Notes from Industry, Playing To Win

Strategists: Stop Obsessing about Averages

The Only Way to Create the Future is to Pay Attention to the Outliers

Roger Martin
6 min readMay 3, 2021


Image Copyright: Roger L. Martin, 2021

I am utterly tired of the modern strategy focus on averages, whether means, medians, or modes. It is Achilles Heel of data analytics/Big Data/Artificial Intelligence. That is why I am dedicating my 31st Playing to Win/Practitioner Insights (PTW/PI) to encouraging strategists to go beyond obsessing about averages. (Links for the rest of the PTW/PI series can be found here.)

The Modern Focus on Averages

In world of strategy, all analytical guns have long been trained on means, medians, and modes. What is the most representative customer behavior? What is the biggest customer need? What is the average cost of our product? How do most people get information about our product? What is the average life of our customers? What is the biggest segment?

Data analytics just exacerbates this tendency. It generally involves compiling a big data set and then applying analytical techniques to look for patterns. What, generally speaking, constitutes a pattern? It tends to be lots of something: the middle of the distribution. What do most customers buy? Where do we make the most profit? Etc.

It particular, data analytics is about finding lots of something that has already happened. As I have argued for decades (for example here and here), 100% of the world’s data is from the past. So, when data analytics looks for its prevailing patterns, it is looking for patterns that are prevalent in the period leading up to the present.

It isn’t wrongheaded to seek better insight on what, in the main, is happening now. It is valuable, but for one thing and one thing only: optimizing what is. That is what comes out of data analytics in strategy. As long as the world stays the way it is, optimizing to how things currently are now, on average, is sensible and valuable.

Aversion to the Outliers

The flip side of obsession with averages is aversion to the outliers. They are largely ignored if not entirely suppressed. Until I spoke to one of the world’s leading Autism Spectrum Disorder (ASD) researchers, I had always thought that the Human Genome Project, biggest medical project in medical history, had sequenced the human genome. Instead, researchers took several hundred individual human genomes and created a kind of average composite genome which was then sequenced. All anomalies were eliminated. They were no longer in the picture during the sequencing process and today it is as if outliers in the genome never existed.

There is a cost to this aversion. While the mean tells you about what is operating today, outliers give you hints about the future. The above ASD researcher, Dr. Stephen Scherer, is responsible for a cornucopia of new knowledge and insights on the genetic roots of autism. He asserts that all of his research breakthroughs have resulted from looking for genetic anomalies — things that shouldn’t be there based on the averages as represented by the Human Genome Project results. He jokingly describes his success as stemming exclusively from rifling through the garbage can of genomics!

Scherer’s method and experiences are generalizable. Outliers are a window on the future. Science fiction writer William Gibson was right when he argued that “The future is already here — it’s just not evenly distributed yet.” The future is revealed in the outliers that will become the mean someday. Bob Dylan was booed at the Newport Folk Festival in 1965 for the outlier affront of playing an electric guitar in a non-rock festival. Within a decade, it would become the mode. Mutual fund buyers were fringe investors in the 1960s and 1970s. But they saw something that mainstream investors at the time had not yet figured out: a basket of stocks was less subject to manipulation by insiders than an individual stock. Now mutual funds (whether traditional or exchange-traded) is the dominant form of stock ownership by individual investors. Steve jobs wasn’t paying attention to corporate users of computers but rather the fringe of hackers and geeks — which in due course became you and me. MAC Cosmetics designed its offering for professional make-up artists but in due course became mainstream. Shopify helped the long tail of small-scale creatives sell online and now is the back end for huge mainstream companies.

The idea that today’s fringe ideas often migrate to the mainstream is not a new insight. In 1997, Malcolm Gladwell wrote one of his legendary articles on this phenomenon in The Coolhunt. There have been entire books written about learning from the fringe. It has long been clear that deeply probing what is going on with and what inference to draw from the behavior of outliers is really smart.

The problem is that our analytical mindset and toolbox has migrated away from paying attention to the outliers. In fact, there is an entire data analytics practice area that provides instructions for preventing outliers from ‘threatening’ your data analysis. You want to remove those nasty outliers from your precious dataset! Much of the suppression occurs without us realizing it. I suspect like me, you thought that medical researchers sequenced a human genome. We don’t realize that we are obsessing more about what is while increasingly averting our eyes to what would help us imagine what could be. What could be doesn’t appear instantly from nowhere. It is already lurking in the fringes.

That is understood in one part of the data analytics/artificial intelligence world, mainly in the world of security, where anomaly detection is key. In that world, what doesn’t kill you is everyday traffic, it is the anomalous terrorist-to-terrorist message. Or what doesn’t kill you is the mean performance metrics of your nuclear reactor; it is the anomalous signals of fission gone rogue. So, it is not as though sophisticated data analytical techniques can’t pay attention to and draw valuable insights from outliers — it is just that in modern business they rarely ever do. And that is one of the reasons that there is a strong correlation between the rise of data analysis and the dissatisfaction of corporations with their pace and level of innovation. You aren’t going to find new things obsessing about averages.

Practitioner Insights

The single most important thing to do is to think consciously and carefully about when you pay attention to averages and when to outliers. When you want to hone and refine what is, focus on the averages and ignore the outliers. Outliers will distract you when creating improvement for the middle of the distribution is your goal. That is why the Human Genome Project focused on a composite average genome — though much of the advance hype about it finding fantastic breakthrough insights was not realized.

If you want to figure out what could be, you have to pay attention to the outliers. If you want to figure out new things about an outlier condition like ASD, then you need to pay attention to data gleaned from observing outliers. You must avoid listening to the siren song of the averages that is often embedded in the analytical tools you are using. If a data analytical specialist gives you results from an analysis, it is more than likely that the outliers have been scrubbed from the dataset without you being told. That is why it is always important to ask to see the outliers along with the averages.

When you think about outliers, ask what about the outlier is totally sensible. Don’t focus on what is silly, stupid or egomaniacal about the outlier. Dylan had a reason for playing an electric guitar. Mutual fund investors weren’t necessarily unsophisticated. Hackers and geeks just were more willing than the average user to sacrifice computing power for power over their computer. Focus on what could be a not yet evenly distributed future. That will help you be a better and more creative strategist.



Roger Martin

Professor Roger Martin is a writer, strategy advisor and in 2017 was named the #1 management thinker in world. He is also former Dean of the Rotman School.