A few years ago I cofounded a company called w00rk, where I was both the lead researcher and the CEO. We were building augmentation devices for offices, physical things meant to help innovation happen in the workplace, and somewhere in the prototyping phase the founding team stopped agreeing about how to define the success of what we were making. One camp wanted a first-of-its-kind graph showing a user growing more innovative over time, the sort of first-in-market result that gets a company noticed. The other camp refused to chase what it called phantoms of innovation, holding that innovation is hard to measure and that a rising line pretending otherwise would not be honest. If you have ever sat where I was sitting, a decision away from splitting your own team into two convinced camps, each holding a position that is defensible on its own terms, you know the seat.
What made it difficult was that both camps were right, and each knew the other was right. Everyone acknowledged that innovation, measured head-on, is very difficult to quantify; the disagreement was only about what to do with that shared piece of knowledge. The pressure in that seat is rarely to be clever. It is to choose without making half the room feel they have lost.
Changing What Gets Measured
My job was to settle the debate, but settling it did not mean refereeing it. I did not pick the more senior camp. Splitting the difference would have produced a compromise that satisfied no one, and waiting for a consensus to arrive on its own was not realistic. What I proposed instead was to change the thing we were measuring. Rather than try to capture innovation as it happened, we would look for the signs of anti-innovative behaviour around the device, and measure whether those were receding.
This came out of a way of seeing I have carried for a long time, close to Kurt Lewin's force field analysis (a frame I have leaned on for years, perhaps more than it can bear). Everything sits in a state of equilibrium until something disturbs it. Innovation is a driving force, and for every innovative force there is likely to be an anti-innovative one pushing back. From experience, both empirical and scientific, we already knew a good deal about what constitutes anti-innovative behaviour, the restraining side. This meant we could capture those behaviours and track their reduction, so that innovation met less resistance as it tried to take hold.
None of this required anyone to give up what they believed, and that was the whole point of it. The camp that believed innovation was real and worth signalling still had something real to signal. The camp that believed innovation was too slippery to measure directly was no longer being asked to measure it directly. If there is a piece of Fisher and Ury's Getting to Yes that I accept without argument, it is the counsel to work on interests rather than positions, and to invent an option neither side had been arguing for. The measure was that option.
Two Versions of Success
What made the change hold was that each camp could describe what it got in its own language. The camp that had wanted the first-of-its-kind result got exactly that. We built seven passive channels for capturing anti-innovative behaviour, seven different ways for a device to notice (without anyone being asked) when a room was working against itself. I believe those were the first of their kind. The findings were written up, and later presented to several large corporations in Japan.
The other camp, the one that had refused to sell a rising line, got what it had been holding out for, which was real and measurable science rather than a promise that our furniture would guarantee innovation. That promise would not have been believed anyway. Nobody likes to be told, however gently a device does it, that you are not being innovative, so the harder problem was turning these deficit signals into a positive nudge, and we managed that too.
By the time we stopped, we had built a full suite, somewhere between thirty and forty pieces of smart furniture with these measures embedded in them. Two or three of the components were later taken up in real offices. Locating your coworkers across a large office floor is something we pioneered, and so was the distinction between innovative and anti-innovative background noise. The same went for the way carbon dioxide, temperature and brightness could be tuned to the time of day and the kind of work being done.
I should be honest about how the company itself ended, because leaving it out would flatter the story. w00rk was a research startup, not a product startup, and it was funded by public money. There was no path to keep producing anything, so when the research was done the team went back to where they had come from to carry on their own work, several of us stayed on as advisers to those corporations, and some of those engagements are still running. What the two camps had actually agreed to, although it did not feel like it at the time, was not a verdict on who had been right about innovation. It was a test, the best way we had to find out.
What a Great Decision Measures
Both camps walked away able to call the outcome a win, which is the point at which I appear to contradict something I have argued before. I have written that your bad decision is someone's good decision, that every decision will be called bad by someone and that consensus-chasing is a trap. I still think so. The two ideas sit together once you notice that a great decision, in the sense I mean here, is not a claim that everyone agreed or that the choice was objectively right. It is a reading of how many of the people involved can call the decision a good one.
Put roughly, when eighty percent or more of the parties involved can call a decision good, people tend to call it a great decision. When about half can, it reads as a decision of compromise. When only a fifth can, it reads as a controversial decision. None of the three is a better way to decide than the others; they are simply different distributions of agreement. And all of that is separate from whether the decision made the business or the project better in the long run, which is a different story entirely.
Calling something a great decision is a little like calling something a great movie. It does not mean the best movie ever made, and it does not mean best for everyone; many people would happily disagree, although the disagreement does not make the description wrong.
Years ago Mary Parker Follett wrote that a disagreement can end in only three ways, domination, compromise or a third settlement she called integration, in which both sides get what they were after without sacrifice. I think we have come a long way from this way of thinking; you cannot really distinguish a disagreement in this simplistic way anymore. I agree with her direction, although there is more to it than how she described it.
Which leaves the question the taxonomy quietly begs, eighty percent of whom? A share of assent is only meaningful once you know whose assent you are counting.
Whose Verdict Counts
There are three groups who might pass judgement on a decision, and they are not judging the same thing. The first is the decision maker, who almost always (and usually without saying so) is aiming for a great decision. Very few executives set out hoping for a merely controversial one, whatever they say afterwards.
The second group is the spectators, and they are the vocal ones. When a company buys a distressed asset that quickly climbs in value, and the market watches in something like awe and says “well played, sir”, that is the spectator's verdict. It is also the cheapest one, because it rests on whatever the spectators happen to know, and a great deal of what they would need in order to judge well is never made public. The applause is real, although it is a reading of the room from the back of the room.
The third group is the one that actually counts, and it is the smallest, the parties the decision lands on. If everyone a decision affects can look at it and call it great, then something unusual has happened, and here I want to keep my own claim honest. That outcome is not just skilful but fortunate as well. You can design the third frame and do that reconciling work yourself, and still need luck for the last stretch of agreement to close. The method does not fully control its own result, and pretending otherwise would be a kind of overreach.
Which is why I want to be careful here. Everything I have described is an account of what sits behind the words when someone says that was a great decision. It is not a recipe for going out and manufacturing one. The idea is not to encourage everybody to aim for the great decision, because different circumstances call for different decision-making processes.
A decision that only one camp can love is not automatically a lesser decision; sometimes it is exactly the decision the situation needed. Reading the room describes what happened. What you should chase depends on the situation in front of you.
When I think back to that debate inside w00rk, I do not remember it as a case of persuasion. I did not win anyone over to the other side, and I did not need to. The people who believed in the graph still believe in it, as far as I know, and the ones who would not chase phantoms are still, rightly, sceptical. If everyone who came through that project would now call it a good decision, I would take that as good fortune as much as good design. What I changed was smaller and more structural than a change of mind.
I didn't ask them to change their beliefs. I just asked them to change how it is measured.
