Can machine learning help improve CLTV in retail?

Customer Lifetime Value (CLTV), along with CAC (Customer Acquisition Cost), are two aspects that will determine the success of any retail company. It goes without saying, therefore, that these are two metrics that need close and constant monitoring.

To be profitable, a company’s CLTV should be less than the CAC. Obviously, the lifetime revenue generated from a customer should exceed the cost of getting them on board in the first place. Whilst getting them on board (the costly bit) is easy, the real challenge lies in retaining that customer.

So can machine learning help with this?

There are a lot of things that machine learning (also Deep Learning) can do, but bear in mind that it cannot do everything. The human element remains, and neither can accurately answer all problems every single time.

A machine learning algorithm can tell you if a customer is at risk of being lost (“churning”). If it is sufficiently sophisticated, it will be able to tell you the reasons why. In order to do so, it needs access to data about the customer journey.

Knowing where certain strategies have worked before (for example, a discount coupon of a specific value), it will be able to tell you what strategy or strategies it ‘thinks’ will work on a particular customer. What it cannot do is come up with a new strategy for customer retention on its own. Everything the machine learning algorithm knows comes from prior observation. After all, it has no capacity for creativity (at least, not yet).

Despite these shortcomings, the mere fact that machine learning / deep learning has the ability to tell you who is likely to churn is hugely valuable in itself. It justifies its use in the ROI to be seen from these observations. It can do this at a level that is much more sophisticated than that which can be ascertained by humans alone, quickly able to establish innovative and meaningful connections between data points and to do so significantly more frequently than a human.

It is this ability to make novel connections that places machine learning over and above the human eye. Without bias, and without any breaks, it tests and processes all manner of variables and relationships between data points, defining criteria for success or for failure and improving its intelligence at every point along the way.

What Of The Human?

Whilst the human is, clearly, less capable of generating answers at this level of sophistication, it is far from obsolete. As aforementioned, the capacity for human creativity and contextual insight is something that a machine just cannot achieve. It can tell you what’s

going on, it can suggest answers based on what has happened in the past, but it cannot tell you what to do in that particular scenario. There are too many contextual variables to take into account (what discount your company can afford to offer, for example), and – ultimately – the machine does not know humans like humans know humans.

Though, by its very nature, deep learning improves its knowledge over time, real-world criteria, particularly in retail, are always in flux, subject to change that is only predictable within the context of a marketing department.

However, this iterative learning ability does make machine learning and deep learning an ever more indispensable tool in your arsenal. The longer you use it, the more insightful it becomes. The more knowledge it is fed, the higher the ROI will be. Cumulative value, ultimately, is what you get.

So, to return to the initial question: can machine learning help improve customer retention? The answer is, yes, absolutely.

Machine learning’s ability to identify relationships between unforeseen data points, combined with the speed at which it can do so, make it a crucial member of the team (as it were). As datasets continue to grow and grow, so must the uptake of machine learning within retail. Without it, your business may already be falling behind.

What do you think? … I’d love to hear your thoughts!