Assessment of the Conversational UX Experience According to Grice’s Maxims Framework— Roo Chat-box
Non-spoken; On Computer; Website-based; by Text Message
About Roo
What is Grice’s Maxims Framework?
via Erika Hall — Defines the “social code” most of us follow for conversation.
A chat-box should be…
Quantity: just enough information
Quality: be truthful
Relation: be relevant
Manner: be brief, orderly, unambiguous
Polite: don’t impose. Give options. Make listeners feel good
Good User Experience
1. Greetings in human-tone
In the beginning screen, there is an emoji-like character. Roo aims to answer questions about bodies, sex, and relationships that cannot talk comfortably in public. I think the greeting is a good design because, as Melaine Polkosky says, chat-box’s tone or voice meets expectations, which in my opinion, means it should have empathy and take care of users’ feelings like a real human.
2. Instructions for the first-time users
People can either follow the instruction or skip it, and also Roo provides examples of what users can ask; this is a polite manner according to Erika Hall states in Grice’s Masims since Roo gives options to users in different needs. Also, the dialog is brief, orderly, and unambiguous.
3. Key feature: ask a question
After clicking “Ask a question”, it jumps to a new page that offers some popular topics. When typing keywords, Roo suggests related questions. This feature goes well with relation metric in Grice’s Maxims because the information is all relevant to users’ queries. As Cathy Pearl says, the chatbot should allow multiple ways to understand the same thing.
SUM UP THE OVERALL POSITIVE EXPERIENCE
Turn-based
In conversational UX design, the most crucial thing is to let users know when it’s their turns, addressed by CathyPearl: more than one conversational turn as well. By asking questions or provide explicit instructions, Roo makes it clear what the user can say.
Goal-oriented
As both CathyPearl and Melanie Polkosky point out, users should be specific about their concerns and then Roos can help find the most appropriate answer. Though some limitations appear, Roo ensures the efficiency of the interaction and how well it serves the user.
Uses common vocabulary and is friendly, supportive, and courteous
Principle by Melanie Polkosky
The sentenses in Roo is easy to understand. Even as a non-native speaker, I feel relaxing to read the dialog.The conversation can leave users a “memory” (Cathy Pearl), thanks to the immitative experience with Roo.
Quality: be truthful
By limiting questions uses could get access to, users can only select existing questions in the Roo’s database. Besides, most of the answers are general and can be applied to almost everyone. Through these ways, Roo guarantees the accuracy of its replies.
Build better voice apps. Get more articles & interviews from voice technology experts at voicetechpodcast.com
Usability Issues
1. Too much waiting time after answering a question
When I was chatting with Roo, I found it needs more than 5 seconds to open the “Want more info?” dialog box after answering a question. This violates Grice’s Maxim quick and clear principle. This problem could be solved technically.
2. Not enough conversational dialog
After I click “sure” to learn more info, Roo gives some access to third websites, which means users will jump into other pages instead of communicating in the chat-box. This feature disregards the quantity principle, where advocate chat-box should have just enough information. However, in Roo, users have to look through a large quantity of text information, which may cause senses of overwhelming and also impede the conversation flow.
Recommendation
Instead of clicking “Learn More”, Roo could let users type in which entrance they want to enter, for example, “first one”. Then it is Roo’s turn to “speak out” some essential information in a conversational way.
3. Users cannot input personalized questions
When clicking on “Ask a question”, Roo opens a dropdown for searching options, and users can only find the most relevant one based on keywords of their questions. This opposes Deborah Dahl’s theory that a chatbox should respond to indirect requests. However, Roo has little error-tolerant.
Recommendation
Roo should be designed to auto-recognize keywords of a sentence that uses feed in, and then predict the possible conserns that uses may raise, continuing dialog in this way. Increase the turn-based sense.
4. Trying to escalate to a person
Access to customer service is hard to find and not intuitive. I accidentally hovered on the Roo’s face, and it became the active status. After clicking, there is a choice of “chat with a person.”
Recommendation
Personally I do not think new users will feel easy to find it. I recommend that the button could be placed in the top navigation level of information architecture and designed to be more visible, or add the text signal.
Compare person service with Roo chat-box
Compared with Roo, the person’s response has more human sense. The person first asks, “Do you have any questions I can help you with”, and then reminds that “Do you want to continue chatting?”. Finally, after confirming users are not there, the service person ends the chat. This series of action is more error-tolerant.
CONCLUSION
The challenge for dconversational UX is to use intentional reasoning to help people make decisions without thinking and acquire information without effort. Roo is far from “just like talking to a person”. It still need to look into more complex, varied and natural possibilities, not only handle the simpler interaction (such limited choices with single clicking) well.
Reference
Erika Hall’s, Conversation Design Ch. 2
Deborah Dahl’s essay from Ubiquitous Voice
Cathy Pearl’s essay from Ubiquitous Voice
