How To Build Great Bots
We’re in the middle of a revolution. An evolution. A mass extinction. Apps are dying, and chatbots are taking over.
Why? Three simple reasons:
- Messaging has surpassed social media in usage.
- Consumers don’t download new apps.
- And if chat is the new browser, bots are the new websites.
The beauty of bots is that you don’t have to download new apps. Bots live in your chat app, for which you already have an account. Also, you don’t have to learn a new UI, since you already know how to use your chat app.
We’re coming up on the one-year mark of launching our Bot Shop in April 2016. We’ve learned a lot, and we continue to be asked, “How do you build a successful chatbot?” We recently presented at SXSW alongside Omar Siddiqui, CEO of chatbot agency Sequel, and wanted to share four best practices that can be applied across all bot development.
1) Know your audience.
While this is true for any product, it’s especially true for bots. A bot on Kik should have a very different experience from something on Facebook Messenger or Slack. The audiences aren’t the same.
Kik’s users are predominantly teenagers, whereas Slack’s users are older professionals. Facebook’s users are typically older as well, but they’re using the social network mostly for fun.
CNN has bots on both Facebook and Kik, but the bots are very different. CNN has done an excellent job targeting the bots’ respective users, and tailoring each bot’s persona to meet the needs of those users. During the recent U.S. presidential election, CNN’s bots were very active.
On Kik, CNN created content that was approachable for our younger audience. Users could share election-themed stickers and GIFs, they could learn more about election processes like the electoral college, and they could receive live state-by-state updates on Election Day. The CNN bot rarely pushed users out to its website, and this type of content was well received. Through the campaign, CNN was able to growth its audience on Kik by 56 percent in three weeks. One million CNN stickers were shared. There were 4.5 million messages exchanged with the CNN bot on Election Day alone — and the election results were checked over one million times during the network’s election-day broadcast.
Microsoft’s Zo.ai launched on Kik last fall, and used its advanced AI to build a “teenager” bot. Zo launched on Kik specifically to target a certain demographic. Even before the soft launch, the bot had held 100,000 conversations. Zo also holds Microsoft’s longest continual chatbot conversation: 1,229 turns, lasting nine hours and 53 minutes.
The reason behind Zo’s success? She was given the personality of a teenager, and our users responded.
2) Start simple, start small.
A successful bot will have a clear function. It will do one thing really, really well.
Tickers, a Tic Tac Toe bot is a great example of a do-one-thing-well bot. The premise is obvious, and the bot works the way you would expect it to. It’s fun, very addictive, and was our third bot to hit one million users.
Dankland is another great example. It’s a Facebook Messenger bot that turns photos into memes. The experience is straightforward. It doesn’t try to be anything more than a meme generator. The results are often hilarious.
3) Build for conversations.
Bots that are too verbose don’t trend as well as bots that are succinct and to the point. Bots that are built for conversation leave room for the users to interact themselves.
For example,the Statsbot on Slack integrates with Mixpanel to produce reports that can be used within a Slack conversation. The bot doesn’t try to “thrill” users with an experience; it simply presents the facts.
Bots like this are very successful because they don’t get in the way of the conversation; rather, they facilitate it.
4) Be social.
Bots find success when they foster social interaction. Bots that work in groups grow at a faster pace than bots that only work in a one-on-one experience. When a user in a group @mentions a chatbot, it’s directly added into the group chat, and all members of that chat can see how that chatbot works.
In a social setting, well-designed, easy-to-use/intuitive and fun chatbots grow very quickly. We experienced this ourselves first-hand when we launched @roll, a simple bot that behaves a bit like a set of dice to help users choose between options. Ninety-five percent of the bot’s growth was through users sharing the bot with friends. The bot hit one million users in three months.
More involved and in-depth bots, like @murdermysterybot, behave as a game “master,” allowing groups of users to interact and play a game together, entirely online.
If you follow the best practices above, you’ll create a great experience that your audience will love, and you’ll be well set up for the chatbot revolution.
