As we have stated many times in this blog, the use of Artificial Intelligence (AI) and intelligent automation is growing. In some sectors this appears to be at an exponential rate and as a result some are leading the way faster than others.
Leading the way is a subjective term obviously. It could mean using AI to solve the most complex of problems but in this case, we mean in terms of its widespread use. In some sectors AI is becoming commonplace and standard.
The prime examples of where this is the case is online retail and marketing. This is often because any improvement in attracting customers or generating sales can have an immediate benefit. If they are able to identify more customers who are likely to buy their products, then sales will increase. The return is worth the investment and, as a result, business cases for technology investment and innovation are relatively simple. This is in comparison to other sectors where the benefits or returns are longer term, but they have the opportunity to catch up.
In online retail and marketing, AI and intelligent automation are being used to enhance customer engagement through chatbots, personalised customer experience and faster, automated processes. These AI tools and capabilities include machine learning, rules-based decision logic, natural language processing and sentiment analysis.
One area that has received a great deal of attention is the use of AI to predict customer behaviour. In particular, customer behaviour when it comes to buying items online and what attracts their attention.
In this blog we will look at this area in more detail and what we can learn from retail and marketing that can be applied to other sectors.
When it comes to predicting customer behaviour and buying patterns there can be quite a lot of variance in terms of quality and success. At one end it can be disappointing, and usually frustrating, when a customer is interrupted during their online activity with an offer that is not of interest. Conversely, when someone receives an offer for just the very thing they were thinking about it can feel like mind reading.
At its most basic level, using AI to predict customer behaviour is based on analysing what existing customers have done, learning from that and then using it to identify what new customers will do.
Rules-based AI can do this through manual programming – where experts analyse customer behaviour and then use incoming data from customer activity to trigger actions to engage the customer further.
To utilise Machine Learning to predict customer behaviour a computer programme is provided with customer engagement data that includes an outcome and then trained to identify the patterns of data that result in that outcome. The programme creates models that are able to predict the outcome when provided with new customer engagement data. Effectively, the machine is learning for itself, hence the name.
So, when a customer is shopping online, the Machine Learning model can use the steps they have taken to predict the most likely next steps – what they are most likely to look at and most likely to buy. To be able to make these predictions with a good level of accuracy requires a lot of data to train and test the model. Fortunately, online retail and marketing tends to have a lot of good quality data that has been collected previously.
The obvious application, and the one we have already referred to, is predicting what a customer is likely to be interested in or buy next. This can then be used to target the customer with products or services at just the right time.
Utilising wider data from social media, location, nationality and the economy can allow models to be built that identify new potential customers. Essentially both these applications can help predict future demand so that supply can be managed effectively.
Similar models can also be applied in marketing to understand what content and engagement approach will work best with potential customers. Then using that knowledge to target them intelligently and effectively. This is now even being extended to automatically create content for specific customer groups that is predicted to be well received.
At the other end of the spectrum customer behaviour predictions are also being applied in areas where customers pay for and receive an ongoing service. Specifically, to predict when a customer is likely to leave, or terminate the services. Also known as customer churn.
By using activity data to predict when a customer is likely to leave, then an organisation can look to intervene early in order to try and keep them as a customer.
The benefits of using AI to predict customer behaviour can help increase customer spending, loyalty and retention and attract new customers. Hence the interest in the online retail and marketing sectors but what about other sectors?