At Culture Amp we call ourselves People Geeks, not Data Geeks. It might not immediately be clear why – after all, we are technologists and we deal in data and statistics. But People Geeks we are.
Culture Amp’s perspective is firmly founded in industrial and organizational psychology. What we do with our data and statistics is make decisions around people. This is different to crunching numbers to figure out how to make a new phone, optimize a manufacturing process or design a large building. A lot of those problems are reducible – they can be pulled apart and put back together. People, not so much.
When dealing with human beings, you’re dealing with culture. The data you collect has to be useful in that frame.
When dealing with people, averages can be precisely wrong
People are comfortable with numbers, with scores, and especially with averages. How many times have you heard in the boardroom: "Just give me a number"? But these numbers can be either meaningless or deeply flawed.
Say you ask all the people in a hypothetical division to rate their job satisfaction on a scale of 1-10. The result comes back with an average of 7.5. From that it would be fair to deduce that overall, people are 75% satisfied, which is pretty good. On that basis you give your two line managers a moderate raise.
But would it make a difference if you discovered half those respondents gave a score of 10 and half a score of 5? That would still be an average of 75%. But you would approach your employee satisfaction problem in a very different way. Why are half perfectly happy and the other half deeply dissatisfied?
Your probing might find that those who voted 10 worked for one manager who is motivational and hard working, and those who gave a low score worked for the other, a lazy bully.
We call the problems that arise with using an average to drive decisions "the flaw of averages." Proceeding on assumptions made from an average can be very dangerous – in this example, you’ve given a raise to a toxic manager, and made a high performing manager resentful of both the survey and performance management system.
The average is precise, but it’s precisely wrong about what’s really going on in our hypothetical division.
Why the Top 2 Box is a better approach
When dealing with people, a much better approach is to use the Top 2 Box. This can be distilled down to: "Rather than telling me the average score, tell me the percentage of people that agreed or strongly agreed."
The Top 2 Box uses a Likert scale, typically with five values, e.g. – Strongly Agree / Agree / Neutral / Disagree / Strongly Disagree. Once collated, we combine the responses of the top two responses (strongly agree / agree). In the above instance, if the statement was: "My job satisfaction working at [X] is high", the Top 2 Box score would be 50%. That is, half are satisfied, half are not.
Knowing this is much more useful for action because the questions describe a situation that we want to occur and the answers tell us how often it is occurring. The number is no harder to interpret – but it’s a much more meaningful representation of what’s actually happening with your people.
“Don’t create a better X… create a better user of X.”
At Culture Amp, the reason we look at people problems differently to data problems can be boiled down to one overarching idea drawn from iconic developer Kathy Sierra: “don’t create a better X, create a better user of X.”
For us, it’s not about just creating a survey analytics platform. It’s creating a better People Geek. How do we make our software increase and improve your capability rather than just giving you the answer?
In terms of delivering that experience, we spend an enormous amount of time on how we present the data so that it helps people act in a way that is likely to get a good outcome. The thing is, we don’t need more data. We’re all drowning in data. We need better data. So our focus is taking this big sea of data and giving it back to you in a form that a) you can understand, and b) can guide you to action.
People who are good at HR usually exhibit a strong understanding of how people think and feel. They’re good at reading a room, at sensing what’s going on. They have an intuitive nose for what needs to be done. Good data can take the rest of the organization on that journey with them. Seeing it in black and white, empirically measured, will help others in the organization to also see that truth for what it is.
People problems are not the same as data problems. We can’t solve people problems by throwing numbers at them. But people problems can be identified using the right data in the right way, so that you can work on real human solutions.