Champion Waste Chatbot
This is a chatbot dedicated to making information regarding products and services easy to find for business owners, construction managers, and homeowners.
Project Specifics
Semester
Spring 2021
-
During
COVID-19
Pandemic
Role
Conversational Designer
UX Designer
UX Writer
Content Developer
Content Manager
Category
Assignment
Team
one (me)
Research
Why a Chatbot?
I created this chatbot in order to give potential customers of Champion Waste information quickly. Champion Waste has so many options for individual companies or products, so the chatbot gives information faster than browsing the website itself (which happens to be page-long descriptions of their services, products, etc.). Since most business owners, construction managers, and homeowners (the main customers they receive), don’t have the time to go through the entire website, they can easily ask the chatbot for what they’re looking for.
I wanted to reflect the website in the chatbot, so the content was extremely trimmed and the chatbot had almost the same organization as the website. When looking at other competitors’ websites, I realized how many options and buttons each chatbot had at first look; therefore, I decided to give the user options they could input into the chatbot. Not everyone knows what a chatbot can do, but with the topics and options the chatbot has available, it’s simple to find almost anything a chatbot can offer. The only other fun function I added was recycling fun facts that were on the website itself.
Competitive Analysis
To get a better understanding of what was already on the market, I looked at specific competitors' chatbots. Surprisingly, major garbage companies had chatbots that help their main customers.
The companies I looked at were:
Key Takeaways and Benefits
-
Each chatbot is targeted towards its main audience.
-
The chatbots only contain information about their services, products, and contact information. Nothing else is included.
-
Tone and plain language are highly valued in each interaction.
Main Issues with Each Chatbot
At the time this was researched, Republic Services did not include ways for the user to explore what the chatbot did. Since chatbots are new, users are still somewhat unsure of what they do. Giving options through buttons really help users understand what they can ask and request. Recently, they changed their chatbot to have buttons.
Waste Connections includes way too many options, so much so, that users on a computer have to scroll to the right to see all the options. It's good to have options, but too many can intimidate users.
For Waste Management, the only issue is that users can't go back to change an option that they picked before. The chatbot even asks if the user needs anything after a few seconds of no user response.
Building
Google Dialogflow
Google Dialogflow is a cloud-based virtual agent builder that is able to connect and integrate with different kinds of digital channels. It even uses AI to figure out what users would input into the chatbot; however, for specific cases like with a garbage company, a UX writer or conversational designer would come in and write down different instances of what users might input for different answers. This means that the chatbot would need to go through some testing.
In this project, I had to create intents that are content topics, and then, within each intent, I had to include the responses the chatbot would say to specific types of input from users.
Organization
Chatbots have a lot of written content, but it all needs to be organized by the response. For this chatbot, I organized it by
-
content topic intent
-
response
-
another specialized response dependent on user input.
-
-
Technically, each content topic intent and its responses are single linear user flows. Therefore, it's important to make sure that every single possible user input is taught to the AI in this chatbot.
Below, I added a few pictures to get a better idea of what it might look like. The first image contains all of my content types. The second is an in-depth view at the organization for one content type.
All intents and content topics.
Service Content Topic Broken Down into its Responses
Testing
Informal Unmoderated Talk-Aloud Protocol
Since this was an assignment, I was lucky enough to have other students test out this chatbot through Google Dialogflow itself. Because of COVID-19, I couldn't ask users who are business owners, construction managers, and homeowners to do proper testing.
Beforehand, I had numbers included in the first chatbot response to indicate the different options users had since the UI/format options are severely limited with Google Dialogflow (the options for UI are super buggy at times).
I had three students test out the bot for anything that I could have missed when connecting responses through the concept topics. I only found one issue. Originally, I didn't include numbers as user input, but I changed it so that it can be easier for users.
Number User Input Gets a Response from the Chatbot
SME Review/Moderated Think-Aloud Protocol
Additionally, I was lucky enough to find an SME who has worked in the garbage and waste industry for 10+ years, both residential and commercial collections. I conducted an informal moderated think-aloud protocol and asked a few questions about the chatbot.
This SME really liked the chatbot and thought that it answered all of the questions most customers come to him with. I can't say which company he works for, but he said that he wished that his company was more transparent to their clients when it came to information like this.
There weren't any changes that needed to be specifically made for the chatbot based on the SME's input, but the more eyes that looked on the chatbot, the better.
Review
What Could I Have Done Differently?
I would've included more ways for users to interact specifically with each response concept topic instead of one interaction at a time. I liked how Waste Connections allows users to get help at any time no matter what users are asking.
I don't know how to change the UI properly without bugs or complications, so I think figuring that out will definitely help since users need options available to them at all times.
My Thoughts
Overall, this was a great project to practice my UX writing and content management. I loved the idea of writing intents and figuring out what users would ask for. Having users think about what they need to write can lead to users becoming frustrated. Through testing, it will be clear what users will want from these chatbots. Since this chatbot's only function was to give information, it's appropriate to have a single linear user flow.
The Champion Waste Chatbot
I cannot link users to an external page of this chatbot but only integrate it onto his page. Below, I have an embed of this chatbot. It's interactive, have fun and check out the fun facts!
The Champion Waste Chatbot Redesigned
Since I was not able to fully edit the UI to my liking, I used Figma to recreate what I wanted the chatbot to look like on the current Champion Waste website. Below is a frame of the redesigned chatbot.
.png)
A Frame of Redesigned Champion Waste Chatbot