Why One Company Uses Online Dating Data to Predict Team Performance — Accurately
As a digital marketing agency with a focus on HR tech, we at Red Branch Media go bananas for companies that share our ambitions and ideals around workplace inclusion, productivity, and engagement. Recently, we came across quite the unique team at Saberr, a people analytics company based in the UK. Founder Alistair Shepherd built the company with one idea in mind: Together, people are powerful.
Naturally, I had to know everything about Saberr, so I sat down with Shepherd to learn more about his beginnings, his core values, and how his team got to where it is today.
Maren Hogan: What are your biggest concerns with traditional workplace communication, productivity, and leadership in team environments?
Alistair Shepherd: Our economic, environmental, social, and political challenges could be solved quicker and more easily if groups of people collaborated without friction. We’d have the opportunity to make faster progress and make important discoveries, and the world could become a better place to live in — but this requires teamwork.
Teamwork can happen without technology; it is nothing new. But we don’t always work together as well as we could. The very best teams have dedicated coaches, people who are responsible for getting the best performance out of the team and ensuring sustained productivity. All collectives need support to succeed, whether we’re talking sports teams on the training ground or executive teams in the office.
Our concern is that most teams don’t have access to this support, and therefore, they don’t achieve as much as they could. We think technology can play the role of the coach and give every team the support it needs to be massively more successful.
MH: I understand you started building Saberr by looking at dating site data. Tell us more about that.
AS: We started with the assumption that the relationships we have with our colleagues underpin a lot of team performance. Anecdotally, this makes sense: Bad relationships are incredibly emotionally draining, whereas good relationships give us more strength.
Our goal was to see if a.) we could predict relationship quality between people who had never met and b.) use that prediction to forecast the performance of new teams that had just formed. The first generation of online dating profiles — think Match.com, eHarmony, etc. — were a rich source of data on the characteristics of users and their preferences in others. We used this data, combined with a lot of academic study on relationships and teams, to come up with our first algorithm to predict relationships and thereby team performance.
Our first test of the algorithm was at the University of Bristol, which was hosting a week-long business plan competition where student teams would compete to win fairly substantial prizes. We profiled the teams using our software at the beginning of the week, and we were stunned when we predicted the correct ranking of all eight teams come the end of the week. We did this without any knowledge of the teams’ skills, experience, demographics, or ideas. All we looked at was our predictions of the relationships that would form within each team.
The ability to predict team performance by analyzing team composition obviously has great application in hiring and staffing decisions, but the real question for us was: How can we help develop teams to perform exceptionally well regardless of how well suited the members are to each other? CoachBot was born to solve that problem, and that’s where Saberr’s really clever stuff happens.
MH: You say robots have more to learn about humans, not the other way around. Explain how this idea can be applied to the workforce.
AS: The rhetoric over the last decade has been, “Hey humans, you need to understand technology.” I think that needs to change to, “Hey technology, you need to understand humans.” Humans have emotional needs, and the “right answer” is not always the best answer. Technology needs to be sensitive to these needs.
A great example is the film Interstellar, directed by Christopher Nolan. In the story, a handful of people set off on a dangerous voyage in a spaceship. The captain asks the ship’s computer what its “honesty parameters” are. The ship replies, “Ninety percent. Absolute honesty isn’t always the most diplomatic nor the safest form of communication with emotional beings.” I find this comment funny because of the truth it contains: People need support, not clinical judgement — even if we say we prefer the latter.
I think the second thing technology needs to understand about humans is that we need each other much more than we need technology. Any AI designed to help improve team performance should aim to increase human interaction rather than decrease it. At the moment, it seems that almost every new innovation is automating away the need to interact with other people. If we continue in this fashion, tech will have a very detrimental impact on society.
MH: What differences in performance outcomes do you see with bots compared to people?
AS: If we take team coaching as an example, the biggest advantage that a bot has over a human is its ability to have cost-effective interactions. A bot can check in with you during your workday without it being a distraction. It can gather data little and often, and it can deliver coaching in the same manner — little and often, without interrupting “real work.” A human coach can’t do this, at least not in a way that is commercially viable.
Being able to hijack what people consider “real work” and ask simple questions like, “Is the work you’re doing today going to help you achieve your goals?” helps nudge people toward behavior change without it seeming like much effort. This is a real bonus.
MH: What is ideal workplace performance to you?
AS: In a collaborative team environment — which is the kind of environment most of us work in today — ideal workplace performance is when the team:
- is motivated by a common purpose,
- has a shared understanding of clear and measurable goals,
- has well-defined roles and responsibilities, and
- has strong psychological safety and open channels of communication.
If the above factors are coupled with the right technical skills, then it’s highly likely the team will deliver a favorable outcome. The important part for me is to focus on the behaviors that lead to good outcomes rather than the outcomes themselves. Behaviors can be repeated or changed, but outcomes rarely can.
A version of this article originally appeared on LinkedIn.