(This post originally appeared on Innovation Excellence)

I’d like to touch upon a subject that doesn’t come up often in innovation discussions: I’d like to talk about how we kill projects.

Everyone would agree that killing projects is a key to maintaining healthy product development pipeline. By terminating projects that are going nowhere, companies free up resources to introduce new, potentially more successful initiatives. When I hear someone saying “We’ve got a lot of great ideas, but have no resources,” I’m always tempted to ask: “When was the last time that you killed a stalled project?”

But killing projects is tough. However good we might be at celebrating failure (remember this “fail fast, fail often” hoopla?), we are actually quite bad at admitting failure. Failed projects negatively affect the company’s reputation and bottom line and they definitely don’t make our career ladder easier to climb (to say the very least). Besides, almost every project is someone’s brainchild, and who likes killing their brainchildren? The great Leo Tolstoy was reportedly crying when writing the scene in which Anna Karenina threw herself under the train.

The way in which companies make their kill/continue decisions largely depends on which of the two schools of thought they belong to: the “pick the winners” or the “kill the losers.” The “pick the winners” approach relies on selecting relatively few projects that supposedly have the greatest chance to succeed–and then investing heavily in these projects. Once the “winning” project has been selected and advanced into the project pipeline, killing it becomes progressively difficult. In contrast, the “kill the losers” strategy is based on launching a larger number of projects, carefully monitoring them, identifying those that aren’t going to succeed and then killing them as rapidly as possible.

At first glance, the “pick the winners” strategy makes more sense. Indeed, if we have established a solid theoretical framework, developed advanced proprietary technology and secured some proof-of-concept data, is it that difficult to identify an eventual winner? Isn’t it where our prior experience in product development really pays off? Besides, isn’t it a proper way to raise (brain)children: to have just a few of them and then invest heavily in each one?

Yet, a recent study on efficiency of drug development, arguably one of the most expensive kinds of projects in the history of humankind, shows that in real life it’s the “kill the losers” strategy that turns to be a true winner. A group of authors at The Boston Consulting Group analyzed 824 individual drug candidates with a known full development outcome coming out of 419 companies. Of these 842 molecules, 637 failed in Phase II clinical trial or later and 205 were approved. For each candidate, 18 attributes were assessed for correlation with success or failure. What the authors have found was that the strongest single factor correlating with success was a high termination rate in preclinical/Phase I stages. In other words, companies making hard decisions about which project to terminate earlier in the project lifecycle do better than companies postponing these decisions for later (and often doing this not of their own volition).

Shall we call it The Terminator Effect?

There are encouraging signs that drug developers warm up to the “kill the losers” strategy: AstraZeneca, the world’s seventh-largest pharmaceutical company, has recently announced termination of 15 therapeutic programs and said that from now on, it’ll be reviewing its pipeline quarterly rather than every 6 months.

“Anna Karenina” would have been completely different book if Tolstoy decided to “kill” Anna at another point in the epic story. But various literature genres demand different rules. Take, for example, crime thrillers where the first dead body is expected to show up on the very first pages of the book. And I would argue that drug development and crime thrillers have something in common: you don’t know the outcome until the very end.

image credit: orionpictures.com