paint-brush
Chatbot lessons from a coffee machineby@vinayakranade
792 reads
792 reads

Chatbot lessons from a coffee machine

by Vinayak RanadeApril 18th, 2016
Read on Terminal Reader
Read this story w/o Javascript
tldt arrow

Too Long; Didn't Read

Your first day at your new <a href="https://hackernoon.com/tagged/job" target="_blank">job</a>, your mentor / enthusiastic coworker shows you where the most important stuff is — bathrooms, water, coffee (or tea).

Companies Mentioned

Mention Thumbnail
Mention Thumbnail
featured image - Chatbot lessons from a coffee machine
Vinayak Ranade HackerNoon profile picture

Not a great user experience.

Your first day at your new job, your mentor / enthusiastic coworker shows you where the most important stuff is — bathrooms, water, coffee (or tea).

Ready to start your training, the first certification you will receive is “Coffee machine operator”. I’ve yet to see an office space where everyone knows how to use the coffee machine. Even after many iterations of the coffee machine as a product, most are still not intuitive to the user. But we have to be forgiving, it’s not like coffee machines come with powerful artificial intelligence computing built into them.

Finite State Machines

In computer science, coffee machines are “finite state machines” or automata. Simply put, each machine has a set of states, often reflected in the new-style digital screens accompanying them like “Fill water” or “Insert coffee pod” etc. When a coffee machine is in the “Fill water” state, you have to fill water. It won’t let you go to any other state without filling water. Of course, sometimes it’s already full of water, and you have to open and shut the water tank to “trick” the coffee machine into thinking that you filled it with water and it can move on.

There’s another kind of annoying finite state machine that we’ve all interacted with — the robot telephone service. “For English, press 1.” (side note: Fortunately our friends at GetHuman are doing something about that.)

To purchase something, please press 2

Given that the original telerobots were designed to interact with numbered inputs only on oldschool phones, and also before artificial intelligence computing, we can be kind of forgiving of this too. But at least we learned that this isn’t a great user experience … right?

Chatbots

Every business is excited about the potential of chatbots and how it provides a new platform for commerce. Unfortunately, most of the bots that are being churned out resemble, sadly, the coffee machine, or worse, a telerobot clown.

Most businesses are on track to deliver a worse user experience via chatbots than their simple website.

This is not because the platforms are insufficient or because chatbots are a bad idea. It’s because the experience isn’t tailored to the medium. For example, if I want to buy some more socks of the kind I already have, it’s going to be easier for me most of the time to just go to Amazon, type in “socks” and just re-purchase my last order, than to go through a chatbot finite state machine that asks me 20 questions before completing my task.

But if I was trying to figure out which socks to buy, a chatbot could be more helpful, by adding value through the messaging medium.

Leveraging the medium

An experience like that would be a bit harder to create on Amazon.com while also making sure that the other use cases are easy to satisfy.

If you’ve ever paid by credit card via telephone (not the Apple Pay kind), you know how frustrating that experience is — “yes it’s Four Four Five Six. Not Seven. No I mean Four Four Five Six, not Seven Four Five Six.” This is also true for chatbots. If a chatbot simply makes you manually type in all your information into a chat window in an unstructured way, the experience is worse than using Google Chrome to autofill your credit card data on the website.

Finally, chatbots have something fundamentally different than most other kinds of app input interfaces. It’s totally free form. To create a chatbot that doesn’t expect every kind of type-able input from a user is to ignore basic experience design. Users shouldn’t have to “trick” chatbots into advancing to the next step in their finite state machine. Chatbots have a lot more artificial intelligence computing power behind them than a coffee machine, and they provide a different method of interaction.

Businesses that shoehorn existing e-commerce flows into a coffee machine chatbot will get nothing from the platform. Simply taking webforms and creating a chatbot that will go down the list of inputs will most likely be a frustrating experience for users, and will have a hard time competing with the existing app or website experience.

Businesses and products that leverage the messaging medium to provide new value to their users will win.

I’ll leave you with this finite state printer from office space. If you enjoyed reading, please click the ❤ below — it means a lot!

PC LOAD LETTER