Phil de-Glanville, former England rugby captain and a player who straddled both the amateur and professional era, talked Deloitte through some of the changes that started to occur as the drive to enhance performance created a desire to deconstruct, measure and improve.
Speaking of his time with England: “Sir Clive Woodward broke the game down into components, very specific parts. It wasn’t just forwards and backs, he also looked at defence, kicking, and ultimately developed a view of how each individual player should contribute. He wanted to make everything measurable and work out how to improve even the smallest component. Defending particularly improved massively as a result.”
This deconstruction of the thing which you are trying to improve is the start. The next step is to represent the relationship between all the components in data - so that each component can be tweaked, and the impact tested against a baseline. For true data-driven thinking, this is a must-do.
Michael Bourne, who is now Head of Science and Medicine for the England and Wales Cricket Board (ECB), having previously held the role of National Lead for Performance Analysis at the ECB, explains how he tackles this challenge: “I build a model of the sport to find the factors which explain the outcome. In the first instance this would be theoretical modelling of the first principles of the sport. You use a hierarchical model where you take a discipline and you define mathematically what it takes to, for example, throw a javelin 90 metres, or run a 100m in less than 10 seconds. You use this model to work out where your interventions are necessary [i.e. where you will try to make a change] and what data you need to collect to explain that particular part of the model.”
In a business context this might involve modelling the relationship between different departments, as a simple example: if the sales team increase their efforts, what impact does that have on the inbound call centre? It is impossible to make an informed decision about a change to strategy in one division, if you do not understand its impact on another.
“In a hierarchical model you know the inter-relationships of all of the factors. In long jump for example, you would know the direct impact on distance jumped if you increased your run- up speed by 2 metres per second, and you would also seek to understand what positive or negative effects increasing run up by this speed may have on other factors. In more linear sports like rowing or cycling you can do this mathematically. In more complex sports like football or rugby you work on more logical principles which you can confirm using statistics.”
It is with this model, and the underlying data supporting it, that you can begin to chip away at the cultural barriers to data-driven thinking, and start to break down a reliance on the HiPPO.
Michael Bourne: “When I used to work in Olympic sports there was a belief that Judo champions needed a minimum of 4 techniques in 4 directions, but when you go back and analyse previous champions, they often only had 1-2 techniques in 1-2 directions. They were so efficient that that was all they needed, contrary to the opinions of many.”
But Bourne points out that there is not necessarily a one-size-fits-all model for a given sport, it is important to understand the specific metrics which pertain to your team’s situation
“You can’t take a non-representative dataset and try to extrapolate that out into the wider population. One example in cricket would be looking at how Australia played when they dominated world cricket. Their process and methods may not be the best way to do things for English cricket, as many things are specific to Australian cricket [for example weather and pitch conditions] and aren’t representative of the English game.”
Many businesses do not truly understand the interrelationships between the various component parts of their operations, and consequently when seeking to make changes they often make decisions based on their instincts or anecdotal evidence, or experience of another organisation which may not be relevant. To properly answer questions such as “why do we keep sending incorrect bills to our customers?” or “how can I reduce inbound calls to my contact centre?” requires a model to be built which reflects all the factors which might impact upon that outcome. With this up-front investment, optimising operations is hugely facilitated.
David is a Partner in Deloitte’s Enterprise Risk Services practice specialising in data analytics, data management and cyber security. David has worked with many of the UK and Europe’s leading Telecoms organisations, and has deep expertise in helping them secure, manage and derive insight from their data.