The best reason to build a chatbot
A bot that can do this one thing well will be a huge success
Chatbots are the new Next Big Thing in product development — every startup and enterprise seems to feel they’re an absolute must-have and potentially the solution to all problems. CNN created one to serve up news, saying “we curate, you query.” Skyscanner released a bot to help users book flights. And there are plenty of startups trying to build your new personal assistant.
Like a lot of people who work in technology, I feel most of these bots are a feature chasing a problem. Asking a chatbot for news on a particular subject is no easier than googling for the same thing, and probably quite a bit worse. Unless you have few constraints on your budget or time, there’s not a lot of value to a bot finding you a flight. And since I’ve never had a personal assistant, it’s hard for me to imagine how to integrate one into my life, human or digital, without significant hand-holding.
Part of the reason chatbots have become so popular is because they’re now so possible. It’s hard to ignore the speed at which services like Api.ai, Wit.ai, and Smooch.io can enable you to build a conversational interface, and for a reasonably low cost at that. But to realize the true potential of the technology, what user problems should we be using chatbots to solve?
Use chatbots to reduce complexity for users
This may seem an obvious point, but it’s deceptively nuanced. Chatbots can be powerful, delightful tools for users if they remove complexity from interactions. I can see two clear ways chatbots can be the best channel for reducing user complexity:
Doing work for users in the background
Digit is a product that analyzes your checking account activity to determine how much you can comfortably save, which it proactively does for you by making small, regular transfers to its own savings account you can manage. Though the company for some reason released an iPhone app, most interactions with Digit happen via text message.
I’ve been using Digit for some months now, and I find it removes complexity from managing my finances in two incredible ways.
The first way is what it’s known for — it’s a terrific experience to see a savings account incrementally grow relative to your checking account cash flow without you having to do anything about it. Does it save a huge amount of money for me? No, but it saves for me here and there and manages the process well enough I’m never worried it’s going to overdraw my account. Best of all, it does this by itself, and just texts me to give me updates on how much it’s saved. It takes my involvement out of the saving process, and that’s delightful.
The second way Digit makes my life less complex is by allowing me to get my checking account balance with just a simple text of “checking.” This may not seem like a big thing, but it reduces the friction of going through the multi-step process of opening up my bank account app and logging in just enough. Even better, it texts me regularly unprompted with the account’s balance at a frequency I can set. This has subtly made me more aware on a day-to-day basis of my account balance and on a couple occasions has even caused me to identify duplicate charges and bank errors.
Digit does a lot of things for me I would normally have to do myself, and better yet, it does them in the background. A lot of the value chatbots can provide come from tasks they perform between chats.
Making complex user interfaces unnecessary
Another way chatbots can reduce complexity for users is by providing a natural language way to interact with complex data. Complex data usually requires a cluttered user interface to comprehend, but a chatbot can make it almost completely unnecessary.
Statsbot is a Slack integration that lets teams get regular reports from and query Google Analytics, Mixpanel, or Salesforce. It’s absolutely worth checking out their demo video to see it in action.
Statsbot is delightful because it takes something that is incredibly complex and makes it delightfully simple. Take Google Analytics — the ease of data collection on sites and apps has meant that we collect everything, and all of that data needs a way to be viewed. Just look at how many options the Google Analytics sidebar now gives you, and this is only the first screenful:
If you want to, say, get the number of users who visited your site last week broken down by their referral source, you’ve got to log in to GA, click through to the correct report, apply the correct filters…..zzzzzzz. By the time you get the right numbers, you may have already forgotten why you wanted them.
The same task can now be easily accomplished by asking Statsbot, “Show me users who came to the site last week by source.” Handling this kind of query is now extremely simple with recent advances in natural language processing. This means a CMO can ask Statsbot this question himself and it doesn’t need to go to the person actually responsible for watching the analytics and back.
Your existing interface might already be great
There might be great utility in building a chatbot to reduce complexity for your users if you can make something easier, but it might also be true that your site or app already does the job better. People booking flights usually want to scan a list of options and weigh several factors: a chatbot actually makes that harder. While CNN may think asking a chatbot to give me news on a particular topic is a delightful experience, I just find it annoying and inferior to simply browsing or searching their site.
Startup or enterprise, make sure the chatbot you’re contemplating actually makes your users’ lives easier, not more difficult.
Nicholas Alan Brown manages and builds products in Brooklyn, NY. Follow me on Twitter: @nicholaslanb
