AI Chatbot for Website – 7 Common Mistakes to Avoid
Customer support chatbots are here to stay. But how do you integrate an easy-to-use chatbot into your customer service offering without making a complete mess of it and frustrating your users?
Although chatbots using artificial intelligence and natural language processing were invented back in 1966, it wasn’t until a few years ago (and the launch of Facebook Messenger) that the bot builder improved their craft and the chatbot platform really took off. You can find out more about the history of chatbots here and look up some simple best practices.
But it’s not all rosy when it comes to chatbots. I’m going to tell you a few things to avoid, so that your chatbot experience doesn’t turn into a nightmare for your users and your company.
Mistake #1: Not having a plan to let a human help
An easy trap to fall into is thinking chatbots can do it all, and enable fully automated customer experiences for your business. They may be great at answering simple questions quickly on your website, but people get very frustrated very quickly if the chatbot can’t find the answer. The solution is to have an easy way for customers to switch to a live agent.
- 23% of consumers still prefer to have a face-to-face interaction when the complexity of the issue increases, such as with payment disputes or complaints. (American Express)
Simplr did some research and looked at how the customer felt after a chat that had human interaction:
- 15%: Appreciated or engaged, like they really cared about me
- 30%: Understood or supported, like they knew exactly how to help me
- 30%: Felt BOTH Appreciated/engaged and understood/supported
On the flipside, only 1% of chatbot interactions (without humans) were ranked as “exceptional.”
Mistake #2: Not thinking first about the pre-sale and post-sale user experience
It’s easy to get carried away with the potential of your chatbot. Like all technology projects your business takes on, a chatbot initiative can soon suffer from mission creep where you lose sight of why you are implementing it in the first place. There are many shiny things to distract you, so before you get started, keep these questions in mind:
- What is the problem that the users are struggling with?
- How can our bot help them solve that problem?
- What might get in the way of that?
There is proof that bots are best used for very basic, simple requests. Returns, for example. However, there is a lot at stake with pre-sale customer service inquiries. Chances are, a bot can’t deliver a low-effort experience when a customer is on the verge of converting. This means missed revenue opportunities for your business!
Mistake #3: Trying to fake it
80% of people who have used real time chatbots report the experience as positive (Uberall). Users are clearly not afraid of using a chatbot. But companies are still attempting to dress up their chatbot as a human, hoping to fool the customer into thinking they are getting live chat support.
Do not do this.
Make sure your customer is informed they are talking to a bot and are not in a live chat. Explain the capabilities and limitations of your bot upfront. Do not over promise. This will set the customer expectations and cut down on frustration.
Mistake #4: Giving your AI chatbots way too much personality
Think about it, this might be the only chance you ever get to create a character and have it live in the real world, interacting with real humans! This awesome power is in your hands, but don’t over do it.
The beauty of the drag and drop chatbot platform lies in its ability to have real time, lifelike conversations and make the customer feel comfortable interacting with it. But don’t take it too far. Do not let the personality take over and get in the way of serving the customer.
Also think about tone. A lot of chatbot interactions are about mundane, boring things. These can be livened up with a light touch from a chatbot, but no one needs to hear jokes when they are getting their credit card balance. Think about the customer service experience your customer wants and treat them in the appropriate manner.
Mistake #5: Not using the best AI chatbot analytics
A lot of companies ignore chatbot analytics. They are completely in the dark when it comes to what works and what doesn’t.
Don’t make that mistake.
Your chatbot should be part of your marketing automation strategy. Use your chatbot software to run user tests and listen to user inputs from Day One. You’ll be able to see how users interact with the bot, where they tend to get stuck, which lines are well-received, which ones fall flat, and where your bot is wasting the customers’ time on your website.
Trust me, you will not get everything right from the get-go. But the analytics data and machine learning can help you optimize your bot over time to make it more engaging, helpful, and fun.
Your chatbot is not a one-off task, but a continuous project that needs constant attention to make it the best it can be.
Mistake #6: Not killing your chatbot when it’s not the answer for web chat
Chatbots are very good at answering frequently asked questions 24 7, consistently injecting your customer conversations with your brand, and taking the heat off overworked agents.
However, even though everyone want o build chatbots, they may not be the answer you are looking for.
Hopefully your chatbot fits perfectly into your customer service portfolio and provides people with an added channel for help. But if it doesn’t, you must know when to pull the plug.
If your chatbot is idle most of the time, doesn’t help with conversions, or it keeps providing the wrong answers, or it generates news articles because it is providing inappropriate responses that are offending your customers, maybe a chatbot is not right for you. It’s your call, but remember, there are multiple platforms for customer service out there.
Mistake #7: Believing the hype – a bot with a free plan and other third party promises
Some promise a super-easy drag and drop chatbot builder. Others offer a free trial. A chatbot builder is currently claiming their bots can field up to 80% of your customer enquiries. We are extremely skeptical about this. There is no way to verify what the enquiries were and how many customers left after a frustrating experience with the bot. We highly doubt a bot is solving 80% of customer questions.
Most of your customers have bigger questions. You might think a bot is answering the customer , but have you thought about the experience?
Chatbot development should also consider uses beyond just fielding customer enquiries. Chatbots can be used to generate leads, to drive to a landing page, or to augment your mobile app.