Where do the really great inventions come from? Many managers understand little about the invention process. They don’t possess insight into the most likely sources of technological and scientific breakthroughs.
What they need to do, argues Lee Fleming, an associate professor at Harvard Business School, is understand that invention is essentially a process of recombinant search. Having studied more than 17,000 patents, he says it’s essential to understand innovation in a very basic light: that is, as a new combination of components, ideas or processes. Every invention can be thought of as an assembling of these parts. For instance, think of a steamship. It’s comprised of a sailing ship and steam engine. Or Apple Inc.’s iPod. It combines cheap memory, digital music and a lightweight battery.
The problem is, people focus only on the successes — the inventions that successfully combine parts to create a startling new whole. But, Fleming says, in order to have more control over the innovation process, companies must understand that inventing is an iterative process, one that has a distribution of outcomes. And, in fact, this distribution is very skewed. According to Fleming’s research, almost all inventions are useless. A few have moderate value and only a tiny, tiny few are breakthroughs.
This profile holds true for every well-sampled distribution of inventive data, whether the measure is patent citations, scientific citations or financial returns. It even holds true in other creative fields: for instance, the number of times a piece of music is performed or the number of times a book is reprinted.
This skewed distribution is rarely acknowledged, however, let alone investigated or systematically managed. This probably occurs for several reasons, including the difficulty of collecting complete data, since failures are often forgotten — people discard rather than learn from them.
But if managers want to understand how these breakthroughs in creativity arise, they need to understand the process that generates the entire distribution — the failures and moderate achievements along with the stunning successes.
In particular, managers need to keep in mind the following three things: First, inventing is a process of putting together a lot of untried combinations. Each combination is an invention. Each provides a shot at a breakthrough. Therefore, to invent a breakthrough, researchers need to try many things. Single attempts rarely work. Think of this in terms of a sports metaphor: Before you get a goal, you have many shots on the goal.
Second, once you generate a sufficient number of shots on the goal, you need to maximize the average of those shots. Here the sports metaphor is a little misleading since, unlike in basketball or soccer, for instance, innovation is an endeavor in which the value of the shot scores can vary greatly. Some could be incredible successes, others only minor achievements. You want to — for the shots that do make it — have inventions that have high average value. If their average shots are low, companies rarely invent breakthroughs.
Third, if you really want to create a breakthrough innovation, you need to do more than increase the number and average score of your shots. Your company needs also to expand substantially the scores’ variability. You need to take wild shots at rich targets: The wider the range, the more likely you are to score a really valuable shot.
Consequently, companies need first to identify how they want to improve their innovation process. If an organization has an adequate number of shots on goal, for example, but a poor average score, it might consider substantially investing in science and basic research. If, on the other hand, an organization doesn’t have enough shots on goal, it could focus on cultivating technology brokers and increasing intellectual diversity in its labs.
This article is adapted from “Breakthroughs and the ‘Long Tail’ of Innovation” by Lee Fleming, which appeared in the Fall 2007 issue of MIT Sloan Management Review. The complete article is available by clicking here.