail - From Pharmaceuticals to Financial Products, From Software to Fine Teas

The formula for success? It's quite simple, really. Double your rate of failure.

Thomas J. Watson

Fast Fail strategies work by rapidly clearing the development pipeline of marginal products releasing development resources to focus on more promising products (Lendrem, 1995).

The Quick-Kill^TM model demonstrates that Fast Fail strategies will:

• shorten the expected time to market, even as the cycle time for successful projects increases (the Development Speed Paradox) 
• increase throughput, even as development speed decreases (the M25 Effect) 
• reduce development costs, even as the burn rate in early development increases (see R&D Savings), and  
• do all this, even as the rate of false negatives - killing projects that would have been successful - increases (see Quick-Kill Rides Again)   

Fast Fail thinking was originally developed in the context of high risk enterprises such as pharmaceutical development where the probability of each molecule screened making it to the market place is low.  However, the thinking is generalizable to a range of products and services. 

The probability of success in the Quick-Kill^TM model can be re-defined, say, as the probability of recovering the development costs; or the probability of securing a given return on development investment; or the probability of securing a given return within a given timeframe. 

80% of all new products or services fail within six months, or fall significantly short of forecasted profits. 

 - Gerald Zaltman 

Sadly, since the majority of new products or services fail to deliver then the model can be extended easily - from software development to fine teas, from financial products to confectionery.

The Quick-Kill model was first developed in the context of pharmaceuticals.  Only recently have the wider implications for business been appreciated. 

References

Lendrem, Dennis. 1995 A Clear Case of More Haste Less Development Speed, Scrip Magazine (December 1 1995): 22-23.

Zaltman, Gerald. 2003 How Customers Think: Essential Insights into the Mind of the Markets. Boston: Harvard Business School Press. 

© Dennis Lendrem, 2011