2.5 million people use Slack on a daily basis. As usage grows, we are only going to see more bots arrive in the Slack app Store and greater admin control of Slack integrations. According to the Slack Platform Roadmap for Developers, it’s a key priority of Slack to make submission to the Slack App Store easier and faster for companies. While more and more bots are being built and integrated into Slack teams, there is still a lot of opportunity to improve how they are introduced to their new Slack team members.
Considering the increasing contribution to GitHub repos such as Botkit by Howdy, building a bot for Slack is not too difficult. This means top developers and companies building bots for Slack must compete on bot discovery and conversational copy (how the bot is scripted to talk). Your bot has to be someone people and teams warm up to and learn how to use.
Slack team owners and admins have the power. Their decision to integrate a Slack app sets off a series of events. Vijay Sundaram, who wrote this great article, details the three ways you can introduce a new bot’s capabilities to your team:
- Spam Them: This is where you program your bot to automatically send a direct message to all Slack team members once it’s been added to your team.
- Show them: This is where the bot’s first user exposes other team members to it naturally through usage.
- Tell them: This is when your bot shows it has joined a #channel through automatically generated activity or a direct introduction from the bot itself.
These bot discovery opportunities will change and grow. According to the Slack Platform Roadmap for Developer, bots that have been added to channels will eventually be listed.
We’re at a cool moment in the world of design when words rule all, arguably called “conversation UI”, so your bot’s first words matter a whole lot. Use the following guidelines to carefully consider the words you choose for the first moment of interaction between your bot and users.
- Your first message should always define context. Your bot should explain who turned it on, what it does, and why it is messaging the user.
- Make sure to provide tutorials. Your bot’s command lines and functionality are not seared into a user’s brain yet, and users might not know your bot can even process natural language requests. There will be an initial period of bot-to-user education taking place. Make sure your early messages contain links or instructions on how to actually use your bot.
- Your bot will need to earn a user’s attention. Unless your bot is adopted by other team members or made mandatory to use, there’s a strong possibility users won’t engage with it the first time around. Each message your bot sends should build credibility and offer direct value. Make sure you clearly explain how your bot makes life easier. Experiment with short-video testimonials of existing users that can be played directly within Slack.
- For a new and untested bot, the odds of passing the Turing test aren’t great, so be honest about what your bot doesn’t understand or can’t do.
- In the beginning, your bot will need to control the conversation, like a dancer who takes lead of their partner. Your bot should give a very specific call-to-action with each message if it is requesting information from the user. Eventually, Slack will introduce interactive messages (adding buttons inside of a message) so conversation UI designers will have additional latitude for creativity.
I hope you found this helpful and will be inspired to share your own thoughts on bots. If you want to talk about conversational copy or bot discovery, or if you want to join a funded startup building a personal financial assistant bot, message me: +1.407.310.8166.
Keith Armstrong is cofounder and COO of Abe.ai.