Why do analytics projects fail?

Analytics is in a state of flux and companies often rush in before they are ready

The integration of analytics with business processes and customer interactions promises great advantages to organisations of all sizes, offering new insights, novel decision-making tools and real-time feedback. But all too often such projects end in failure. In fact, Gartner reports that only two out of five analytics initiatives go beyond piloting and experimentation.

Not all organisations have the necessary structure, experience and culture to deliver big data and analytics initiatives without some serious preparatory work. The success of such projects is dependent on having a number of factors in place, for example the right skills, agile methodologies, data integration, support from the top and proper data governance. Further, a project needs to focus on a discrete business outcome.

Typically, analytics projects are initiated by an enthusiast in the IT department keen to play with supposedly world-changing technologies such as Hadoop, Spark or NoSQL databases, or by someone in the business who has no idea about the technology involved. This can lead to the cart being put before the horse, with businesses value tacked on almost as an afterthought. More deliberative questions such as "Why should we do it?" and "Is there another way?" tend to get drowned out by more imperative ones such as "What can this new technology do, and how can we use it?"

Initiatives based on shaky foundations will fail, but even those that make it past the pilot stage by involving the business early and focusing on delivering value may face issues further down the line. They need to deliver RoI on the budget allocated to them and this may be easier said than done as success is predicated on broader changes throughout the business. Change is likely to be slower and more expensive than anticipated.

Recent Computing research among more than 100 IT decision makers asked about the goals of analytics programmes.

In terms of business outcomes two thirds of our respondents were focused on operational efficiencies and cost reduction - classic IT goals. A close second came improving customer experience, which ties in very closely with the digital agenda being pursued in many organisations currently; obtaining instantaneous feedback from customers' behaviour is valuable for all sorts of sectors, not just the obvious ones like e-commerce. And third came greater visibility and speed of decision making, again something that places the emphasis on real-time feedback.

However, when asked whether they have a well-defined strategy for delivering these outcomes, the answer was far from clear.

Only 25 per cent said yes, outnumbered by the 31 per cent saying no. For the largest portion such planning was still a work in progress. This shows that as analytics moves from the rear-window view of BI towards real-time and predictive tools, organisations are in a state of flux. Another question showed the majority of respondents to be in the middle of an ongoing analytics programme.