In recent years the customer experience landscape has seen the emergence of chatbots, virtual digital assistants, and AI. By automating repetitive tasks these tools have saved costs, allowing humans to focus on more complex issues. The long term impact of AI and machine learning applications, however, is potentially tremendous and far reaching.
Machine learning refers to the ability of information systems or computer programs to learn and improve from experience, without being programmed. Essentially, the machine interprets existing data using algorithms allowing the computer program or information system to find hidden insights without being explicitly programed where to look.
“At the moment”, says Peter Doherty, principal solutions consultant from ServiceNow, “Machine learning is very good at correlating vast amounts of structured data to make fairly simple decisions, but it needs to be trained. As such, current chatbots need to be trained with structured or pre-determined conversations.”
However, Doherty believes that machine learning has tremendous potential, he predicts, “Machine learning will take a major step forward when it can interpret unstructured data and improve the speed it takes to interpret and apply learnings from the data collected. At the moment it takes vast amounts of data to have learning applied to it, and it takes much more time to collect this data than it does to perform learning against it.”.
In terms of AI and machine learning’s impact on the customer experience, there are three key areas to pay attention to:
- Automation
- Analytics
- Personalisation
Automating repetitive and laborious tasks
Organisations find it a challenge to keep up with changing customer expectations. According to the 2017 Customer Service Benchmark report 41% of companies do not respond to customer service emails and only 11% answer questions in full on first reply.
Instant resolution of a problem or a question can be very powerful in terms of creating a positive customer experience. Machine learning can automate simple interactions allowing customers to resolve issues and questions in real-time regardless of when or where they are located. Faster resolution and response to customers means a better experience, yet customer expectations are constantly changing.
On one hand it can provide automated responses to the simplest enquiries or problems on the other it can route the query to the most relevant agent or team, providing them with up to date data about the customer and their history with the organisation.
The automation of simple and repetitive tasks frees up time for employees to focus on more strategic or value added work. Doherty comments, “Machine learning can automate plenty of internal processes andtasks. This may be as simple as speeding up the resolution of customer issues by getting them to the right people, at the right time, at the right priority. We’re working with a number of clients to correctly categorise, prioritise and assign customer interactions to the right person or group automatically, saving time, improving the customer experience and leading to greater productivity for businesses.”
Predictive analytics
Machine learning will enable analytics to predict future trends and customer behaviour, allowing issues to be rectified before they become a problem. A customer who is about to abandoned the shopping cart on a website can be connected to a live agent to have their issue resolved. Agents can be given the necessary information and insights, based on the analysis of customer data and past behavior, to assist customers in resolving support issues or making purchasing decisions.
“Currently we’re at a stage with chatbots where the technology has to pass over to a human at the end of their predefined conversation possibilities. The next phase is where the technology still stays connected after the handover to a human customer service representative and learns in real time how the successful outcome is delivered”, says Doherty.
Companies like Amazon already provide promotions and recommend new products based on past purchases and interactions with their websites. Doherty observes, “As machine learning capabilities grow, we will see more examples of pre-emptive or suggested delivery of products and services based on better understanding and analysis of past behaviour.’
Personalisation
“Consumers have become both critics and creators, demanding a more personalised service and expected to be given the opportunity to shape the products and services they consume.” From research by Deloitte Consulting
Karl Worth and Katie Sweet in their book One-to-One Personalization in the Age of Machine highlight how machine-learning and predictive analytics is critical in offering a scalable and personalised experience to every customer. At each and every interaction, based on analyzing past data and behavior, a customer can be presented with the most relevant content or experience as possible.
Personalisation is the holy grail of the customer experience. According to a range of opinion and research, personalisation, over any other factor has the most significant impact on advocacy and loyalty.
The need for quality unbiased data
The future success of machine learning is dependent on data – having plenty of accurate and relevant data. According to recent research from ServiceNow CIOs cite data quality (51%) and outdated processes (48%) as substantial barriers to the broader adoption of machine learning. If the data collected is inaccurate or biased, the machine learning algorithms will extract the wrong interpretation.
Rather than improving the customer experience, poor data will lead to catastrophic errors and bad experiences for customers. Not only is accuracy important but speed is also a major factor. It’s the ability to interpret and learn from the data as quickly as possible that will make or break the success of a machine learning application.
