Why chatbots fail and how retailers can make them work

Most chatbots fail because companies either do not clearly define their purpose, or they set goals that are far too ambitious, according to a recent study by Forrester.

The analyst group’s research, ‘Chatbots are Transforming Marketing’, suggests some bots are being launched prematurely, and report author Xiaofeng Wang argues the success rate across all industries is currently very low.

She says a more focused approach to chatbots is required to meet customer expectations, and today’s consumers may be sceptical because the technology often overcomplicates interactions – the exact opposite of what it was originally developed to do.

These artificial intelligence (AI) technology creations are seen as a way of relieving the burden on customer support centres, providing automated responses to customer queries or helping execute targeted marketing. But, at present, success stories are rare.

Following the early stages of chatbot usage and AI in retail, there have been some illuminating findings for those deploying the technology.

On the wider benefits of bots, Siobhan Gehin, managing director of Kurt Salmon, part of Accenture Strategy, notes: “The development of AI is likely to accelerate this year and become part of the next generation of digital services that retailers will adopt to build closer relationships with their customers.

“Through text analytics, machine learning, robotics, virtual agents and image recognition, retailers can make smarter recommendations, locate products in-store, and capture unprompted and authentic feedback."

Meanwhile, Wang is confident bot use in wider business has significant potential. She says: “The fact that most chatbots disappoint today doesn’t mean that marketers should give up and stop experimenting.

“If marketers can clearly define the purpose and scope of a chatbot, thoroughly evaluate the benefits it can bring, and plan and execute it well, it can deliver business and customer value.”

Where is retail at with chatbots?

A week hardly passes without news of a new chatbot being used in the retail or hospitality space, which is indicative of the test and learn phase the industry is in as it continues its digitisation path.

Pizza Express launched a bot in February, allowing customers to book a table at their local restaurant by conversing with AI technology through the popular social media channel, Facebook Messenger.

Brands such as Burberry, Estee Lauder and TK Maxx have all dabbled with Facebook Messenger chatbots, too, using special events such as London Fashion Week and Christmas marketing campaigns to test the tech. They have had varying degrees of success.

Stuart Ramage, eCommerce director at Dixons Carphone, says the group’s Currys PC World arm is in the early stages of working with Microsoft to develop chatbots that will help customers better understand its “a complicated product set” online.

“We want to help our customers navigate [our products] better online and then get the added value of our expert colleagues in-store, who can answer specific product questions and know how products work best together,” explains Ramage.

“We turned to Microsoft to utilise its technical expertise, software innovation and unique bot proposition to develop a bot tailored to the exact needs of our customers.”

However, it is Very.co.uk that claims to have stolen a march on the industry with its use of chatbots. Deployed within its iOS app in November 2016, the bots help support the 'Very Assistant' mobile app function which answers customer questions in the style of an instant messaging service.

Parent company Shop Direct calls this platform “conversational user interface” technology. It has been set up specifically for customer service, and it is supported by long-term tech partner, IBM.

Jonathan Wall, the retailer’s group eCommerce director, says: “Our customers have taken to Very Assistant in a big way. Over 180,000 people have engaged with the chatbot since we launched it late last year and you only need to look at the comments next to our app reviews to know it’s a feature people love.”

In its current guise, the platform asks users if they need help before providing a sequence of questions and multiple action options. Later this year, Very.co.uk will apply AI to allow customers to ask service questions in their own words, with the chatbot offering up tailored responses.

Wall adds: “Our vision is to use an AI-powered chatbot for inspiration and sales too; giving each one of our customers their own personal shopper, who gets to know them better every time they interact. It’s a long-term goal, but one we think is achievable with the right investment and skills.”

Additional advantages

Kurt Salmon’s Gehin believes chatbots will increasingly be used as personal shopping services. She says they will create custom-style profiles, offer options for shoppers to select from, and learn from people’s choices, as well as help consumers purchase gifts for family and friends.

“Additionally, there are advantages of using bots in contact centres,” she notes.

“Employing advanced AI software to categorise customer emails, an automated response can be delivered with a human touch. As the system can be trained using millions of past messages it can process thousands of customer emails per day, so customer service representatives don’t have to spend hours categorising them manually.”

For Wall and Shop Direct, the chatbots are helping to generate “a huge amount of insight”. Shop Direct offers its customers credit facilities, and from its deployment of AI tech it knows, for instance, that “next due date” is the most requested query and payment queries are the most popular service among its mobile customers.

Commenting on why he supports the use of the technology in the wider retail industry, Wall remarks: “It makes the customer’s life easier because they get their answer quickly and we benefit from the efficiency and insight that an automated system gives us.”

Despite initial teething problems for businesses and only a small handful of meaningful user cases, there is clearly huge potential for chatbots. Forrester’s Wang is confident the tech will become even more sophisticated in the coming years.

“Today’s successful chatbots are driven more by keywords than by machine learning,” she notes. “In the future, advances in AI will move chatbots’ potential from ‘question and answer’ to ‘human-like’.”

In retail, that has the potential to reshape marketing strategy, mobilise customer care teams and help businesses move closer to personalised one-to-one shopping.