Sample frameDay 304·Innovation · Variance

Your best bet for next year is already somewhere in this year's pile. Nobody can pick it out, and that isn't anyone's fault.

Innovation.

The Innovation Tournament

5 min read·Apply by lunch

The question

Count the genuinely separate attempts you have running at the problem you most need solved — separate in approach, not merely in name, so that one of them failing would tell you nothing about the others — and if the honest count comes to one, what exactly is meant to happen on the day that one turns out to be wrong? Most plans have no answer, because the plan was that it wouldn't be.

The idea

Christian Terwiesch and Karl Ulrich both teach at the Wharton School at the University of Pennsylvania, and in 2009 they published Innovation Tournaments, which treats innovation the way a knockout competition treats players. A large field enters, rounds of filtering remove most of it, and a few exceptional survivors come out the other end. The arithmetic underneath is extreme value theory: if ideas are draws from some distribution of quality, the expected quality of the best one you're holding depends on how many draws you took, the average of the distribution, and its variance. Notice what that last term does. Variance is the enemy everywhere else in operations, and here it's an asset — a field of near-identical candidates doesn't have much of a right tail, however good its average. A fourth factor sits alongside the three, because a field you can't sort is a field you can't collect on. With Karan Girotra of INSEAD, the pair ran an experiment reported in Management Science in 2010: groups whose members generated ideas alone before meeting produced about three times as many as groups working together throughout, and those ideas rated higher on average. The same paper's less comfortable finding is that both structures were poor, in absolute terms, at picking out their own best work.

The framework

01
Two attempts that share a premise are one attempt.
Two projects resting on the same belief are one project with two teams on it. If the reason the first one fails is also the reason the second one fails, you funded a single draw and paid twice for it. So walk your live attempts in pairs and ask whether one collapsing would leave the other standing — different mechanism, different customer, different assumption about why the thing works at all. The count that survives that test is your real number of starts, and it's usually a good deal smaller than the number on the slide.
02
Variance is the enemy everywhere else, and the asset here.
Everywhere else in the business you're paid to narrow spread: a process that varies is a process that fails somebody, and consistency is most of the job. Here the arithmetic runs the other way. What a field is worth sits in its best member rather than its average, so a wide ragged spread beats a tight cluster around a better mean, because only the wide one has a long right tail. Which is why the sensible-sounding moves are the expensive ones — the single shared brief, the early convergence, the odd proposal quietly dropped before anyone had to defend it.
03
Alone first, together after.
The three-times result is a change in seating rather than in budget, which makes it the cheapest of the four levers you own. Give everyone the problem and a week at their own desk before anybody convenes, ask for ideas in writing rather than a conversation, and let the meeting be for combining what came back. What a room that starts together does is start from one stimulus: the first idea spoken becomes the frame, and much of what follows is a variation on it. Separate desks raise the count and the average together. One room raises neither.
04
Your filter is worse than you think, and that changes what you do with it.
The uncomfortable half of the Girotra, Terwiesch and Ulrich result is that both structures were weak, in absolute terms, at assessing the quality of the ideas they had generated. Take that seriously and two things follow. Carry more candidates than feels tidy, because a noisy ranking makes dropping the eventual winner a live risk rather than a theoretical one. And spend on evidence that sharpens the ranking — a price actually quoted, a customer actually approached — rather than on more discussion. How the survivors are then gated, and what written condition ends each one, is the next two mornings.

Members only · 4 principles + template + AI mentor

364 more frameworks are waiting.

The Innovation Tournament is Day 1 of 365. One framework every morning for a year — across strategy, sales, negotiation, leadership, and more.

$1 /day

Billed $365/yr · cancel any time

365 frameworks

12 topics · templates · AI mentor

Reserve your place — $365/yr →

One payment a year · the whole library · no auto-renewal.

More Innovation frameworks

All 24 innovation frameworks →