Transforming Ask Sora — A Consumer Chatbot [UXDI Project 5]

[A General Assembly User Experience Design Immersive final project retrospective]

Prologue:

“She’ll be their best friend, someone they’ll look to having a conversation with.”


She stirred from her nap.

Though time is always in a state of flux in this world, it sometimes get boring for a youngling like her who just arrived in this area of Messenger Land.

She looks out of the window, gazing at the large group of visitors gathered around the uncle’s house across the street. He has his way with people, aside from being able to provide arrival times of the local bus service, it’s his jokes that people usually come back for. That probably comes with experience, she thought.

She stretches as she sat in front of the three doors her creators built into her house, recalling what they told her:

The first door, show your knowledge to the visitors who knocked on it, they will be keen to know, her creators assured as they handed her the Book of Smart Facts.

The second one, they said, play with them, ask them which do they prefer when given two choices.

And finally, the third door, her creators instructed, invite them in for a chat, let them ask you anything, be a friend to them. Don’t worry if you don’t have an answer immediately, just hand them the cards of subscription and let us know. She nodded eagerly.

She exhales, looking out again while waiting patiently for someone to knock.

…This is the story of SORA.


Ok, let’s get this out of the way first, SORA stands for Smart Objective Robotic Assistant.

Moving on~

With the redesign officially launched, it’s probably safe to publish this article. As a final project for the UXD course, we were assigned to actual clients, I was fortunate to be regrouped with my Project 4 teammates again and given the task to redesign Ask Sora, our client’s consumer chatbot. Something very different from what my other course-mates received.

What is a Chatbot?

Wikipedia: A chatbot is a computer program which conducts a conversation via auditory or textual methods.

I must admit, though I’ve heard of chatbots, I’ve never really used them or have in-depth knowledge on how they work. Which means time to do our research.

Know Thy Product and Competitors

We started our research by first understanding our client, Aitomon’s intent, expectations and target audience. These information helped in establishing our user interview questions.

Then to evaluate the current state of the chatbot, we needed to know the following:

  1. What is a chatbot? (As mentioned above.)
  2. What makes a good chatbot?
  3. How to use heuristics to test a chatbot?
  4. Who are the competitors? What can we learn from them?

Ask Sora the Chatbot

How Ask Sora looked like

Ask Sora was created with the intent that she’ll become a friend millennials want to converse with, using the “Ask Me Anything” feature. It’s through this interaction, data can be mined for machine learning.

“Smart Facts” and “Vote Now” were secondary features to entice users to revisit the chatbot.

By evaluating the chatbot, we identified the various heuristic violations summarised in the following points:

  • Objectives of the chatbot and its features are unclear.
  • Limited by Sora’s answers, users are unable to interact much with it.
  • Chatting with Sora can be frustrating.
  • Too much textual info, making it hard for users to digest.

With this exercise, we also considered potential scenarios whereby users may experience confusion or difficulties, and prepared ourselves to probe further into their thoughts.

Targeting the Audience

As the target audience were millennials, we visited various tertiary institutions to conduct our interviews and testing. Interestingly enough, there were several similarities in the users’ responses, categorised below.

Key insights

Something I’ve learnt through the sessions is to understand a user, one doesn’t always have to ask questions directly related to the product. It could be, for example, what mobile apps they use daily, why and what do they like about them. You’ll never know when that bit of information can be useful.

Putting Our Heads Together

Collaborating with the men-in-black (our client)

To get our client on the same page, we held an ideation workshop where we presented our findings, personas created from the user research data and brainstormed potential features.

Persona #1
Persona #2

The workshop proved to be extremely useful:

  • Our client have a better understanding of what issues the users faced and what their interests are.
  • A focus on essential features best for both users and business, based on client’s input on the ease of implementation.

The Proposal

Restructuring the Content

With the research data, we decided to redesign the entire content and user flow. This was supplemented by the good industry practices we have learnt during our initial research.

Sora’s New Form

Throughout our interviews, users have feedback that they couldn’t relate to Sora’s anime girl image, even commented if the chatbot was aimed at the female audience.

So we have to change that perception.

Meet Sora the squirrel, an animal that represents nimbleness and intelligence, it brings nu(t)ggets of information and entertainment to users in the same way as a squirrel would source and store nuts.

Of course no good chatbot would be complete without a personality. The way to let its personality shine is by crafting the dialogue in a fun and witty manner. Also equally important is a response must be provided with every user selection as feedback closure, even when Sora doesn’t have an answer, it will tell the user that it will get back later.

Using micro-copy to keep users interested in interacting

Validation through Testing

Unlike the usual app or website testing, messenger chatbot testing really focuses on the content. Key feedback from our users were mainly on the micro-copy, in which we made further simplification and clarification.

Revised content

Wait There’s More

Being a Facebook Messenger chatbot, we felt Sora also needed a new Facebook page. It will be easier to use it for spreading awareness and posting of seasonal events; helping to increase and enhance user interaction.

A potential feature for future development is pushing of exclusive content such as promotions, depending on the user’s geo-location. Partnerships can be formed with businesses at said location.

Another future feature will be content curated for users, personalised to their preferences. Something that was brought up during our interviews but currently difficult to execute.

Epilogue

This was an intriguing project. Compared to mobile apps or websites, redesigning a messenger chatbot comes down to its core functionality which is the content. Through this project, I’ve come to appreciate well-written copy and content even more.

Another lesson would be the importance of collaborating with the stakeholders. It is when they understand and see from the users’ perspective then the product can be improved, realising its potential. The same applies for any product. I think sometimes as designers or creators, one must know when to loosen the hold on your product. It was a great idea brought in by my team-mate, Xian, leveraging on her past experience.

The world of chatbots is expanding exponentially and will only continue to do so. I highly recommend any UX designer to work on one when the opportunity arises.


Lastly, as mentioned in the beginning, our client has since taken up our redesign proposal (though not entirely) and launched it officially. Here is the link to the new Sora.