How AI is writing the future of CX

Share this content

Customer experience (CX) is one of the areas of business that has been most comprehensively changed by digitisation over the years. Since the start of the information age, we’ve moved from a primarily face-to-face customer experience, in which you got what was available, to a fast-paced, automated, omnichannel world, where recommendations and offers are (or should be) given to you based on your wants and preferences. Data analysis has changed the way businesses delight their customers.

That innovation is far from over. With the rise of AI, there are further opportunities for CX to change and grow, and provide an even more helpful, targeted service to customers and help organisations differentiate themselves from the competition.

The Experience Economy moves in

Consumers today have more choice than ever before, they’re more empowered and they have much higher expectations of the organisations with which they do business.  Welcome to the experience economy, where the biggest differentiator is no longer having the best product or the lowest price - it’s all about the customer experience.

Today's consumers compare all products and services with the best service they ever received – from any company or person. That’s why we’re seeing industry disrupters like Netflix, Uber and AirBnB forcing organisations in other industries to adopt new business models and digitally transform themselves to meet changing customer expectations.

To deliver the personalised and relevant experiences that today’s consumers have come to expect, brands need to process and analyse the vast amounts of data now available to them. And they must also action the insights in a timely and cost-effective way to impact customer experience whilst managing costs and resources.

It's no surprise, then, that organisations are increasingly looking to AI to help address this challenge and create more meaningful customer experiences.

Manage your customer journey with AI

Customer journey mapping and optimisation is a common challenge in today's omnichannel environment. Marketers need to be ready to guide customers along the journey with the right ‘next best action’ content, message or offer at any point. But traditional marketing automation and journey orchestration solutions are hitting a wall. There are many permutations of paths in any single customer journey and myriad different actions, content, messages or offers from which to select. The scale and complexity of this challenge is a perfect opportunity for the application of AI in the shape of ‘reinforcement learning’.

Reinforcement learning is a type of machine learning. At its core is the concept that the optimal behaviour or action is reinforced by a positive reward which allows the algorithm to learn on its own very quickly.

A good analogy would be to think of a toddler learning how to walk. When they first try to walk they might take a big step and fall over. Next time they adjust their step making it smaller to see if that's the secret to staying upright.  If that works, they try another step. And so the toddler continues learning and adjusting until they can walk confidently. Staying upright is the reward that helps the toddler reach their goal.

Reinforcement learning takes the manual guess work out of marketing and optimises the customer journey by continually looking for the next best action to deliver the best outcomes. It continuously tries different actions to work out which will deliver the best outcome in the long term.  In time, the system learns which action will deliver the best outcome – but every now and then it performs a random action just to be sure that the model is still fresh and up to date.

This is an exciting step forward for marketers searching for the holy grail of customer journey optimisation and ensuring next best action across every touchpoint.

Recommendation engines hit top gear

We’ve all had those online shopping experiences where we’re recommended an irrelevant item and realise that the company does not understand us as a customer at all. 

There are typically two common approaches to recommendations. The customer-centric approach looks for similarities in the behaviour or characteristics of customers and recommends products that other similar users have bought. A product-centric approach looks for products that are associated with each other. This is helpful when you don’t know anything about the customer and their characteristics other than that they’ve shown some interest in a particular product. In this example the product most associated with what the customer is interested in is recommended. 

Whichever approach is taken there are three common challenges with every recommendation engine.

  • Scale: As a recommendations database grows, the performance decreases and you start to get trade-off between performance and prediction accuracy.
  • Cold start: This challenge appears at early stages of a recommender system’s life cycle, or when a new visitor or product is added to the system. If you don’t have enough data about customers or items, the recommendations won't be that smart.
  • Sparsity: Many customers may only view only one product, or no products at all. Alternatively, many of the products may have never been viewed or rated. Both these scenarios are examples of sparse data sets and the recommendation engine struggles to find the common behaviour patterns or relevant associated products. The result is a long tail of products that is ignored by the recommendation system.

Artificial intelligence can give recommendations engines a real boost.  Machine learning algorithms that combine both customer-based and product-based recommendation algorithms can overcome scalability, cold-start and sparsity challenges to improve results for the retailer as well as delivering a much-improved customer experience.

Making analytical insight stronger

These days, customers interact with brands in many ways: social media and digital channels, phone, email, webchats, letters, surveys and reviews, and in branches and stores. Each of these interactions contains an opportunity to garner rich insights into customer intent and sentiment, what customers want, and how they feel about the products, services, and experiences being delivered.

Natural Language Processing combines machine learning, AI and linguistics and can automate and quantify customer feedback and insights across every customer interaction, regardless of channel.

Capturing insights at scale allows you to accurately quantify the size and shape of opportunities to improve the customer experience and prioritise which investments will deliver the most impact, both to customers and to the business. What’s more, this level of insight down to an individual customer, gives businesses the power to make more effective and profitable decisions about customer strategy and fuel 1:1 customer experiences.

Nationwide Building Society uses Natural Language Processing from SAS to take their industry-leading customer satisfaction to even greater heights. The solution analyses data from textual interactions to identify root causes of customer dissatisfaction and implement improvements. Nationwide has used the results to identify several concrete ways to make their service legendary. At the same time the discovery that 25 per cent of interactions could be moved to digital channels is supporting the company’s digital transformation journey. 

Each of these examples really highlight how artificial intelligence is taking customer experience to the next level by injecting machine intelligence directly into the customer interaction.

The future of customer experience will see more widespread adoption of AI across brands of all sizes, but it's the organisations that are embracing AI today that are already setting themselves apart from competitors and delivering winning customer experiences.

About Tiffany Carpenter

Replies (0)

Please login or register to join the discussion.

There are currently no replies, be the first to post a reply.

Related content