Emma Herrod takes a look at chatbots and machine learning and how automation is changing shoppers’ interactions with retail contact centres.
RETAILERS REALISED early on in the infancy of online shopping that customers sometimes need more information about a product, brand, delivery or returns than what is shown on the site. FAQs and delivery and returns pages were introduced in order to help customers to answer the questions themselves rather than make a telephone call to the retailer’s contact centre. Simple questions, easily answered, left shoppers happy with the level to self-serve.
Self-service technologies evolved, along with ecommerce and customer expectations of online shopping as it moved from handling product enquires to queries such as “where’s my parcel”. The transparency of parcel tracking became the norm, again taking pressure away from an increasing number of contact channels.
“The days of ‘do not reply’ emails and texts, and ‘we can only deal with this over the phone’ statements are over,” believes Stephen Ball, Senior VP and General Manager in Europe & Africa at Aspect Software. His comments come of the back of a survey conducted by the company which discovered that customer service across all industries can be so bad that two in five people find customer service so frustrating, they’d sooner visit the dentist than contact a brand.
The majority of Top500 retailers in the UK still take an average of more than a day to respond to a customer’s email, according to InternetRetailing’s Customer Performance Dimension report. Some retailers, particularly those within the fashion and consumer electronics sector, responded within 10 minutes, though.
When it comes to the telephone, 8% of the researchers’ phone calls were not answered at all, and on social, Twitter was responded to faster than Facebook. InternetRetailing’s researchers warn that “retailers than don‘t compete on customer service will certainly be left behind by consumers who now value convenience (and, by extension, service) above all”.
Shoppers can be empowered through self-service enhanced with natural language enabling interactive, intuitive text dialogues with an automated system. Natural language technology cuts the cost of customer interactions since text-based channels are capable of addressing more enquiries with personalised, conversational self-service. “Additionally,” says Aspect Software’s Marketing Programme Manager, Maddy Hubbard, “effective issue resolution on the channel of the customer’s choice increases customer satisfaction. Using applications like Interactive Text Response and Facebook Messenger powered by natural language will keep companies in sync with consumer communication preferences”.
This then frees up time and resource to handle more complicated enquiries and those which require a human touch as technology takes over the more repetitive responses or frequently asked questions. Ocado, for example, has developed a natural language algorithm to analyse the thousands of emails received by its contact centre on a daily basis.
Inbound emails are automatically analysed and prioritised so instead of being handled in the order that they are received, customer emails which need urgent attention are dealt with ahead of ‘thank yous’ or those giving general feedback.
The analysis is done by a machine learning-enhanced contact centre application which was developed in-house by Ocado’s technology division. The processing grunt of Google Compute Engine and Google Cloud Services and its Tensor Flow machine learning service are utilised, along with a natural language processor developed by Ocado which has been trained using millions of past messages from Ocado customers.
As Alex Voica, Technology Communications Manager at Ocado Technology explains, as a pureplay retailer the only points of contact for customers are the van drivers and the contact centre which customers can contact via social media, a UK landline or via email. Customers aren’t asked to put a specific heading on their emails so each day thousands of uncategorised emails arrive in a centralised inbox requiring a response. The contact centre handles 6,000 or more emails per day at its busiest such as during times of extreme weather, so it can become a stressful place to work and one in which greater efficiencies could be achieved.
Prioritising emails which require speedy resolution also improves the experience for customers.
A NATURAL RESPONSE
Automation will take customer service much further than analysing inbound emails. Machine learning can be applied to social media – something which Ocado could do but has chosen not to do yet. The rise of machines understanding natural language and the ability to train them with a retailer’s own data sets is how automation will really free up customer service agents’ time by allowing automatic, but natural ‘human-touch’ responses to customer enquiries.
US retailer North Face launched a digital shopping assistant last year to help online shoppers choose the right jacket. The recommendations served are based on a series of questions around where and how the jacket will be used and the shopper’s style and colour preferences. The shopper is able to respond to each question in a natural way, with the IBM-Watson powered assistant able to understand and learn from the ‘conversations’.
This natural-language intelligence is something which Shop Direct is working towards enabling for its Very.co.uk customers. Rather than being used as a sales tool though, the AI-driven system will work from within its MyVery app and will cover 13 different customer service scenarios.
Very is working with IBM Watson to replace its existing Very Assistant, which it launched as a first step in November 2016. Currently, customers have to follow a sequence of questions and tap the relevant response from multiple action options within the chat environment of its app rather than being able to ask questions in their own words. The customer’s answers enable the Very Assistant to serve up the information they are looking for. Self-service options include help to track an order, make a payment on their Very.co.uk account, confirm that recent payments have been processed, check payment dates and request a reminder of their account number.
When launched, in a couple of months’ time, The new Very Assistant will have been trained for 13 different use cases enabling shoppers to further self-serve by typing a query in their own words. As Jonathan Wall, Group Ecommerce Director, Shop Direct explains, the upgraded Very Assistant is being trained currently with past data which includes 250 ways in which customers have asked “where is my order”.
Since launching as a minimal viable product in 2014, every part of the Very app has been developed in line with feedback from customers via the App Store (where it is rated 5 stars) and the company’s own conversations with customers. In this way, the retailer discovered that customers not only want to chat in a natural way but would prefer to interact with a chatbot about orders and self-service matters rather than as a sales assistant.
Wall says that the usage of Very Assistant has been “phenomenal” in terms of unique usage and the volume of chats which have reached resolution. A “significant proportion” of Very’s mobile sales come through the app and the company has ambitions for it to rise to become the majority. Altogether, mobile accounted for 70% of Very’s business over the peak season.
With the app accounting for an increasing level of sales – and therefore customer enquiries – the aim is for the app to reduce the number of inbound calls to the contact centre and free up agents’ time to answer more complicated calls.
Shop Direct was the first UK-based retailer to offer a WhatsApp-style conversational user interface (CUI) platform for customer service and Wall believes that other retailers will follow suit.
AI has the ability to transform customer service and natural language whether typed or spoken – as Alexa, Siri et al have shown – offers a better experience for customers used to messenger apps. It offers choice to shoppers, shows that retailers understand their customers and the way they are interacting elsewhere while also freeing up customer service agents to handle calls where the human touch can increase customer loyalty. A win:win all round.