This is a guest blog post by Nick Clarke, Head of Analytics, Tessella
Many successful companies, large and small, are running data analytics projects and initiatives. These projects can and do deliver huge impacts to the bottom line. But they also have a bad habit of taking so long that corporate backers lose interest or move on to new projects. If analytics projects are to succeed – and teams secure funding to keep doing them – they need to deliver on their promises rapidly.
A reason many don’t deliver is a lack of appreciation of the portfolio of skills and level of experience required for different projects. Much of this stems from the use of the term ‘analytics’ to refer to everything from simple number crunching to incredibly sophisticated maths and programming based on diverse, unstructured data. Different projects require many different skill sets, yet ‘data analytics’ often gets grouped under one umbrella. The result is people who understand one approach to data end up on projects that require a completely different approach, and end up taking too long to get it started.
Identifying the right people is indeed tough. The visible rise of analytics and data science has led to a plethora of new training offers – from short courses to complete data science degrees. Alongside this new wave of data professionals, an ever growing range of self-service tools are appearing, all designed to provide easy-to-use analysis and visualization of data sets.
Many of these courses and tools are excellent and bring good value to a wide range of problems. For clearly defined needs, such as customer churn or retail analysis, a well designed platform can often solve your problem quickly and effectively.
But these advances also lead to a false sense that data analytics can be commoditized. Some can to an extent. But the most valuable analytics are often about tackling business transformational projects in diverse and completely new areas of the enterprise.
These high-value projects often need people with high-end statistical and data skills to be successful. But even more critical is battle-hardened “been there, solved that” experience of applying analytics to hard real-world problems. They need people who are familiar with how a wide range of models and analytics tools do and don’t work in the real world, can identify which one to apply to which problem, and can refine and refactor those models to the degrees of accuracy required by the challenge at hand.
Such projects also require data be understood in context. Data is not just a string of numbers, it represents something – a machine or an engine or a plane or a person. Someone experienced with oil well drilling for example will be able to spot if a change in a data stream is indicative of a problem, or a normal part of drilling operations. Someone trained in only data, will only know that a new pattern has emerged, not what it means.
That’s not to denigrate newly trained staff or self-service tools; both can play a role in complex projects. Enthusiastic graduates should be supported to gain this experience. But organisations must recognise that these skills are complicated and take time to acquire and mature – far longer than the duration of any single project.
An enterprise shouldn’t fall into the trap of thinking that data is new and must therefore be managed or manipulated by a new generation of analyst. Data analytics projects must be recognized for what they are, and an appropriate level of experience applied to them. If the goal is business transformation, they need people with proven experience of delivering business transformation data projects.
Understanding what your project truly involves and resourcing it with the right level of experience is critical to its success. Get it wrong and the project may suffer false starts, take longer and maybe never deliver its business case at all.