In spite of the growing attention enterprises attribute to data, it often isn’t clear where responsibility for data analytics lies within large organisations. It could be the Chief Analytics Officer or Chief Data Officer, but it may just as likely fall into the camp of the CIO, CMO or even CFO. More often than not, whoever is responsible for data analytics is likely to hold more than one of these jobs at the same time. Let’s call them the C#O.
Companies have made huge progress in tracking data, analysing it, and seeing the value in the insights, but the lack of consistency in job title is indicative of the state of analytics more generally. The climate is still like the Wild West, with models for governance, prioritisation, data and information architectures, talent development, and approaches to problem solving either completely absent or, at most, incipient. As the person in charge of data analytics, it’s your responsibility to bring law and order to the Wild West, or face the consequences.
As is often the case, this is easier said than done when you are operating without any frame of reference or creating it as you go along. Common stumbling blocks for the C#O include juggling multiple sources of data, prioritising where to use analytics and creating real, actionable insights, among many others. As C#O, your concerns can typically be divided into two categories: those about data and those about decisions. Of course those two categories are tightly interrelated, but this divide is a fine line for any C#O to walk, and very tricky to get right.
While the C#O’s background is a big factor in terms of which side they fall on, there is a prevailing tendency to focus on dealing with data related issues above everything else. It’s logical: we need data before we can do anything with it. However, without knowing what decisions we need to make, how can we know what data to collect in the first place?
In striking a balance between data and decisions, there are five specific balancing acts C#Os need to consider.
Creation and consumption
There are so many statistics you can cite predicting the massive future growth of data. IDC thinks we’ll produce 44 zettabytes by 2020. The problem here is that the production of data is outpacing our ability to analyse it. We can create metrics, graphs, and dashboards, but unless these generate actionable insights, they’re useless. To ensure that data collection does not become ‘art for art’s sake’, it needs to be driven by the people making the business decisions and their need for information and insights.
Questions and answers
Anyone who has played a game of 20 questions knows how tricky it is to ask the right questions to get the right answers. We’ve all felt under pressure at one point or another to have all the right answers. Your team is in the same position but they’ve got to make sure they’re asking the right questions first.
When we’re busy, it’s easy to jump conclusions but the risk of this is lower if we start by asking the right questions. One method to get at the root cause is to ask a series of funnelling questions, even adopting the 'Five Whys' approach whereby you repeat the question 'Why?' and the resulting answer forms the basis for the next question.
Short term gains and long term success
Members of your team have probably felt like everyone wants something from them at the same time. Firing off answers and possible solutions as quickly as possible may seem like the right thing to do here. However, time pressure should not stop the team from thinking carefully about the problem that needs addressing. It is up to the C#O to help their team come up with a problem solving process rather than quick-fire responses. Spending a bit of time to scope out this process will save time further down the line and deliver better results.
Here’s one process outline to consider:
- Start with measurable outcomes and define the behavioural changes needed to drive these desired outcomes
- Map out the relationship between different business problems – you will often find that solutions emerge thanks to identifying the links between problems rather than by addressing the problems separately
- Select the most suitable combination of analytical methods – such as descriptive, inquisitive, prescriptive, predictive – for the problems you are aiming to address
Governance and empowerment
In line with the Wild West metaphor, governance when it comes to analytics can be very convoluted. Information governance and data governance get mixed up with analytics governance, creating confusion for C#Os.
To satisfy governance needs, most C#Os focus on risk mitigation, whether it’s through decision rights, policies, or standards. This is understandable given the number of threats business face on almost every week: LinkedIn, Google, and TalkTalk have all been recent victims of a data breach, impacting their customers’ data.
This control over data needs to be balanced with access afforded to those in the business who really need it.
Also keep in mind the role that Gartner calls the Citizen Analyst. They’re out there. They’ll create new Tableau cockpits, hire their own data scientists, and procure their own modelling tools. You won’t stop them. But you must find a way to give them some fuel to go along with the guardrails you set up for their work. That fuel can manifest in the form of making a clear litany of services and tools available, offering flexible resource capacity, helping to assemble funding requests, even training on different problem solving techniques.
Technology and culture
Our society places a strong focus on technology - to automate processes, communicate, and enable greater productivity.
This makes us more efficient, but can affect the internal culture. The commonly cited example is when we choose to email a colleague rather than walk the three metres to their desk. For the sake of protecting our professional relationships, we need to strike a balance between technology and culture.
In the same way, we need to balance technology and culture when it comes to data analytics. It isn’t just about having the latest software and functionality. Your team needs to have the right problem solving mindset to go with it.
With this problem solving mindset, be sure to encourage a fail fast and often, but cheap mentality. Make sure your strategy involves tight feedback loops so lessons learned are incorporated into future problem solving. We’ve found that this makes us more efficient and improves performance.
Defining the analytics needs of your company and implementing them is part of your job but it isn’t easy. We’ve seen how understanding the Wild West landscape is a challenge for C#Os, but what is particularly difficult is knowing what data to collect to inform business decisions. That said, the way C#Os can succeed at this is with a careful juggling act of different aspects of data analytics. Moving the focus towards a balance of data and decisions will go a long way to satisfying your board.
Tom Pohlmann, Head of Values & Strategy at Mu Sigma