To Build a Successful Chatbot…
Start with these 5 questions.
With advancements in Artificial Intelligence and Natural Language Processing (and with companies like Facebook & Apple opening up their messaging platforms for developers), we’ve seen an explosion of new chatbots. Just three months after Facebook announced their platform for building bots, Facebook Messenger had more than 11,000 bots.
Now, a year after Facebook opened up their platform for bots, doubts are being raised about their impact.
“…despite the promise of a [chatbot] revolution in how we interact with services and companies online, progress has been utterly miserable — the vast majority of chatbots are gimmicky, pointless or just flat out broken.” Dave Lee, BBC Technology
“Facebook’s Messenger bots may not be having the impact the social network desired. Just yesterday, online retailer Everlane, one of the launch partners for the bot platform, announced it was ditching Messenger for customer notifications and returning to email.” Sarah Perez, TechCrunch
So, what happened?
Like any technology going mainstream, a lot of companies and eager developers jumped onto the chatbot bandwagon. But, not everyone got it right.
So then, how do you do it right? By asking the right questions.
After spending months building a messaging platform, interacting with chatbots and understanding the conversational UI paradigm, I took a quick stab at creating a list of 5 questions you should ask before embarking on the road to building a chatbot.
So here we go. Let’s start with the basics.
Question 1
The first question to ask yourself is NOT “Should I build a chatbot?” but rather, “What is the outcome I want for users of my product or service?”
Don’t just build a chair, build a chair so people can rest on it.
Conversational UI or chatbots are a mere vehicle, one way of many to deliver what your customers need. Reframing the question will help you focus on “The Why” which is the core of every action and every solution.
Question 2
What is the business outcome you want to drive? For instance, does the chatbot platform enable you to reach an untapped market?
According to Chris McCann, president of 1800-FLOWERS, 70 percent of the company’s chatbot orders on Facebook messenger have been from new customers. Maybe your chatbot idea can get you a more diverse and larger user base too.
Or, it could be about customer retention and customer service, if you want to provide technical support without having to wait for a live technician to be available. You don’t have to create a list of extensive business outcomes, just the key ones (one key business outcome is good too).
The first two questions will help you understand the high level use cases you want to address with your chatbot idea.
Question 3
Now that you have established the outcomes or “Whys,” the next question you should ask is if conversational UI (chatbots) are the best medium for your outcome. Do conversational characteristics like flexible dialog based information exchange help with use cases otherwise difficult to solve with traditional UI?
You can book tickets for your next family vacation at the best price using a website in a few minutes. You can get the weather forecast at your destination using a mobile app to prepare for your trip. How is conversational UI better at these tasks?
To help with this, try the “So What?” technique. Imagine you are sharing your chatbot idea with a very skeptical friend, who always responds with “So what?” Let’s take a hypothetical scenario of Excited Gary who thinks he should build a chatbot for cooking. Here’s how the conversation plays:
Excited Gary: I want to build a cooking chatbot
Skeptical Friend: So what?
Excited Gary: So people can find the best recipes by interacting with the bot.
Skeptical Friend: So what? You have tons of websites and YouTube channels offering the exact same thing.
Excited Gary: This is different. You can interact with it in a less-structured, more friendly way. For instance you can say, “I Want a Chili Recipe that isn’t too Tomatoey,” and the bot will give you that.
Skeptical Friend: So what? How many people do you think search for recipes this way?
Not so excited Gary: …
You get the picture.
Question 4
Does your idea try to replace tasks that people love doing? If the answer is yes, you have an uphill challenge of convincing some very critical and unforgiving users. On the other hand, if you focus on the tasks that people hate, you might have one less problem.
I was talking to a friend who loves planning trips with his friends and relatives but hates the additional back and forth communication to make sure the trip aligns with everyone’s preferences and schedules. If my friend was your target user, you can see which focus area would be the best to pursue.
After you have answered all 4 questions above, you should be in a position to answer the fifth and perhaps the most important question of all.
Question 5
When asked about the huge markup on Ray-Ban and other luxury sunglasses, Andrea Guerra, then CEO of Luxottica Group (which owns Ray-Ban and 80% of the US eyewear market) said, “Everything is worth what people are ready to pay for it.”
So the question to ask is — Is the cost for using your chatbot-powered product or service less than the perceived value. By cost I don’t mean just the monetary value but instead the effort, attention, time and (perhaps) money it will require of users?
About 2 years back, I tried to convince my wife to switch from handwritten grocery lists to an app so we can be in sync and always up to date when either one of us goes shopping. I couldn’t. The reason? Most of the times, when she’d realize we ran out of milk or eggs, she’d be in the kitchen. But not all of those times she’d have the phone with her. For her, the convenience of pen and paper by the refrigerator was much higher than finding her phone which might be in the bedroom or on the couch. Not to mention, after you get the phone you need to unlock it, find the app etc. Too much effort (high cost) to just add eggs to the list.
Last year, I bought an Amazon Echo. Within the first week, my wife happily switched from handwritten lists to Echo. Of all the things Echo does, one of the most common use cases for us is “Alexa add Item X to the list”.
What happened? Both the shopping list app and Echo came with the advantage of having an always available, synced up list (value). However, the cost (effort and time) of using Echo was much less than the cost of using her phone or paper.
If your prospective users are convinced the value your chatbot brings is significantly higher than the cost, you have cracked it! Celebrate! But then get back to work, you have a chatbot to build.
Read about how to get started with chatbot design here
What have your biggest challenges been in building a chatbot? Do you think there’s a sixth question I should add? Do you think I forgot anything? Please leave your comments below so we can all learn from the collective wisdom.
Thanks for reading. And good luck with your next big idea!
A big thank you to Jessamyn Miller, Krishna Vadrevu, Meg Cramer & P.Fallon Jensen for their help refining this article.
