Launch Goes to Geekend – Olivia Rado Presents on Chatbots

Article by Olivia Rado
5 minutes 31 seconds read

UX Designer Olivia Rado participated in a presentation at Geekend in Savannah, Ga., on Feb. 2-3, 2018. In “The Art of User Experience,” she discusses why designing for a content preview within a chatbot isn’t as simple as it may sound.

Olivia joined moderator Charley Erdman of Dfuzr Industries and presenters Christopher Via of Avesca and Alyssa Greenman of Gulfstream Aerospace. You can replay Olivia’s presentation and we’ve also her slides below. Read on to learn more about the UX of chatbots.

“User experience design plays a huge role in the systems that are a part of our everyday life.”

— Olivia Rado, UX Designer

From taking Lyft to performing daily tasks on the job—user experience a huge part of our economy, and it’s fantastic to play a role in that as a designer.

My project is content preview within a chatbot. It is exactly as it sounds. You’re looking at a chatbot and using it to answer a question. It sounds simple, and it’s assumably simple with good reason. Upon further research, there are a lot of things to discuss.

So without further ado, I’m going to introduce you to Matchbook and the design problem.

Matchbook chatbot

Meet Matchbook, a chatbot who supports customers with helpful information and automated transactions. In this project, the customer-facing display was enhanced by adding a content preview feature with helpful information.

On one side you have the customer who asked the question. In this case, I like to call “the user” the customer because that’s exactly who they are. They have bought a product and have a question specifically relating to that.

Matchbook’s intelligence is powered by the artificial intelligence service of IBM Watson. When a customer asks a question, this is sent to Watson in the cloud. Then the answer is sent back to the customer-facing UI. Each transaction has a load on the cloud, which accumulates in a variety of ways. With that said, it’s even more important for the customer to accomplish their goal as soon as possible.

Watson’s knowledge base is reliant upon the amount of data received, which requires hand-holding from a variety of specialists with strategic thinking. Within the Autodesk digital ecosystem, this includes pulling information from a variety of complex resources.

When the customer comes to Matchbook, they are looking to get a very specific answer. All of these answers are matched (this is why it’s called Matchbook) then squeezed onto the back of the matchbook. It’s tons of information iterated upon through various teams and requiring tons of effort to make it simple for the user (the customer) to give them the answer they’re searching for in an efficient manner.

matchbook chatbot ecosystem

This is the digital ecosystem that Matchbook lives in. There are tons of environments being supported by this tiny little matchbook. The chatbot does a great job of delivering answers but it’s still learning how to support the customer through iteration on the back-end and front-end.

Anytime a customer asks a question, Matchbook actually thinks about it. Matchbook sends the customer’s request up to IBM Watson in the cloud where the information is processed, sent back to Matchbook, and then shown on the front-end.

Within this situation, the chatbot is answering with a content preview. These are content cards (pictured above). This is the current state template used within various parts of the ecosystem.

When the stakeholders handed off this project, they proposed using a current state template because it was proven successful for users. So it makes sense to assume that it would continue to be used modularly. Plus, it doesn’t hurt to save time and effort on the back-end. 

But is this appropriate for Matchbook? Does it feel like a lot of information to squeeze into this space? Is an image actually relevant to the context of a solution?

As a UX designer, it’s my job to do a lot of digging to turn over stones and discover solutions.

I got to work with a really great team to answer all the questions that I had in terms of what would be the most appropriate MVP for this project.

Assumably, from a stakeholder’s perspective, this was it—to take what already exists to be consistent for the user plus save time and effort. But is that really what’s best in terms of answering these customer questions efficiently?

These are the three types of content represented within the chatbot. We have: a form, a troubleshoot, and an article. These are abbreviated wireframes of the current state with the inline links driving to the browser view.

*Note – For more background on Geekend, check out this post from ATDC.