How to avoid turning data analytics into an art for art’s sake

Big Data analytics has become a major focus for companies of all sectors and sizes, with global revenues predicted to reach $16.9 (£12) billion in 2016, an increase of 5.2 per cent from 2015.

In spite of all these investments, a new report on behalf of Mu Sigma has revealed major gaps when it comes to managing and getting the most out of analytics.

While two-thirds of the senior decision makers in large multinational businesses who were surveyed acknowledge the impact that analytics can have on business growth, 41 per cent felt that their ability to glean actionable insights could really improve. What this says to me is that everyone is aware that data analytics can really help them, but they don’t always know the best way to use it.

There is a strong indication that a deluge of data in large corporations has contributed to issues of quality, consistency and usability. Getting on top of these issues is critical, especially given that the study highlights that a well-structured and -managed approach to data analytics is essential for business success.

In my experience, there are a number of possible causes for the patchiness in how multinational enterprises tackle problem solving and analytics.

What comes first - data or decisions?

To start with, organisations do not apply the same rigour to analytics and problem solving as they do to other more established functions like corporate IT or governance.
For instance, we found nearly 40 per cent of those surveyed indicated they do not have a standard methodology for solving business problems.

Furthermore, only a quarter of companies said that they begin with a business outcome in mind, while nearly three quarters start from the data they have access to and build their analytics from there.

While it is laudable that existing data gets used, this is like putting the cart before the horse. Data-related considerations should not be prioritised over what the company needs to know in order to make critical business decisions.

Thankfully, the overwhelming majority of participants (70 per cent) acknowledged that, to varying degrees, they plan to make improvements to their approach and have a clearer roadmap of analytical business problems they want to address in the coming year.

A lot of businesses are also on the right track in terms of identifying which areas they need to focus on. Over a third pointed to tackling data challenges such as quality, consistency and availability and a fifth highlighted organisational challenges such as setting up and governing analytics. Talent shortages and lack of training were also highlighted as significant areas of improvement.

As they go about enhancing their analytics function, they can take heart from the strong connection between business performance and analytical rigour that the survey revealed. Those firms who have met or exceeded stakeholders’ expectations are nearly four times more likely to use a consistent methodology for analytical problems.

Who’s in charge?

Another issue that needs to be addressed is the disparate ownership of data analytics within organisations. When asked who has overall responsibility for analytics within their organisation, specialists such as Chief Data Scientists, Chief Data Officers and Chief Analytics Officers jointly made up only 25 per cent.

In many large organisations, analytics is still owned by the Chief Information Officer (CIO; 23 per cent) or the Chief Finance Officers (CFO; 17 per cent).

Of course the background and the respective priorities of whoever ‘owns’ the analytics functions – whether it’s finance, IT or marketing – strongly influences how data strategy is formed.

Add to this that responsibility for data is often shared between various functions, and it’s easy to see how the large disparity of approaches to analytics comes about.

It is not surprising, then, that successful firms are three times more likely to have a specialist such as a Chief Analytics Officer in charge. Making a dedicated, expert board member responsible for data analytics emerges as a major success factor that other companies should learn from.

Governance is a tug of war

Perhaps unsurprisingly given the multifaceted approach to ownership, the report also highlights a diverse range of governance models. Most use a centralised model (44 per cent), where a single group provides analytics services to the rest of the company. This is understandable given that analytics is still largely owned by CIOs, and IT is typically organised in a centralised manner.

Another 22 per cent apply a decentralised model, where individual business units are responsible for their data, and 16 per cent have adopted a federated approach, blending the two other models.

Among those who were intending to change how they govern analytics, there was a pronounced preference for greater centralisation. This is likely a sign of those companies seeking to gain more active control over their data.

While having a centralised governance model is better than not having a governance model at all, it is going to be less flexible in terms of tailoring data collection to the varying business needs across the organisation. One size doesn’t fit all, and you may find a better candidate in a federated governance model which marries centralised tools and capabilities with empowering front line staff.

Looking beyond the ‘here and now’

What also becomes apparent is that too much time and effort is spent simply organising and reporting on what is happening in their business. The bulk of analytics work still centres on the descriptive, so the ‘here and now’ (39 per cent).

Instead, companies should focus much more on asking ‘why’ and ‘what next’ in order to prepare better for the opportunities and challenges that lie ahead. Right now, only a fifth of fims’ analytics efforts are spent on predictive analytics to answer these forward-looking questions.

Why this still happens may be due to a lack of bandwidth and limited resources being dedicated to analytics, stopping Big Data from delivering on its potential. It’s also conceivable that decision makers prefer to stay within the comfort zone of the descriptive rather than sticking their neck out and trying to second-guess what the future holds.

Overhauling company culture

Underusing data analytics may also be rooted in company culture. Among the most successful companies, 60 per cent adopt a ‘fail fast and fail cheaply’ mentality to help them identify the right mix of different types of analytics to achieve a competitive edge.
Furthermore, over two-thirds of these companies say that they look outside their industry for learnings and practices to make improvements to their business.

This suggests that a key ingredient to problem solving and analytics success lies in taking a more creative, experimental and interdisciplinary approach to problem solving rather than sticking with established methods and processes.

Where next for data analytics?

In summary it seems that, to reap the rewards of Big Data and justify investment in the analytics function, it will be critical to treat analytics as a formal discipline – much like IT or HR, for example.

This means putting specialists in charge and empowering them to create a stringent governance framework. The latter needs to be driven by business problems rather than being led by the technical possibilities that ever-growing data streams afford us.
Finally, communicating actionable insights is essential to business success. Three quarters of the most successful companies we surveyed are focused on making their insights both consumable and actionable within their organisations.

Data analytics should form the base of a new ‘art’ of problem-solving, not become an ‘art for art’s sake.’

Tom Pohlmann, head of values and strategy, Mu Sigma