Pizza Bot
Ma & Pa Pizza delivery bot
Overview
Ma & Pa is a nationwide pizza chain that takes order by phone & online. They want me to implement the following:
-
A digital bot with agent escalation on the web
-
A digital bot with agent escalation in facebook messenger
-
An IVR that routes to their voice agents
My role was to research, sculpt out the user flows, write the sample dialogues, and prototype conversations that the bot would take.
Tools
Draw.io
Google Excel
MS Word
Facebook Msg
Sketch
Role
Design Lead
Sample Dialogues
Duration
Nov 2020
(3 weeks total)
Opportunity
Ma & Pa's Pizza wants the bot to find the quickest path for the customer to place an order (including authentication, order details, and payment). I have an omnichannel conversational bot that uses Natural Language Processing to understand customers' intent.
Hypothesis
By providing consistency in all their channels (IVR, Facebook Messenger, and Web) this will make their customer base feel welcomed and familiar. This will direct customers to use the bot more often, cutting the time on human labour.
Research
Based on the criteria given, I am designing a helpful bot. I looked into competitors such as Domino's Pizza to learn more about how their Facebook bot operates. I also took inspiration from their use of cards and images to incorporate into my own bot.
Ideate
I created a high level flow for customers ordering pizza, taking into account the many turns in collecting information.
Then, I wrote sample dialogues taking into consideration how different scenarios may impact how conversations may flow for IVR, Facebook Messaging and the Web.
User Flows
Sample Dialogues
It's important to take into account for different designs based on channels. I focused on 4 journeys and wrote sample dialogues for each:
-
Via web that results in bot answering successfully
-
Via IVR that completes successfully
-
Via Facebook Messenger where the bot did not answer successful and the user needs to be escalated
-
Any other channel where the user goes 'off the rail' and bot needs to steer user back on track and complete the journey
Page 1
Page 2
Page 3
Page 4
Conversational
Prototype
I developed chatbot and IVR prototypes based on the scripts to deploy to users. Here I have showcased a mockup of how it looks in Facebook Messenger.
Facebook Messenger/Web
Conclusion
I didn't realize how many different turns I had to account for (no match, mismatches, visual buttons, word placements). I learned that designing for conversations on IVR are significantly different than designing for a chatbot. Conversations on IVR need to get to the point quicker than a visual interface or people will tune out. It's so exciting to be writing conversations and finding opportunities where chat and voice could benefit in the future!