Case Study: Textie
How we designed a better way to text.
We have a bad habit of neglecting unread messages.
It’s stressful to see our unread texts pile up, and many people cope by putting them off until later. Especially as we’re physically distanced from friends and family right now, our team wanted to make sure people aren’t also relationally distanced because of notification anxiety.
Getting familiar with the challenge
We began with a diary study tracking the unreads of 6 college students over the course of a day to quantitatively observe the rhythm of notification build-up and resolution.
From this information along with qualitative interviews, we made a persona combining behavioral patterns, mental models, and characteristics of people affected by an inability to manage their texts.
Experimenting with a rough intervention
Initially, we wanted to change how notifications are surfaced on phones. This was our best-guess solution direction as we didn’t actually conduct an ideation session at this point in the quarter.
We asked participants to write a sticky note for each time they deferred a response, over the course of a week. We were open-ended about what went on those notes and where to put them. The idea was to get a sense of what information users expected to see on a reminder and how they would organize them.
Our team gained insight into what users want to see on notifications but, participants actually found the activity to be harmful to their stress levels. We discovered by system mapping that nearly everything in the user experience of texting is already additive in nature, making things overwhelming. Of course the intervention failed: all it did was add another obligation into the mix.
Defining a better way to intervene
After learning from our participants in the intervention study, our team came together for a brainstorming session where we landed on three top ideas.
We were most intrigued by idea 3: an end-of-day text summary. What if your unread messages could be fed to you at the end of the day in a story? We’d reduce the friction of seeing conversations and link it to an already easy-to-fall-into browsing behavior.
It’s rare to reinvent a completely new paradigm for a common experience like texting, which excited us a lot. It would also make use of one of the hottest interaction trends (stories), too. We imagined it living in Arjay’s life like with a product-as-hero storyboard.
Task and wire flows
Taking our idea from this promising concept to an actual product meant first defining what tasks users would experience when using the app:
By thinking about which of these tasks belonged together IA-wise, we identified 4 main interfaces:
- Onboarding. A step through series of screens to help users set up the Textie app.
- Home screen. A landing zone with conversations you have to visit for the day, as well as your current progress through them.
- Intermediary screen. Like Snapchat’s screen in stories that previews the person who’s pictures you’ll be viewing, except here used to provide a subject line for what the conversation contains.
- In-story screen. The actual conversation interface showing texts as a stack of story cards, including reply suggestions and a reply later option.
Then, we translated these into wireframes.
Making it real with a visual design and mockups
Visual identity
To go to higher fidelity, we thought about brand elements first. How might we visually express that texting can be less stressful? We leaned into cool colors, roundness, and the human touch to convey a sense of calm to our users.
👆 Style tile, 👇 Sample brand collateral showing our aesthetic.
To support the transition from brand identity to actual product, I did some additional work by creating a design system and component library in Figma to make interface creation across the team consistent and aligned.
Mockups
With the brand created and a design system ready, the team turned their wireframes into high fidelity static mockups.
Grouped by feature, our mockups looked like this:
We even included some additional features that users would expect when using this app, such as conversation details and settings.
Creating these was very iterative. For instance, the final design of the summary stack was something I revisited multiple times to make the expected interaction design of the story view more obvious.
Linking it up and testing it
The final step in the process was linking together static screens and handing it off to users. We were anxious to see whether people would understand our new metaphor for texts, and if the intended use case of messaging only once in a day was acceptable. We created 3 scenarios for our usability test participants to walk through:
- Setting up a Textie time. We wanted to answer: do those anchors make sense for establishing the replying habit? Are people good with texting only at that time?
- Responding to an uncomfortable message from an individual conversation partner. We wanted to answer: will people get spurred to respond even in hard cases if given a helping hand? Can people find those reply suggestions?
- Deferring an extremely busy group chat. We wanted to answer: do the times for delays make sense? And again, can people find them?
Ultimately, the app had notable success in more overarching areas:
“I’m uncomfortable responding to messages after a while, and this would make things way easier for me.” — K.P.
“I like the focused swiping story. It felt better than the chaos of a normal list of convos.” — D.P.
But within the finer details of the interaction, there were some usability concerns we didn’t see ourselves until our testers pointed them out:
“I’d just exit this instead of trying to see. I’m nervous about whether read receipts would be sent.” — D.P.
“I thought the X would remove the conversation from the summary, not kick me back to the homescreen.” — K.P.
Collecting feedback from here and elswehere, here are 5 directions for changes to make if we had a little more time with our project:
- Clarify the navigational elements of the story view. At a high level, people understood swiping between items but had some confusion about what exactly the X, info icon, and progress bar pertained to.
- Improve the recognition of intended swipe direction for reply suggestions. Because moving between conversation and viewing more smart replies is done with the same gesture (a right-to-left finger swipe), unintended progression can easily be activated.
- Clarifying integration trickiness, e.g. through read receipts. Since Textie sources conversations from other apps, we need to be clearer about whether read receipts (or other things) that don’t natively exist on Textie are still going to take place by proxy.
- Better tailor the copy of suggested replies. People generally felt that the short suggestions were too curt to feel right.
- Make suggested replies populate the text field instead of auto-send. People found the smart texts useful as a starting point but wanted to add onto them in their own words rather than immediately send.
What more is there to do?
Textie could have real potential—if not as a standalone product, perhaps as a new way for native OS’s to show messages or to ping people who struggle with notification overload. However, a lot more would need to be done to make this possible at scale.
For one, there are the ethics considerations to work out. Texting is very personal, and few would want messages to be dissected by us even if it reduces their stress. We also want to be inclusive of all kinds of messaging styles, and should be very cognizant about suggesting “the right way” to text when the same solution for interpersonal connection might not work for everyone.
A longer-term study where Textie is now the intervention is also a natural next step we didn’t have time to do. Do people’s overall number of unreads get reduced by using our product? Do they reply back to partners with greater frequency? Do they feel better about communications once they’re on the platform? These are all core questions to our success that we could answer by releasing our trial into the world and seeing whether we successfully solved a need.
For now, doing the project was a wonderful exercise in learning about the challenges of behavior design, and how we can apply better product and design thinking to solving those challenges.
Special thanks to Paulina Anzaldo, Medha Verma, Mary Y Zhu, and Jennifer He, who were amazing collaborators that helped to realize this project.