How to Kill Bad Projects

When entrepreneur and former Accenture IT researcher Adam Siegel worked as a consultant, he would see how information often went from true to much less true—from fact to rosy fantasy—as it moved up the hierarchies of client companies. “What were mostly red lights at the project level became mostly green lights—with maybe a few yellows thrown in—by the time [a report] got to the CEO.”

Project owners creatively spun results for political reasons—mainly to prevent funding from being yanked. Consequently, there was a gaping disconnect between the project people down at ground level and the business leaders farther up the food chain when it came to understanding how projects were actually progressing. The leaders tended to think things were going much better than they actually were.

The problem of corrupted information flows stayed with Siegel and ultimately led him to found his current company, Inkling Markets, a software-as-service venture aimed at helping companies conduct successful prediction markets. What does a prediction market have to do with eliminating spin? Siegel sees an opportunity to produce higher quality decision support in businesses by tapping anonymous input “from people who aren’t normally asked their opinions, in samples large enough to filter out individual agendas.”

In the case of an internal prediction market, employees might be asked to weigh in anonymously (wagering a sum of token currency) on a statement like this: "The Voldemort Project will meet all of its defined performance targets by the end of 2008.”

Based on the level of agreement with the statement, an aggregated assessment of the Voldemort Project’s prospects emerges. If that assessment were to disagree significantly with highly optimistic official reports, then the project owners would have some explaining to do. Markets can aim either for wide input on questions of enterprise strategy, or for narrower input on specialized matters from populations of more knowledgeable employees (say, everyone working in the R&D function).

Siegel’s software has a do-it-yourself component so that participants can suggest their own questions. Inevitably, “this surfaces good questions that market owners might not think of.” He adds that this is “hugely important if you’re trying to find the next black swan.”

While many are naturally captivated by the black-swan-finding potential of prediction markets, another sweet spot may be their use as a form of institutional lie detection—guaranteeing the integrity of internal reporting and keeping the progress of business initiatives transparent.

Have any of you used prediction markets for this purpose? If not, what other methods have worked in combating grade inflation awarded to undeserving projects?

You might also be interested in this free HBR in Brief about other methods of institutional lie detection: “Delusions of Success: How Optimism Undermines Executives’ Decisions” .

- harvardbusiness.org