Success with analytics: what does it take?
There seems to have been a steep change in how companies are approaching the question of analytics. In the last few years or so, we have seen a much greater acceptance that analytics is an essential. Companies have woken up to the need to be data-driven. They understand that there is value to be created through the use of all the data that they have kept and stored, often since the creation of the company.
However, understanding that there is value in analytics, and actually generating that value, can be a world apart. Increasingly, we are finding that we are being asked how to work with analytics to best achieve success. The first answer is that it depends. There is no ‘one size fits all’ recipe for success. However, I suggest that there may be two key issues that all companies would do well to bear in mind. One is strategic, and the other is very much operational.
A strategic decision
The strategic issue is about the approach that the company takes to analytics. Many companies look for examples and case studies of successful uses of analytics, and then try to emulate them. The problem with this is that the most high-profile users of analytics are companies like Netflix, Amazon, Spotify and Google. Most of us are aware of their algorithms, and how they help us to find what we want even—or perhaps especially—when we don’t know ourselves. However, not every company is a Netflix or Google—and these companies’ approach doesn’t work for everyone. The big tech companies have a huge advantage over many older or more traditional companies: they are digital natives. They were born of technology and analytics, and the use of data is hard-baked into their DNA. That is just not the case for many other companies. Retailers may be keen to emulate Amazon, but that may be impossible with legacy systems and ‘bricks and mortar’ stores. Instead, you have to find an approach that fits your company, your customers and your needs. This may seem harder than picking up a template from elsewhere, but in the long run it will pay off much better. It may seem odd to need to say this, but your strategy needs to fit your business, not Google or Amazon.
Into operational issues
The second reason why I think organisations can sometimes struggle lies in the Venn diagram below.
Traditionally, IT (or computer science in the Venn diagram) has been a support function for the business, a bit like the human resources department. It generally exists to serve the business. However, in the new data-driven world, IT is a key player. It has a seat at the table, and is on equal terms. If the IT function is not engaging with data science, you have a problem. I have written before about this partnership. I commented that data science functions can be split by the level of expertise required from each of IT and the business side. For example, deployment work might best be seen as an IT function, but selecting data is far better done by business users. However, moving to this partnership approach can in itself create friction from both sides. The business side may not like seeing IT as a partner—but it is also not necessarily how IT teams like to work. Many of those in IT like their separation from the main business, and it can be hard to make the leap from being a supporting player to becoming one of the stars.
There may also be a fundamental friction in how different groups view data. Data scientists and analysts tend to see data as a resource. However, IT teams tend to see it as something to be stored and managed. To IT teams, all data is basically the same: it’s a problem that needs managing. It can therefore be challenging to uncover what data is available and put it into a usable form—and it can also be difficult for analysts to understand this.
A changing landscape
The road to success with analytics may therefore lie through foothills of organisational culture, data access and quality. Not to stretch the analogy too far, you may have to make considerable investments in building basic roads and bridges across functions before you can cross those foothills. It may not sound as interesting as analytics, but it is an essential underpinning. Bad data makes for bad analytics and poor outcomes.
