Chatbots, Please Don’t Make Me Think!

It was almost 16 years ago that Steve Krug’s instant classic hit bookshelves across the United States. “Don’t Make Me Think” quickly became the bible for web usability and in these 4 short words, it captured the essence of how websites should be designed.

It obviously doesn’t stop with websites. “Don’t make me think” applies to mobile apps, and even to our chatbots. Now that our lives are filled with many more distractions and way too many options, the words from this book are more timely than ever. Do you remember that adorable dog in the movie UP that would drop everything he’s doing if he sees a SQUIRREL!? (watch clip here) That’s who we need to design our chatbots for. Doug the Dog reminds me that our users have very short attention spans and we need to be remarkable to even have a shot.

If you don’t make it blatantly obvious and effortless to perform the first step and then next step in your chatbot, the user is going to get frustrated and leave. Paul Graham often shared this advice with startups while he was running Y Combinator.

“What actually kills startups isn’t competition, it’s the back button. If your product is something that users see and go “Meh,” that will kill you. It’s not Apple or Google going after you.” — Paul Graham, Y Combinator.

In websites, it’s the click of the back button.
In mobile apps, it’s the tap of the home button.
In chatbots, it’s the flick of the thumb to go to another conversation.

Software within Messaging apps is a blessing and a curse — because every single second, our chatbots are competing with the people that our users care about the most — friends, family and Tinder matches! SQUIRREL!

“What am I supposed to do now?” and “Huh?” is what will kill most chatbots.

It’s so easy for a user to leave your chatbot at any time. In Facebook Messenger for instance, there are probably a dozen other conversations that are far more interesting and pertinent than your chatbot. We need to recognize that reality and try to overcome that friction. But how?

What can you do to improve the experience of your Messenger chatbots?

1. Tell me when the chatbot is thinking with the typing:on indicator. Otherwise, I’ll assume that nothing’s happening and leave.
2. Guide me through your chatbot. Buttons and Quick Replies are awesome for accomplishing this. If I can’t figure out what to do, I’ll leave.
3. Meticulously think about the user experience for a user’s first 10–30 seconds on your chatbot. If I feel “Meh”, I’ll leave.
4. Reduce your chatbot’s response times as much as you can. The snappier the better, but anything more than 2–3 seconds is pushing your luck with the user’s patience.

User Retention with “The Aha Moment”

User acquisition for chatbots is a popular discussion in chatbot communities right now, but user retention is going to be the real pain in the neck. If you want the user to return to your chatbot, you must get them to, what Chamath calls, “the Aha moment” right away!

You must show immense value in the chatbot upfront, or the user is going to see a SQUIRREL and bounce.

Last month, while talking to BotList, I said “the most common replies from people while talking about your chatbot is still ‘huh’ and ‘what’s a chatbot?’” That’s why it’s imperative to take special care of our chatbot’s early adopters — these are the select few that have navigated across the internet and landed upon your chatbot.

My Messenger Chatbot Success Stories

1. Smokey — This puts real-time air quality data at your fingertips! Did you know that 6.5 million people die every year because of air pollution? While watching users interact with Smokey, I quickly realized that I can’t assume that my chatbot is intuitive, it must be glaringly obvious what the user’s next step should be. Smokey was a Featured Bot on BotList in July 2016 and served up hundreds of real-time air quality reports to users from 18 countries.

2. India or Pakistan Game — I decided that the only way users could talk to this chatbot is with buttons! No text. No emojis. Just tap buttons, and guess what? All engagement metrics are through the roof. The users spend several minutes talking to the chatbot (aka, playing the game), and an average conversation has 50+ messages. This UX decision won’t work for most chatbots, but if it makes sense for you, I encourage you to try it out.

tl;dr

1. Don’t make me think!
2. Talk to Smokey to get real-time air quality updates in your city and to see the tips of this blog post in action.