Changing behaviors: do you use numbers or stories?

I am sure you are as convinced as I am: numbers tell the facts, and drive behaviors. We all know it by heart: “You cannot understand what you cannot measure; you cannot control and manage what you cannot understand.” This is definitely true, but is the reciprocity true as well? In other words, is it true to say that once you see the numbers, behaviors will change and performances will improve?  There are really two folds in the question: “behaviors will change” and “performances will improve”.

Starting with the first fold, is it true to say that numbers change behaviors? I hear you, the answer is obvious: “YES”. When you see the numbers, your behavior changes – consciously or not, everybody is influenced in a way or another. I agree; this is true and hardly challengeable.

So moving to the second fold, my next question is: is it true to say that when you see the numbers, your behavior will change AND RESULT IN IMPROVING the performances of the processes and people you are managing? The answer is not that straightforward this time.  

In reality, if you answer YES to the second fold of the question, it means that, if you are able to measure process performances and show your dashboards, you expect people to make the right decisions, execute the right remediation or improvement actions, so that eventually performances do improve. If you answer NO, it means that the data that you report has either no impact or is degrading the performance of the process you are measuring. With my eyes of a LSS MBB (Lean Six Sigma Master Black Belt), I would tend to jump to the conclusion: once you see the numbers, your performance naturally improves. And if you want to change behaviors, show and explain the numbers; people will naturally adapt, and, assuming nobody has bad intentions, processes will naturally improve.

It would be true if we were all automats. But the trick comes from the psychological aspect of how we – human beings – make decisions. Sadly (or luckily?), our decisions are driven by data – very little, and by psychology – a lot. This has been demonstrated by top notch psychologists and is explained by Daniel Kahneman, 2002 Nobel Prize in Economic Sciences, in his best-seller “Thinking, fast and slow”. The study proved that our actions are influenced by data and facts only once the individuals are surprised by the facts.

The experiment consisted of some individuals being asked a certain number of questions. Each individual sits alone in a booth, answering questions with headset and microphone. Everybody can hear everybody else answers, but doesn’t see the other individuals involved. One of the individuals is a stooge: he simulates having a stroke when answering one of the questions asked by the facilitator: “I am having… er… er… er… a stroke… er… er… please, er… er… er… help… er… er… er…” Which reaction would you expect from everybody else? The same as you would do: jump out of your booth and help. No! In average, 27% of the individuals go and help the stooge. 73% of the individuals don’t move. When asked why, they respond they rely on others to go and help. No need to mention that the experiment is done on a set representative of the whole population of “normal” human beings, with the statistical precautions necessary to avoid Alpha and Beta errors (ie false positives and false negatives).

Then the interesting part of the experiment starts. Three groups of people are asked to predict their own reactions if they had been in one of the booths when the stooge simulated the stroke. The first group is not being told anything about the actual results. The second group is being told the result (27% jump out to help, 73% don’t move). The third group sees the video showing 73% of people staying in their booths. The second group only knows the facts (the dashboard somehow), while the third group is being given the full story.

The results are striking: people from the second group broadly ignore the facts, and predict that, if they had been in the same situation, they would have gone and helped the stooge. Their predictions are similar to the ones of the first group who knows nothing about the actual outcome. However, people from the third group accurately predict the reality and answer in majority that, probably, they would not have moved.

What is the lesson that a change leader can learn from this? Well, we cannot assume that you, I, or the people we work with are statistically different from everybody else in general. So should we all be so convinced that showing the numbers will naturally drive changes in behaviors, and naturally result in performance improvement?… I don’t believe it anymore. I am not saying that I recommend stopping showing the numbers, certainly not. On the contrary, showing the numbers is an absolute necessity. But that’s far from being enough: that’s no more than half of the job of the change leader. The story behind the numbers should be told in its context; the numbers alone do not effectively work. Furthermore, how to publish the numbers has a huge influence on behaviors.

If you now think about what it means in your own world as a change leader, truly proud of branding the “data-driven decisions” concept: how many dashboards or metrics documents have you seen, full of numbers, without a comment and without a link to what the numbers do mean to the broader picture of the company performances, bluntly sent in a huge email that takes 2 minutes to open? And still, managers are expecting the processes performances to improve since the numbers are here ! In reality, when receiving the reports, people who take the time to open them merely flip the pages, look at the red items; at best they will be able to say: “oh, it’s better or worse than last month”. But no significant improvement action will result from a bare and bold dashboard. Big Data is everybody’s favorite topic those days: true that collecting data is unavoidable, and however time-consuming consolidating and digesting the data in a nice dashboard can be, producing the dashboard is only a “business-as-usual” task among others. Don’t expect people to know what to do just after seeing the numbers. And you and I are no different: don’t expect to know what to do just because you have seen the numbers. Go and look for the story, go and look for the explanations, go and look for the trends and lessons hidden behind the numbers. Only then you will understand the numbers.

So, to summarize, I am now saying that (1) numbers are required, so that (2) interpretation of the numbers can take place in the relevant business context. At this stage, you understand the meaning of the dashboard, behaviors start to change.

Moving on to the next step – improving the performances – still requires another level of maturity though: (3) be surprised by what you see, and surprise those who are expected to change. It’s a psychological fact: until people – including you and me – are surprised, behaviors will not be truly influenced by the numbers they see, and at best understand.

To conclude, here is how my behavior has changed, after being surprised by the reading of the above experiment. From now on, when I see a dashboard, or any other form of numbers or facts, I ask two sets of questions:

1)      Is the data transparent, ie can I read through and understand the true meaning of the bare facts?

Answering this question helps me to find out if the dashboard has the effect to move the reader from step 1 (seeing numbers) to step 2 (understanding the numbers and changing behaviors).

2)      Is the data surprising enough to change behaviors?

Answering this question helps me to find out if the reader is truly going to climb to step 3 (proactively using the numbers to improve performances)

Finding the true answer to these questions is long and difficult though, as this requires a deep reflection to put some contents and perspective into the numbers. As such, another important element to pass step 1 is that too much information kills the information: if your dashboard has too many components, answering these questions will be a lot more difficult. You simply won’t be able to find any perspective in the huge amount of information you have in front of you. On the other hand, if the data is too poor, the questions are not worth it as they won’t lead you to any surprise or breakthrough. So taking the time to report the right data with the right level of granularity is the very first step that should not be missed. The worst attitude to have when designing a dashboard would be to put “as much as possible in it, so that the information is there”. Fine the information is there, but so what? What can you do with it? How can you possibly answer the two above questions when drawn in an ocean of numbers? Streamline, point to the 3 or 5 important topics, question the meaning, trigger the surprise. Then only your numbers will contribute to make a difference.

Eventually, I am still a believer of the “there-is-no-one-single-answer-to-all-problems”. So keeping asking the right questions at the right time is what should drive us, change and innovation leaders, in all what we do, so that we can truly influence behaviors and help people to improve their processes and performances. At the time of obtaining the data after a most likely long and painful data collection process, I found that the two questions above play the trick. This should eventually result in surprising the people reading the dashboards, which is the only way to make people change their habits.

Going further, which question to ask at which stage of a process improvement journey is the core topic of my book: Lean Six Sigma: Coach me if you can – Take a look? You won’t find answers though, just the questions you need to ask so that you can bring your own answers to your own problems, in your own world. Bringing your own answers on your own is what will make the difference between average and good. Enjoy the reading, and good luck with your own process improvement journey!