Chatbot Horror: How Conversational Apps Ruin Customer Experience
Bots can be so bad, they scare customers away
We’ve all got used to some sort of automation in nearly every aspect of our lives: from learning weather forecasts to shopping to making travel reservations — Conversational AI apps are here to make our lives easier. And this trend is on the rise — 70% of white-collar workers will interact with conversational platforms on a daily basis, according to Gartner.
However, some customers do not feel particularly optimistic about chatbots: research from PwC found that 71 % of U.S. consumers would rather interact with a human than a chatbot or some other automated process. What’s more, according to Forrester analyst Ian Jacobs, 54 % of U.S. online consumers believe that interacting with a chatbot will have a negative impact on their quality of life.
Why do such things happen? While we all know what chatbots are, this industry is still rather young. Sometimes businesses and people behind chatbots make rather blunt mistakes, that others can learn from. And here’s how chatbots scare customers away.
“Stupid” bot
This is a simple one. According to research, people get more frustrated when chatbots can’t answer their questions than when human agents in a similar situation. So, do invest in a good chatbot design and be patient, otherwise, you’ll just drive people away. Just like Poncho weather bot, which peaked in 2016. The screenshot below has made it to the hall of chatbot shame.
Also, human conversations are never linear and you can’t expect your customer-chatbot interactions to always run smoothly. Users may want to change the subject or ask a clarifying question. A chatbot is truly conversational only when it can properly process abbreviations, misspellings, jargon, colloquialisms, and subtleties.
Pretender bot
According to research, 75 % of people want to know when they are talking to a chatbot, with nearly half of them called a chatbot pretending to be human “disturbing.” Also, more than half of the people surveyed say they can tell the difference, and they don’t like to be fooled. Do you remember how much media backlash was caused by Google Duplex’s presentation when a bot called to a hair salon to make an appointment with an unknowing stylist? The company promised this feature would have a built-in disclosure, so learn from the best and don’t try too hard to create a human impression. By the way, in California, it is illegal for bots interacting with consumers to pretend they’re human if they sell goods, services, or try to ‘influence a vote in an election.’
Persistent bot
Sometimes customers want to be left alone when they are done with your services — and a bot needs to understand that. Becky Beach, eCommerce owner and Business Blogger, who uses Facebook chatbots to interact with customers, shared a story of her chatbot mishap. Last year, a customer emailed her about a chatbot “ stalking him… He said that the chatbot kept messaging him even when he told it to stop.” So Becky wrote a cease feature in the chatbot code to prevent anything like this from happening again.
But it’s not just small business clients who have to deal with pushy chatbots. In 2016, CNN news bot wouldn’t understand the simple unsubscribe command — the bot properly processed “unsubscribe” only when it was used with no other words in the sentence. So, make sure your bot understands when to let go of a customer.
Overly wordy bot
Just like with pushy chatbots, customers hate it when a chatbot sends them large blocks of text. They are difficult to make sense of, so such messages simply discourage, and/or overwhelm users and make them want to leave without ever coming back. So, make sure to break up the information into digestible pieces, if you want to keep your users.
Negligent bot
Once your bot has prompted a user to act, it has to promptly address that action. If the bot provides instruction, but never reacts to the action that follows, it demonstrates its unreliability. When customers see that bot is online, they expect a response from it. So, make sure it lives up to their expectations.
Limited functionality bot
Unlike the shopping websites which usually have up to hundreds of various items on display, bots can offer a rather limited number at a time. For a smooth shopping experience, bot designers need to get creative and showcase the offering while trying not to confuse customers. Giving your customer just a curated collection comprising a dozen items is just as frustrating as pouring your whole online catalog into a chatbot. Also, when people make orders, they tend to change their minds on the go and your bot needs to be prepared for that.
“Dead” bot
Just like everything in life, bot projects end. Some bots simply have an expiration date — the promo ones, for example. Whatever the reason for your bot to stop working, your customers will still try to use it. If your app exists online and people are able to find it, you should provide some sort of information to them. If you don’t expect your bot to be running again, your customers have the right to know it. God forbid there’d be any more installments of the 50 Shades franchise, so Christian Grey Chatbot is not likely to respond anytime soon.
It’s rather frustrating when a once heavily promoted chatbot, stops functioning without any explanation. Einstein bot, which had been designed to promote the National Geographic’s show Genius was considered one of the most innovative conversational apps. But now it simply ignores messages.
On the contrary, there’s Eugene Goostman, a bot that has been around since 2001 and “that some regard as having beaten the Turing test” is still up and running.
Humans want a human attitude — even form a bot. When we feel that the conversation is going nowhere, or that our questions have not been answered, we get frustrated. Make sure your chatbot APIs are on point, otherwise you’re not very likely to see people coming back to you.
