AI hype fools a lot of the people a lot of the time. Many still claim AI is revolutionising customer service in every aspect — including powering chatbots to handle all manner of complex customer conversations with ease, replacing humans as they go.
But the bot backlash is already underway and customers are voting with their feet.
While AI offers many potential benefits, our typical advice is to stick to rules-based bots for now to help improve customer service performance and productivity then build out.
Why? Read on to find out.
What’s the deal with AI Chatbots?
AI chatbots and rule-based chatbots are two different beasts. AI chatbots are super-smart bots that can (or, try to) ‘understand’ what a customer is saying and try to respond with a unique answer tailored to those specific needs.
AI bots rely on tech such as natural language processing (NLP), voice to text translation capabilities, the availability of data lakes, visual recognition and so on.
For medium-sized businesses though, this AI tech is not usually reliable enough to warrant putting them at the heart of your customer service strategy.
AI chatbot fails can be epic even for big companies. For customers, AI chatbots are mostly just frustrating. The AI often misinterprets and misunderstands a customer’s needs, sucking up valuable time and good humour by trying to direct people towards something they’re really not asking for. Interestingly, Gartner predicts that 40% of the bot/virtual assistant applications launched in 2018 will have been abandoned by 2021.
While it is good to explore AI use cases — especially if you have your systems all speaking to each other and lots of data to work with — for the most part, we advise companies not to go full-on AI with customer-facing chatbots until they have their own customer service house in order and AI tech matures.
What about rule-based bots?
Many chatbots marketed as AI are actually rule-based applications that have been around for decades in one form or another. Maybe they employ a little bit of NLP or similar, but you’re basically buying a rule-based bot. That’s a good thing! There are tons of examples of great rule-based bots; from LEGO to Sephora.
Chatbot conversations are designed by mapping out the questions a customer might have, and determining what the appropriate response should be. These types of chatbots can’t answer any questions outside of the things you make rules for, but that’s fine!
While they cannot handle complex questions, rules can easily handle simple repetitive questions that often make up a big part of a customer service team’s workload. Automate these questions and you can speed your response times, gain massive efficiencies, and free your team to handle more value-adding conversations in pre-sales, for example.
Some claim that human customer service agents handling about three chats per hour can increase these to nine chats per hour when helped by a bot. That’s an incredible efficiency boost.
Plus, rule-based bots are quicker to design, deploy and optimise than AI chatbots. They integrate well with legacy systems. They are much faster at responding than a human — virtually instantaneous. And speed is, as we shared earlier, one of the four ‘must-haves’ for today’s shopper.
Everyone wins. But again, there is no silver bullet. Before you think that we’re being too conservative about AI and chatbots, bear in mind that one study from CGS shows that an incredible 86% of consumers prefer to interact with a human rep rather than a chatbot. That’s because most companies get their approach to chatbots wrong.
Here’s how you should do it:
1. Make sure you automate the right conversations
Accenture claim that 80% of chat sessions can realistically be resolved by a chatbot. But there’s a huge difference between ‘can' and ‘should’. Our research and experience show that only around 15% of an eCommerce company’s questions should be botted.
The questions you should not bot include pre-sales questions and last-moment purchase problems in the cart. There, you want a real live human as a live rep - according to our own data - can generate conversions anywhere from 20% to 50%.
Download our free white paper on AI and Chatbots to find out more on what to bot and what not to bot.
2. Get your handover to a human right!
When automating customer service conversions, it is essential to get the handover from bot to rep, right. That is notoriously difficult to do with AI bots. It can be much easier with rule-based chatbots.
Rule-based chatbots are built with a menu of operations — seen in carousels, buttons and so on — that they can perform. If your customer wants to do something that they cannot see in those options, they should see a super-big, bright, ‘Speak to a rep!’ button staring them in the face. Don’t hide it! Your goal is not to avoid questions, but to better help your customers. Sounds simple, but you’d be amazed how many people use bots badly.
Assuming that all the information from the bot conversation is pulled forward into the conversations with the live rep, from all channels — as is the case when we integrate chatbots into the ROBIN Conversation Console — then your handover will be successful.
The takeaway: Analyse customer service conversations before deciding what to automate and what not to automate. Deploy rule-based bots to handle repetitive questions to free reps for more value-adding tasks. Experiment with AI bots and build the business case before implementing