Moneyball for Business
Today it’s possible for an algorithm to predict if you’re going to ‘fit’ into a business. Answer a series of questions on your tablet and that could be how you’re hired (or told ‘no thanks’) by the employers of the future. We talked to Alistair Shepherd, one of the founders of Saberr, who describe themselves as ‘Moneyball for Business’. They provide tools that allow employers to optimise the efficiency and effectiveness of their workforce through the ‘intelligent’ design of teams and the recruitment of candidates who are statistically more likely to be high performing. This is based on two main beliefs. One that a spread and diversity of personal characteristics are good for teams, and two, that a team that has a good alignment of ‘values’ has a good chance of succeeding.
Tell us about yourself, what was it like growing up?
I’m 25, from Fort William in the Highlands of Scotland – my parents moved there when I was six. I spent all of my youth just mucking around outside – it’s very rural – had a brilliant childhood, with sports, the rest of it. I did terribly at high school; the drop-out rate was really quite high. That was the one thing: quit school as soon as you can, get a job, or learn a trade. I was really into sailing though, and I sailed for Scotland for a while. I just wasn’t engaged at school; very distracted and interested in everything else apart from what I should be focused on. My mum always said my concentration was terrible.
So, you weren’t academic then?
Not really, but I wanted to go to university. It wasn’t a common thing from my high school, which almost made it worth wanting to achieve. My parents had been to university, and I guess I thought it was a reasonable thing to do. I went to Southampton University (UK) and did aerospace engineering there, I didn’t study particularly hard, but I did the four years. More interested in other things…again. I graduated though!
After University did you work for other people? With the lack of interest in school, it sounds like you were either destined to drop out or be an entrepreneur.
I’m accidentally entrepreneurial. I never intended the plan to be the starting of a company. In my final year at university, we did a project, there were four of us. We came up with a novel wave energy device, it sorts of got a lot of attention both in and out of university. It was the first time that for anything I had done, anyone outside of my immediate environment had paid attention to. It was definitely the first thing I had done that perhaps had a commercial value.
It was really interesting, and I became fascinated by the business side of things, the economics of it all. People were looking to buy and use the product straight the way, I found that really compelling.
This led to me being interested in business, and I got offered a scholarship on the Kauffman Global Scholars Programme in America. All of a sudden, I’m being sent to Harvard, MIT and Stanford. A long way from home, an unbelievable course, with 20 passionate, driven and motivated people around me. That was it. It kick-started me into wanting to do something commercial.
Did you come across something whilst in America that sparked the idea for Saberr?
There was a series of moments. The first at Harvard. Noam Wasserman who was a professor there who had written a book called ‘The Founders Dilemma’ and in class once he said that 83 per cent of startups fail, and I had heard this number before. I found that quite alarming.
However, the interesting thing was that 65 per cent of these failures were because of team dynamics, issues between the founders or early employees. One of the only things you have control of when you start a business is who you work with, and for that to be the dominant cause of failure just didn’t make sense from an engineering perspective. Why aren’t we paying attention to this?
Mathematically the problem is hugely complex, but I’m obsessed with simplicity and how things can be modelled. That’s the beauty of mathematics. I was familiar with the team dynamic theory from my engineering studies. They covered models like Myers Biggs, the Big Five, personality profiling tools. They are all fascinating, but none are predictive. They all tell you stuff about yourself, but there’s lots of debate about which one stands up to academic rigour and lots of disagreement.
So how do we simplify the situation? Can we predict a good relationship? The answer is yes, we do it all the time in daily life, it can happen in 3 seconds, bang. We don’t have a checklist, we just do it, and we’re pretty good at it.
So, can we find data to model this? To me, online dating was a hugely rich dataset, which is essentially a digital record of two people’s search to find a good relationship. We looked at patterns in successful matches when people close their account after finding someone online. I found consistent patterns in the way that they answered particular questions. I took these insights from online dating and combined them with the consistencies from academic behaviour modelling. An example being: diversity of personality, regardless of how you measure it, is beneficial – so the more you have in terms of diversity, the better it is for team dynamics.
The next thing was to test out the predictive model, to see if we could predict what would make up a good team. The first test was at the University of Bristol, a business plan competition with eight teams, and roughly eight people in each team. I wanted to see if I could predict which team would win without any knowledge of their skills, their experience, their demographic, and crucially, no knowledge of the idea they were working on. In fact, I never met the people, I just wanted to measure them with my questions.
We were correct in predicting who would win, but what was more interesting was that we got the ranking of all eight teams correct. That was the point at which I thought we had something that hadn’t been done before.
Was this just you?
No, it was me and my friend Sam. The model was mine, but we worked on the business together. There’s no way I could have done it without Sam. The classical career path just didn’t appeal – you’re working to make someone else’s dream come true. Somebody in a successful company, once upon a time, had that idea, and you’re just fulfilling their ambition. I’m not saying that’s a bad choice, but it just didn’t appeal to me.
What steps did you take to make the idea a reality?
Lots of steps, sometimes backwards (laughs). The first thing was to test the ideas, to see if they stood up. We did the test 20 or so times, and got in front of as many people as possible – we did the Microsoft cup and some other events. I was fairly cavalier with my statements.
So, the idea is that you use questions to try and find the best team dynamic?
It’s based on two things. We measure behavioural preference and value alignment. It’s always contextual. Classic personality tests always say that you are ‘X’, but that’s not true, we change, and my behaviour in different situations changes. It’s all relative. Two extroverts together will mean that one of them is more extrovert than the other. We measure my preference to exhibit certain behavioural traits in the context of the people I’m with.
The other thing is value alignment. We look to not explicitly ask people what their values are because it’s very difficult. When you meet other people, you don’t ask what their values are, you’re just interested in knowing that you share them. We look for value alignment without explicitly stating what those values are.
What kind of things do companies wanting your services need to tell you then?
If you were recruiting, we would measure your existing team – all of you individually – so this would give you insight into your existing team; whether you’re a high functioning team or where the tensions lie. We then measure the candidate, and then we get an idea of how they will match with you.
Does ‘fit’ necessarily equal performance?
Well, we wanted to see if individual ‘fit’ made a difference to individual performance. We found a company that had 20 or so people in one of their development teams, and one employee they were having doubts about. They challenged us to find that person using our algorithm. They asked us to rank the people from the best ‘fit’ to worst ‘fit’.
Well, we did the test and we found that person. However, what was more interesting was that our ranking of fit almost precisely matched their internal ranking of key performance indicators. This obviously becomes a more compelling business case and model.
What’s the response been like? Are people interested in your product?
The interest level has been enormous. We are making a process which is typically very difficult to quantify into something which is efficient, and effective. We are starting to deploy data into human resources or any scenario where people are together: sports teams, military, government and commercial situations. It’s fascinating.
We’re also tapping into latent human curiosity, the “tell me about me” that interests us all. That’s surely why horoscopes are so popular?
I’m not so sure if I would want to do one personally.
That comes down to the method, the communication of the system. It’s about trust, but we do have that curiosity to know about ourselves. We have a relentless urge to know, to learn. Commercially it’s working, so we are very hopeful.
You’re based on Google campus, what’s the reality of working here, in Shoreditch and Tech city?
It’s less glamorous than it may seem from the outside. It’s really hard work, long hours, with no reward at the start. The hipster thing is not true either (laughs). None of us fit that stereotype. Fundamentally if you have a great idea, it could be crazy, but you may just make it.
By Mostafa Sayyadi Knowledge can be classified into individual and collective knowledge. Executives recruit followers based on their individual knowledge which refers to the individual’s skills, prior-knowledge, and proficiencies or sometimes referred to competencies....
Here at Global Innovation Magazine, we like nothing better than talking to people who are making a difference in people's lives. Not just because of the technology they develop, which of course can have a great impact. The great stories for us are the ones about...
Can innovation help the petroleum industry to survive the perfect storm created by the oil price war and COVID-19 pandemic?
I believe innovation not just can help here but will become the key part of the solution. However, to benefit from innovation and achieve sustainable growth and value creation, companies have to adopt a systemic approach to innovation. In particular, companies should...