How Machine Learning is Improving Real-Time Customer Experience

As a marketer it’s impossible to avoid discussions about how machine learning (ML) and artificial intelligence (AI) are improving the real-time customer experience (CX). And while we might think we know what that sentence – and all the terms in it – means, there’s a lot of confusion about what exactly does improve customer experience and how exactly it does it. To help navigate what is certainly the most important development for businesses in several years, Modern Marketing Today sat down with Dr. Chris Nolan, Chief Technology Officer at Alterian who has a Ph.D. in Neural Networks and Genetic Algorithms.

Here’s what Chris had to share with Modern Marketing Today about how machine learning is improving real-time customer experience by helping marketers put what they know about customer interactions to work.

Modern Marketing Today (MMT): Let’s start with the basics. What’s the difference between artificial intelligence and machine learning? Why does this matter to CX Professionals?

Chris Nolan (CN): The two terms are used so frequently together that the fact that they’re two distinct technologies is often missed. Machine Learning (ML) is a subset of the field of Artificial Intelligence (AI). AI is by far the more popular term, but it is still a future-focused technology. In other words, with notable exceptions we’re a long way from capitalizing on AI’s core capability which is in applying learnings from data and facilitating machines to become genuinely creative.

What CX Professionals are actually using is ML. That is, they are using algorithms that ‘learn’, i.e. improve over time by using real world behavior as feedback into the system. This means information fed by data collected from customers’ interactions with the company website, or social media pages, or email newsletters, can be used to continually improve the customer experience. The real advantage for marketers with ML is that a customer’s journey will get continually more relevant & engaging resulting in a greater lifetime value without human involvement. With the right application of ML tools, the improvements in the customer experience apply to real time interactions in the customer’s journey as well as all the other steps.

MMT: How can ML improve the real-time customer experience?

CN: There are so many ways in which ML can improve real-time customer experience. From streamlining support ticket flow, to improving a customer’s call center experience, and of course, personalizing a customer’s experience on a brand’s website or social platform, to name just a few opportunities.

While CX Professionals can attempt to take on these types of activities without ML, ML helps manage the scale and complexity of journey orchestration enabling marketers to personalize the customer experience in real-time for hundreds of thousands of interactions across hundreds of product lines or services and, of course, a machine can continue to process at scale and speed 24 hours a day 7 days a week without fatigue or complaint.

MMT: What does an organization need in order to take advantage of machine learning?

CN: Some people want machine learning projects to be highly complex. They want to rip and replace entire technology infrastructures in order to able to rebuild using proprietary tools. But to me that’s the same as trying to boil the ocean or being complex for the sake of being complex. Today there are plug and play machine learning tools that can be added to existing customer experience systems so all that is really needed is a desire to get started.

Where I’ve seen the greatest success is when organizations start small with a project that has buy-in from the relevant departments and at least one executive, and that has a defined set of goals and tangible business metrics or KPIs. The guiding philosophy behind what project to choose is obviously determined by the organization’s goals, but I recommend looking for a use case where a real difference can be made for the customer by putting them back at the center of customer experience.

One way to get started is to use ML to update the layout of the home page of the website based on what is known about the customer such as purchasing history, geo-dem info, but also on their real time behavior. Just as retail stores,improved customer experience and revenue by re-imagining the physical space, so to can the web team re-imagine the web site to draw the customer in. And with webstores there is no need for a ‘one size fits all’, every customer can have a different, unique and customized web experience.

Over time, the organization will aggregate yet more data that can be put to work about their customers to continue the process of improving real-time customer experience. Organizations can gather both ‘fast data’ – which is information about behavior in the moment – and ‘slow data’ – which is who the customer is, and how they behave over time. The ability to leverage both sets of data about the customer allows for further refinement of real-time customer experience and customization of their journey. While we’re still quite a way from evolving an organization’s website from ‘The’ website to ‘My’ website, this is what ML is designed to do.

MMT: Are there any pitfalls? Can they be avoided?

CN: As with any new application of technology there are definitely pitfalls and lessons that can be learned. One easy way to avoid most of them is to work with a trusted technology partner so you can leverage the wealth of their knowledge and expertise over hundreds, if not, thousands, of engagements and sidestep the traps entirely.

There are general project errors such as failing to coordinate departments or failing to get executive buy-in, but there are also specific errors that come with integrating ML into a customer experience system. For example, while algorithms can detect patterns within the data the algorithms have no social context or knowledge within which to place that information. So, data that might seem innocuous within the context of the machine – such as information about race, gender, or financial information – could be applied in ways that have negative impact on a brand or are downright illegal. Equally, ML will optimize parts of the organization without thinking about how it will affect other parts. Say ML increases calls to the call center by 30 percent but staffing levels in the call center weren’t increased in anticipation of a successful campaign and so wait time increases. What is a customer experience success in one part of the journey, is overall a customer experience failure and could have significant negative impact on lifetime customer value.

In the end while ML tools are powerful, they still need humans to oversee them to ensure that they stay within bounds, are inline with brand values and identity, and that all teams affected by the algorithm are prepared for the impact.

MMT: Do you have any final insights to share?

CN: We’re at the beginning of the golden age for improving the real-time customer experience of consumers. All the tools are in place – compute and storage have never been cheaper—and this puts the power ML within reach for nearly every organization. By starting small, building cross-functional teams, and deciding on success metrics you’ll be 90 percent of the way to a successful project. But always remember that algorithms are just machines and that humans are still the most important element since only they can watch for bias and ensure that the outputs from all that data are in alignment with the brand or company values.

Written by Jenna Sindle - Modern Marketing Today

This site uses cookies to enhance your experience. By continuing to the site you accept their use.

Accept Privacy Policy