How can AI support the customer experience?

In this video we focus on how customers interact with a brand and how AI can be used to capture and measure this. Typically, customers interact with brands in multiple ways and through multiple channels. An example of this is shopping on a website for clothes. A potential customer Jane, may already know the brand from physical stores, social media or TV. After seeing an ad, a user may search for the brand through a search engine, and visit the website. When browsing they may find something they want to buy. If they can’t find their size in stock, they may engage in live chat to see if it will be available again soon.

Once their question is resolved they may continue to browse or leave the site. When they are next on Facebook or reading an article on a news site, an ad may appear reminding them what they browsed for and nudging them to complete the purchase. All the actions a brand takes through advertising and marketing are aimed at supporting a user towards a purchase, and are part of the customer experience. In many cases AI is used to support this customer journey. As we can see from looking at this one simple journey there are multiple touch points and opportunities to capture data about the customer behaviour.

If we organize this experience into a simple customer journey map we can gather more details and discover the phases and stages a customer may go through on their path to purchase. This includes searching for clothing, browsing and selecting a specific item, and buying the item. We also describe some actions a person may take on that journey including where they first saw the brand, which channels they engaged through and how they interacted with the brand after buying. It also shows what a customer’s mindset might be when engaging with the brand at different stages. Keeping this customer journey in mind, let’s look at some AI techniques and how they can support the understanding and improvement of the customer experience.

Some of the most common and relevant techniques include text analysis, a Natural Language Processing method. Behavioural Analysis, Classification, Clustering and Predictive Analytics which are all techniques based on machine learning. Speech recognition which allows computers to understand the human voice and image and object recognition, a machine vision and machine learning method. Let’s take a look at each of these techniques in more detail and understand how they can be applied to the customer experience. Text analysis uses a method called Natural Language Processing (NLP) in order to understand the meaning of sentences. By breaking down the text, AI can understand what key words and phrases your customers use, group similar comments together and find common themes.

It can even tell you if there is a positive or negative association with what your customers are saying. Let’s consider how this relates to the customer experience. In this example, Jane goes to the website and chats via the live chat function, she also comments on social media and completes a review after buying. It is possible, and likely, that Jane will have used similar words and phrases across these different platforms. Is her message echoed by other users and reviewers for the same or similar products? Without AI, identifying these patterns could be a slow and labour-intensive process.

Text analysis helps to organise this vast amount of data into an easy-to-understand format, allowing you to summarise general trends so you can take effective action. This data can be especially helpful when thinking about how to easily address frequently asked questions. It can also help when developing your brands advertising messaging and site content. Let’s now look at classification and clustering. Classification is a supervised machine learning method that helps group customers into clear pre-defined buckets. For example, people who typically spend more than £100 and people who typically spend less than that. It can also help understand whether someone is likely to take an action such as purchasing. This is called propensity modelling.

Clustering is an unsupervised machine learning technique that lets you explore what customer groups might look like if you didn’t have any pre-defined ideas about how you want to split the group. Understanding customer groups and attributes based on information you collect including what they buy, how they behave on the website, what type of questions they ask and what they say about you on social media can help to provide a more personalised experience. Going back to our example of Jane. At the beginning of the journey we may not have that much information about her, but as she spends more time on our website, social media pages and talking to us we start to collect data and understand her in more detail.

However, Jane is one of many thousands of people that might be interacting with your brand online on any given day. It’s not always easy to give each user a unique experience by understanding their needs and interests. Using techniques such as classification and clustering we can find groups of people that behave in a specific way, for example they browse and buy items on their mobile and look for certain product. We can personalize the ads and website to focus on the right content and message. With the assistance of AI, there has been huge steps forward in helping customers feel unique.

Predictive analytics uses the AI method of machine learning to capture historical data and use this to predict what might happen going forward. Predictions can be based on any data points that are relevant for the business. For example, predictive analysis can help forecast growth metrics such as sales, website traffic or the number of people searching online for your brand. You can look at predictive analysis in combination with correlation; a technique which helps you understand which data points influence others and how much i.e., as the number of people visiting your website increases so do your sales. Correlation will help you understand what factors influenced this and to what extent.

When we look at our example of Jane, we may need to know how many potential customers are like her. In other words, want to buy a shirt, engage with your brand on social platforms and would buy online through the website. Knowing this can help you plan more effectively in terms of your marketing and communications focus, helping you decide which products to promote. There are two more AI techniques that brands have begun to apply to understanding the customer experience; speech recognition and image and object recognition. Speech recognition is the process that enables a computer to recognise and respond to the spoken word.

It converts the word into a language that the machine is able to understand, which in turn the machine converts into an action, based on those words. Although still in its infancy in relation to the overall customer experience, there are many possible uses for speech recognition. The most common use is in call centres. AI has the ability to transcribe a phone call in real-time. It also allows for automated questions and answers. For example, wait times are made shorter by automatically asking for an account number or the reason for calling to help speed up the process, leading to greater customer satisfaction

Although not common place, there are other uses for speech recognition which are being experimented with. For example, cars that have voice-activated air conditioning or Smart TVs that can change the channel on demand. As technolgy evolves, we see adoption in the customer experiences growing rapidly. In recent surveys by eMarketer in the UK smart speaker users increased from 10 million in 2018 to almost 16 million in 2020. In the USA, voice assistant users have grown from 104 million in 2018 to 128 million in 2020.

Finally, let’s discuss Image and Object Recognition. Image recognition is a machine vision method. This field is a rapidly growing area in AI focused on identifying key objects and details in images and videos. It can help identify if an image contains a person or object such as a glass or book. It can also recognise colours or what is in the foreground versus the background. This technique has been used significantly in medical analysis and security. However, brands are increasingly using these images and videos to promote products and services. These are shared on their website, social media and advertising campaigns. This content can be expensive to create involving lengthy photo and video shoots. It is important to understand what works and why.

With the use of image recognition, it is much easier to identify common themes across images.

It can help answer questions such as: do users respond better to content with people in it? And which colours work best? There are many elements of an image that would be hard to break down manually. AI supports this process by providing data that will help you make creative decisions in producing content that is more likely to increase user engagement.

Through Jane’s simple shirt buying journey we have seen a number of places where AI can help us get to know Jane better and improve her customer experience by capturing and analysing information about all her interactions. When we present this in a single view, it is amazing to see that AI can help us at every point in Jane’s journey from discovery through to purchase and beyond.