Call us bleeding-heart liberals, but we’re generally not big fans of dictators in human societies. We’re all for people weighing up the pros and cons of different stances, making a decision of their own accord and then giving their opinions democratically.
Data, on the other hand. Data is lazy. Data could contain world-changing information but would never volunteer it through sheer apathy. Data needs to be told what to do in no uncertain terms. Make no mistake about it: data needs dictating.
You may have no experience of being a dictator, or you may feel that you lack the required skills to be a proficient dictator. Don’t worry, though, because to ensure that your customer data pulls its weight while remaining under your total control, you just need to remember one simple thing: divide and conquer.
When organisations have large amounts of customer data in one place, it can be difficult to manage. The more there is, the more it obscures any attempt to try and find trends or patterns that could be valuable. What you need to do is split it up into smaller, more manageable and more coherent chunks.
Historically, customer data has been segmented by demographic information or by the groupings that are perceived to be those that an organisation should be targeting. While not arbitrary, this approach still uses external parameters to sort and split data in what may not actually be the best way.
In reality, the best means of segmenting data is based on the patterns and trends within the data itself. If an organisation’s marketing groups are on point, then it may be that the patterns and trends indicate that the data should be segmented in the same way. More likely, though, they will point to another means of segmentation. The problem is, the patterns and trends at play can be well hidden within the data, making them impossible for humans to spot.
Using what are known as clustering algorithms, though, it’s possible to identify which variables can best be used to group customers most effectively. These can then be used to build models that predict how different groups of customers will behave in different scenarios, allowing organisations to plan activity in better and smarter ways.
Of course, being a data analytics provider, we’re bound to harp on about the importance of customer segmentation. You don’t have to take our word for it, though. A great example of how an organisation can use customer segmentation is the Telegraph Media Group. Like many media organisations, it is dealing with an increasingly fragmented landscape, with customers consuming its content in more ways than ever.
At a recent CACI seminar, head of customer insights at the Telegraph Media Group Paul Hatley explained that, to get a handle on its huge amount of varied customer data, the firm has begun to pool it all together in a “data lake*,” from where it begins to identify core segments and then convert unknown customers into identifiable people. The project outputs are, unsurprisingly, very valuable:
“We used to show the journalists a dashboard of things like most-clicked and most-shared articles,” Hatley explained. “The language we used wasn’t about people – it was about audience, and IDs. Now, we show the journalists how their content is driving registrations and subscriptions. It’s about people, and what our content inspires them to do. It’s a fundamental change.”
*When you choose to work with Peak, we give your company access to your very own data lake, a cloud-based data warehouse powered by our own Peak platform. It acts as a one-stop shop for all of your business’ data, with Peak’s in-house team of data scientists on hand to help you uncover new insights and make sure everything is ticking over nicely.