Interviews, insight & analysis on Ecommerce

More “Chat” than “Bot” – making AI chatbots work for your online business

By John Walls, Regional VP, Customer Success, EMEA at Zendesk

Thanks to increasing technological advancements and the boom in ecommerce, it is almost impossible to shop online without coming across a chatbot. That’s not a bad thing – a well implemented chatbot is a valuable resource for your online presence, and can direct customers through the customer journey.

However, there are some pitfalls to avoid too – a chatbot that leads customers in circles, or doesn’t provide quick resolution, is possibly worse than no chatbot at all. When 50% of customers worldwide say they’d leave a business over just one bad customer experience, you’ve got to get it right first time. That places added pressure on having effective chatbots – but it’s a challenge that I think they’re ready to rise to.

Embrace AI

Historically, chatbots have had a bad reputation. In fact, for many, the word ‘chatbot’ conjures up an image of a prescriptive, ‘click your answer’ conversation that is essentially a flowchart in chat form. That bad reputation came from early iterations which used limited language processing and offered only a few pre-set responses that can frustrate more than assist. 

This just isn’t how chat works. You wouldn’t message a friend to say “How are you? Please select an answer from below: Very good, OK, or Fine.” So why would your customers want to talk to your business this way?

Throughout the past year, there has been a dramatic increase in customers using conversational channels to communicate with businesses – in fact, Zendesk research shows that one in 10 Millennial/Gen Z customers used an AI chatbot for the first time this year. This goes to show that consumers are now more willing to communicate with businesses through  bots. How should organisations take advantage of this? 

The first step here is to really assess what your chatbots are there for and the role they play in the customer journey. In the broadest sense, a good chatbot serves two functions: it makes it easier for your customers to get answers, 24 hours a day, 7 days a week; and it saves time for your agents who can focus on more complex and high value requests, rather than basic or repetitive questions. 

To do this, context and connection are the keys to success. First, can your chatbot understand your customers? 

Thanks to Natural Language Processing (NLP), bots can now understand customers as they would usually speak, picking out key words and phrases to create an accurate assessment of what they are being asked. For example, rather than a customer having to ask ‘What are your opening hours on Bank Holiday Monday?’, they could message a chatbot ‘I’m planning to pop by around 5, will you still be open?’. NLP will understand that this is not a request for general opening hours, but a request for same-day information and can respond to let the customer know of any changes to usual times.

In addition, today’s bots are advanced enough to answer questions that were once directed at the customer service team; suggest complementary or upsell purchases for sales; and collate feedback for the marketing department. For this to be effective for your business, you need to train your AI. Using your existing customer chat logs, you can set your bot to work using machine learning to find out the types of enquiries customers typically make, and ensure that the bot has access to all the information it is likely to need – from FAQs on opening hours, to APIs with your distributors for enquiries on order shipments.

Connecting the dots

The second half of the equation I mentioned was connection. The best customer service teams know how to work together to deliver a seamless and effortless service. And today, this includes the bot. But having a chatbot can’t be an excuse for a fragmented and difficult experience where the customer has to keep repeating themselves. Too often, this is where the chatbot experience fails. 

If you’re deploying a chatbot, be transparent with your customers about the fact they are engaging with a bot. Depending on your brand, give it a persona and even a name! And, importantly, make it clear to customers that they have a choice if they aren’t getting the answers they need. The most effective bots are trained to understand when a request is complex enough to require human intervention and can seamlessly pass the request on to the best-placed team to support – with both information and context. We are even getting to the point where sentiment analysis can help the chatbot flag customers who are particularly frustrated as part of the escalation process. 

Data collected by the bot at the start of the conversation – such as username or order number – can be passed to the agent picking up the request and actually enable a more personalised and unified customer experience. Connecting the dots on all of the customer information is invaluable for both the immediate customer request and resolution and to help understand wider trends that can help improve your business strategy. 

Make chatbots a strategic imperative

Across industries and businesses of all sizes, we’ve reached a digital tipping point, with 75% of company leaders stating that the pandemic accelerated digital transformation. Every business now has some sort of online presence, and your customers’ last interaction with another business sets the bar for their next digital experience. Whether they spoke to a person or a bot, the feeling they took from the interaction will remain front of mind and set expectations for the level of service they expect next time.

Intelligent chatbots can do more than just serve a customer’s immediate need and represent the start of a personalised experience. A well set-up chatbot can learn to recognise repeat customers, welcoming them back and picking up their conversation from where it last left off. If deployed correctly, chatbots can become an extension of your team, keeping the same name, speaking in the same tone of voice, and becoming a part of your brand identity, giving reassurance to your customers that help is always on hand, like a real person would in a bricks-and-mortar business.


More posts from ->

Ecommerce Age

How do we solve the issues with lockers?

We’ve all seen the banks of parcel lockers that we’ve all seen outside supermarkets and train stations. They are clearly a sensible idea, and one that I’ve been more than willing to use, but I’m very rarely given the option to do so when ordering goods online. Part of the problem, according to Gary Winter, VP of global strategic initiatives for parcel lockers at Quadient, is that they are invariably linked to a single delivery firm – such as Amazon or InPost – and this limits traction.


Related articles