Business lessons from the world of sports
The story of ADVANCED STATS ( basketball analytics ) and why they are relevant for consumer businesses ?
How to use analytics for real impact
I will start with an old ( this was a few years back) question about analytics and data in the world of professional basketball.
Why do you think coaches and team executives had no confidence or belief in analytics and data ?
Their critics would argue that these guys were biased by their old methods and not objective enough to see the power of data.
They would also make the case that these coaches and sports executives were poorly equipped to operate in an analytics driven world and were therefore just fighting for control and their own survival by stopping the advent of data.
They were partially right on both counts. But
What about the fact that these guys had actually built successful teams, managed them well and won games and titles. They definitely knew something useful about the game and its tactics and the best practises of coaching.
On the flip side, how many games had been won by the analysts.
Zero.
In the end, we realised that both these groups were right and wrong at the same time.
They were just not understanding each other
The coaches were not against data and analytics per se. They wanted better data which was more meaningful in the way they broke down the game and made sense out of it. Which enhanced their understanding of players and tactics and answered the questions they wanted to ask.
Like, how to identify a player who doesn’t score a lot but makes everyone around him better. And measuring assists is not the answer to that.
Or, how to identify the opposite. A player who puts up a lot of numbers but never wins anything.
Or how to identify the best combination of players to play together.
Or, how to to value the worth of a player who takes on the most important defensive assignment of stopping the opposition’s best player and does it well.
They had learned to answer these questions based on observation and instinct. While these methods came with biases and were not fully scientific, they had been right enough times to justify their utility somewhat. These guys had enough wins and titles to show for it. They also knew that getting better answers to these questions would make a massive difference. They were likely to be more open to advanced methods and tools which gave them those better answers and allowed them to improve their decision making on things like player selection and match strategy.
In the sports world , they had a term for these instinctive things which you could “see”. They called it the “Eye Test”. If a player made others better, then he should have passed the eye test. They could see it in the way the game started flowing better once that player came on. And if everyone loved playing with him. It was a sacred thing. Knowledge and wisdom gained over generations. There might be some biased observations, but a lot of it was right. It was still a game and you could “see” how it was being played.
The problem often happened because the insights from the analytics and data would either not match the eye test answers or not manage to provide any answers at all.
You needed analytics which worked with the eye test. Made it better and more scientific. Allowed the coaches and scouts to make better sense of things they knew. Not tell them that everything they knew was wrong.
At the heart of this problem were two fundamental problems with data and analysts. ( Read Nate Silver’s book The Signal and the Noise to know more about this subject in greater detail)
1)Most problems in the real world are fairly complex and actions are impacted by multiple variables. Many times, if you are not very well versed with the subject matter, you won’t know all the variables that were relevant to the problem. (A variable is a factor which affects something.)
Eg. Basketball teams often play their reserve players if they are leading or trailing by a lot of points in the final stages of the game. Any stats accumulated during that time don’t matter. There’s a reason that type of situation is called garbage time.
Your shooting percentage varies significantly depending on whether you get a pass in an open position or whether you have to dribble and create space for yourself to shoot.
These are all the nuances that a professional player or coach would know but an analyst would not.
2)Second, you would often not be able to measure or quantify a variable. There would be no available data for that variable.
Eg Good defense means not letting a player take high percentage or easy shot. You do that well but the other guy still makes that difficult shot. How do you factor this variable into that player’s defensive impact. If you watch the game and note it down every time, you will know. But its not available for use.
In most cases, the analysts world by ignoring these two problems. Needless to say, they often failed to give answers to the problems or gave poor ones.
This is exactly what happened in sports. You started with the available data and built models and insights. And you wanted people who knew the subject really well, to unlearn their partially proven knowledge and relearn this new incomplete imperfect model.
Limited models like this couldn’t explain the real world.
You needed a model which first understood the real world and its important variables and then found the right data ( or workable proxies ) and built a model to better explain the real world and the eye test.
Luckily for a lot of sports teams, that eventually HAPPENED.
There were new models which were built keeping in mind eye test problems and they made a massive difference. New data points were created and measured for this purpose. Analytics companies and teams which invested in creating these benefitted hugely. Everyone could use the same statistical models and data science techniques. The teams with access to better data won.
The world of ADVANCED STATS was born. The power lay in these new data points which didn’t exist traditionally. New technology and cameras were invented to capture the data.
Like, knowing that when a certain player was defending the basket, what was the shooting percentage for shots taken from within 5 feet. Or how many shots were attempted as a percentage of total possessions. As compared to other players. This was done by some scouts manually for a few games in the old days. It had all kinds of manual errors. Now you had special cameras capturing this data perfectly.
Coaches always knew that great rim protectors players forced other teams to change their approach. And now they had data which showed how these players impacted shot attempts and shooting percentages.
Coaches knew that all shots were not equal. And therefore players couldn’t be compared by points and shooting percentages. Now they have stats which measure shooting based on whos guarding, difficulty, game situation everything.
The eye test and the analysts were now talking the same language.
This became a huge competitive advantage for certain teams. Like the medieval colonialists who used guns to decimate the helpless opposition who were armed with bows and arrows.
If you are interested in ADVANCED STATS, then start following the NBA and check out Cleaning The Glass and Second Spectrum.
So how does this ADVANCED STATS apply in the business world ?
In exactly the same way. Start with the eye test understanding and then create data and build models. It can be used for consumer analysis as well as employee performance measurement.
As we saw in sports, the success of ADVANCED STATS will depend on superior eye tests and access to better data.
Let’s take an example of both and explain. (Please bear in mind that I am not going to share the data or variables being used for these examples. They are my guns)
Let’s take the sales teams first
The easy traditional way is to measure everyone on revenue, conversion and NPS. Everyone does that. But a great eye test will tell you that it doesn’t capture a lot of things.
One useful eye test for a Sales leader is the “Right Behaviour” test — the set of things they want their teams to do and not do for every customer interaction.
Like shooting opportunities , all customer conversations are not the same. Anyone can convert an existing customer who just needs to confirm whether the order will be delivered in 3 days or not. It’s much harder to convert a new customer who is calling from a small town and has never made a transaction online in her life.
You know what you will be able to do if you get the right data to understand these nuances.
Now let’s look at a customer example. Let’s talk word of mouth.
There are enough studies which tell us that the digital share of word of mouth is only around 15%. Which means that word of mouth is one of those variables which won’t get measured and won’t factor into your consumer and marketing analytics.
Now imagine your eye test i.e consumer feedback is telling you that word of mouth is very important for you. What do you do now?
A smart company on the other hand will find a new data point to measure it. This will mean that it can segment customers based not just on their LTV (Lifetime value ) but also their word of mouth propensity. It can then factor into acquisition costs and become a basis for measuring campaign performance.
A company without ADVANCED STATS is doing what everyone else is doing. And if you do that, HOW can you expect to have an edge over the competition?
You will be the one with a bow and arrow when the others will have guns.
