Zoom’s Chatbot
Previous reviews: McDonald’s, Inbenta, Wysa, Tonal, NASA, Planned Parenthood, Lowe’s
Recognize the screenshot below?
I bet you’ve seen it a thousand times! It’s the web page that shows when you click on a Zoom link to launch a video chat. Mostly I just wait for it to do its job and then close the window. But today I saw this chat bubble in the bottom right corner… let’s investigate!
The bubble itself is blue and white, fitting with Zoom branding. It also doesn’t use a character, but simply a chat bubble. This seems to deprioritize the personal and prioritize the informational or task-oriented goals. It’s also subtle, without text-filled chat bubbles popping up, without audio, with a simple two-color scheme and a simple icon. It seems to be a fallback option for people struggling to connect, which is a different use case than a sales conversion tool like some of the other reviews I’ve done (Tonal, Lowe’s).
If we click the bubble (at 100% zoom in Chrome on my external monitor), we get this:
I am surprised that the first experience doesn’t fit on screen. This seems to be a common thing for bots. Zooming out on Chrome doesn’t help fit it all on screen either. You have to scroll to see all the buttons.
From the initial view, the bot has a name (Bolt) with an associated icon ⚡:
Why is Bolt introduced three different places? This seems like overkill, particularly since one consequence is that the initial content can’t all fit on screen, but it’s not the first chatbot I’ve seen that presents the bot’s name so many times (e.g. Tonal).
I find it a bit confusing that the Zoom logo (blue camera) is used next to the opening chats. I feel a conflict between whether I’m speaking to Zoom or to Bolt, the Zoom assistant. Bolt says I’m speaking to Bolt, who has a ⚡, but the chat uses the blue camera icon. How might this be different if the Zoom icon were to the left of the chat bubbles?
So, what about the dialogue?
The initial content has a greeting, introduction, orientation, and a question. It seems like the introduction is redundant with the text above:
Let’s dive in to the orientation and question.
The orientation seems designed to train the user on what is most likely to be a successful interaction… if they enter free text. And yet, the user is presented with five buttons, which don’t relate to this guidance at all. I wonder if the orientation is more confusing than helpful. You could always give this orientation text after a free-text input that fails to match.
The content of the buttons themselves (the intents) seem appropriate for a Zoom context. I can imagine this bot does a lot of helpful triage and user support. Clicking through them, I see that they link to how-to articles, Youtube videos, and in-line explainer text.
Having focused most of my attention on the initial welcome screen, I wondered what it would look like to compare the Zoom chatbot to a one modified to include my proposed changes.
I like that my alternative design doesn’t require scrolling to see all of the initial content. It avoids some of the original’s redundancy (intros, icons). It feels like a more coherent relationship with Bolt, who speaks on behalf of but is not Zoom. I think the spacing is better.
Powered by Ada
The bottom of the chat window shows that this is a bot powered by Ada:
Clicking “ada” here launches a Zoom-specific version of Ada’s web page!
https://www.ada.cx/?utm_source=zoom&utm_medium=referral&ref=zoom&utm_content=user-group-b
If you go to the non-parameterized web URL https://www.ada.cx, you get a general welcome:
Bravo Ada! I like this customization. It’s also great for analytics, since you can track how many people click through from the Zoom bot to Ada’s website. I guess I added a few more clicks to their logs :).
Having done a few reviews now, I am starting to see patterns in my recommendations. For example:
- Get all initial content on screen without forcing the user to scroll
- Avoid redundancy
- Cut unnecessary content
- Be clear and consistent about the bot’s persona
Perhaps there are counterexamples, and people may disagree about what is “unnecessary” or “redundant” or “clear”, but these are principles that jump out to me as ways of improving conversational designs.
Thanks for reading!
Copyright 2021 Joseph Tyler All Rights Reserved
https://josephctylerwords.medium.com/zooms-chatbot-f39fc29b6e4f
