Startup Teams: There is no such thing as a bad team, just a bad fit! – Saberr explains why.

“It’s all about the team.” It’s actually becoming rather annoying to keep hearing this. At one point or another in your early entrepreneurial ventures it is almost guaranteed that someone will offer this pearl of wisdom …Like you didn’t already know that a good team was fundamentally important to your success?! It’s like heading to the airport and the taxi driver saying “make sure your plane has wings”. The unfortunate thing about clichés is that they are usually based upon solid truth. The hard part about this particular cliché is that although it’s undeniably true, it’s also very hard to measure, prove or even improve your team.
I think the only thing more annoying than hearing “the team is important” is the inevitable anecdotal advice that follows as to what makes a good team. “Are they someone you could grab a beer with?”, “You need a good mix between developers and designers, forget the business people”, “have you gone through a tough time together?” and my personal favourite “make sure they fit your culture”.
I’m an engineer. I don’t like it when people give advice that can’t be measured. Take my personal favourite as an example. How does one measure culture? How do I measure if one person has better culture than another, how do I even know what good culture looks like?!
Two years ago, at a Startup Weekend in Seattle, driven by a statistic from Noam Wasserman of the Harvard Business School that 65% of startup failures are due to issues between the founding team, I set out to attempt to quantify the ‘team’ and with it make a prediction as to which team would win Startup Weekend. The following is a brief account of the process.
My first question was “how do you measure team dynamics?” So I turned the vast library of academic theory on team dynamics. Tools like Myers Briggs, Belbin or the Big 5 are all psychometric models that give insight into the behaviour or personality of people and how they might fit together. Unfortunately none offered the ability to predict future real world outcomes nor the likelihood of a team winning Startup Weekend. There was also disagreement between various academic factions as to the validity of the differing methods.
So I simplified the question to, “Is it possible to predict a good relationship?”. Intuitively this is much easier to answer. The answer is yes. Human’s are a social species – we’ve evolved to be good at predicting relationships. But I wanted data. The world of online dating represented a vastly rich dataset which is essentially a digital record of two people’s search to find a good relationship. So I looked for patterns in the way successful matches behaved online. I defined a successful match as two people who had met online and then closed their accounts. Now you could argue that this isn’t representative of a ‘good relationship’ but in the world of online dating it’s a significant step and a good place to start. There were, unsurprisingly, very consistent patterns in the way successful matches answered questions or expressed themselves online. So I took this insight and combined it with the consistencies in behaviour theory from the academic world.
My next step was to test it. Unfortunately the process above took somewhat longer than a weekend so having missed the end of Startup Weekend in Seattle I turned to the University of Bristol in the UK. They were running a week long business plan competition with 8 teams of roughly 8 people in each team. I wanted to use my algorithm to predict which team would win the competition without any knowledge of their skills, their experience, their demographic and crucially no knowledge of the business ideas they were working on. Submitting my predictions ahead of time I turned out to be correct in predicting the winner. What was more interesting however was that I was correct in predicting the precise ranking of all teams 1 through 8. I’ve repeated that first test 21 times and the model has >95% accuracy in predicting the ranking of teams at short term entrepreneurial competitions, anywhere from 1 day events through to the annual 8 month Microsoft Imagine Cup. As it happens, the model is also very good at predicting the performance of individual employees – again with no knowledge of skills, experience, role, seniority or demographic. Simply measuring their ‘fit’ with the rest of the team.
Now I’m not saying this is a silver bullet and of course there are many more questions to be answered. Does the model hold over longer periods of time? Am I in fact measuring the ability of a team to deliver a killer presentation? etc etc but it does point to the idea that the model is at least asking the right questions.
The model measures two key aspects. The first being the diversity of the individuals contextual behavioural preferences. That is to say, how does the person prefer to behave when surrounded by his or her team mates? And how much do these behaviour’s differ throughout the team? In this case more diversity is better. This can be measured through various online tests – the ‘Big 5’ is a good place to start.
The second is the degree to which team members are aligned on their core values. What I mean by this is not “what are your values?” because that’s a difficult question to answer. In fact if you were to ask me what my values are I would probably tell you every value ever. Do I value trust? Yes of course I do. Do I value freedom? Yes of course I do. Do I value trust more than freedom? I’m not sure, maybe, maybe not. As people we’re bad at explicitly defining what our values are. So the key is value alignment. Now the model measured this through pattern recognition in a carefully selected set of weird questions similar to what you might find on a dating site. Essentially though the model is looking for agreement on controversial and politically sensitive subjects. Things like sex and religion… but mostly sex. And even more so that there is agreement on how strongly you feel about these subjects. t’s key to note that the model looks for disagreement on non-sensitive subjects like sports interests or hobbies. Remember diversity of thought is good, diversity of core values however is bad.
So there you have it. To win at Startup Weekend (and perhaps longer term in the business world) find a way to measure behavioural diversity and value alignment.
Alistair is a co-founder of Saberr, a people analytics company helping large enterprise make data driven decisions about their workforce.