Data analytics: everybody knows why - but few know how
If you’re from France or Italy, you’ll know about mirrors and nightingales. If you’re from the UK, you’ll know about red herrings. Different languages have different idioms to describe distractions – things which stop you finding what really matters.
In recent discussions with water companies, I found out that big data can do exactly that.
Everybody knows that data analytics is a hot topic, and that the growth in affordable sensors and the IoT invite data-driven insight into business operations.
But if you start discussions with analytics, there’s a risk to move into territory which is at best a distraction, and at worst can lead to reinforcing false suppositions.
It’s a risk we all know well: that without a wider and more open context, all data does is telling the story you want it to tell. Let’s look at a real example.
Define purpose – what is data for?
Never lose the big picture. It’s easy to say but not so easy to do. When companies start analytics initiatives they rarely get the expected results. The use cases they imagine at the start are rarely the ones that emerge as the program progresses. So how can you avoid this mismatch between expectation and results?
Before even beginning a proof-of-value initiative with a client, at Atos, we apply a strong methodology based on “return-of-experience”. We call the first step “design thinking”, an approach combining analytical and intuitive thinking. This shared creative process involves a wide group of interested parties including final users, operators and IT specialists, together with sales, legal, finance and directors.
I was recently invited to take a closer look at “revenue protection” by a utility company. They suspected that some customers were by-passing meters to use water without paying for it.
Initial thoughts focused on applying analytics to identify the fraudsters. Seems obvious enough. Identify customers whose water usage deviated notably from that of their peers, and make them the focus for fraud investigation.
Fortunately, we got a red flag before going much further. Representatives from the legal department pointed out that because water is an essential service, you cannot cut supply even in cases of fraud.
The director then asked why the company would invest in a solution if they have no leverage and cannot take action.
So instead of blindly following the “meter data points to fraud” logic, we stopped and took a step back. Rather than simply looking for another clear target for analytics, we zoomed right out to define the most added-value use case that at the end was much more technical that initially predicted.
Zoom out not in
Another example of complex use case definition is summarized below:
Figure 1: Is water quality the use case with the biggest impact?
It shows us that to improve water quality we can develop a predictive maintenance system, a leak-detection system or tools for use in pursuit of operational excellence.
The lesson here is to make data work for you: data value is not always where you think it is.
With the company that started looking at fraud, we went back to their strategic objectives and asked what we, as business technology partners, could do to make the greatest contribution.
There were two reasons for this. First, every new initiative must be paid for, and in an environment in which every organization seeks to minimize operational costs, initiatives without Board support have little chance of success. Leading with analytics is not a recipe for success: the “C” team want to know the objective, not the means.
Secondly, to develop and deploy such a solution requires a strong change management support.
Are you ready for the change?
Defining your objective, making data work for you is great but not enough. The next step is to make sure your company is on board and open for change.
A “water quality” use case may show that it is more profitable to develop tools for operational excellence than to develop smaller non-integrated solutions. The solution will then lead to dramatic changes in the way operations are conducted.
New solutions based on analytics often require new ways of working and of doing things. If you don’t have the commitment at the right level, great ideas may never see the light.
Ensuring executive level support helps address three main points:
- ensure that the use case supports strategy
- safeguard funding not only to develop but also to industrialize the use case
- make sure that change management requirements are covered.
Although we are naturally enthusiastic about data analytics, I learned the importance of a “top-down” approach – about starting with strategic priorities and also of working with as wide a group of contributors as possible.
But perhaps most importantly, I learned just how important it is to let the data do the talking – and to be prepared to be open in analysis. Because success of data analytics is not rooted in confirming prejudice, it’s about revealing substantiated truth – to everybody’s benefit.
Now it’s true, that water companies, like any other organization, are wary about embarking on initiatives where the outcome is unknown. But with a rigorous methodology, this becomes a bonus not a risk.
Working with design thinking workshops and highly-structured iterative reviews, for example, we can ensure that go/no-go decisions are made in a way that is both disciplined, substantiated and, of-course, cost-justified.
With this approach, we can be sure to avoid mirrors, nightingales and red herrings.