In the last decade, communication through online text has skyrocketed. Our culture has transformed to include texts and social media messaging as some of our main forms of communication. With only a quick taps on our phones, we can instantly connect with anyone across the world.
As we have adopted the use of text messages, we have also adopted different habits regarding it. We all know someone who takes forever to respond to texts. Or someone who uses way to many Emojis (I'm looking at you Mom!). And the person who uses "haha" and "lol" way too often. The list goes on and on.
Likely, you can easily describe someone's texting "type" after only a few conversations. You probably even have a pretty good idea about your own texting habits. But it can be hard to know for certain without quantitative data.
We wanted to start with one of the most common form of communication in today's world, especially among college students: iMessage texts. Our goal was to analyze iMessage conversations on macOS computers, and share the data back with the user.
We talked to friends, classmates, and faculty about our idea. Although some were skeptical about the impact that the data would have on them, everyone seemed to be curious in at least viewing analytics about their text message habits. And that was enough to propel us towards building mockups and an initial prototype.
By allowing users to view analytics about their text messages, we wanted users to learn about their texting habits and better understand how their communicate with others. We wanted to tap into that feeling of curiosity that people initially mentioned in our first conversations, and we hoped to eventually create a platform that allowed users to explore things that they didn't know about their texting habits.
Through our user interviews, we gathered information on features that our users would be most interested in, including the ability to filter analytics by conversation with a specific person. But one user quote defined new focus for our team...
From this point, we knew we had to focus on creating a comfortable and trusting environment for the user to interact with the website. We looked at how other websites that handle sensitive information, and drew inspiration from Apple, Google and even 23andMe.
By creating user personas for different user groups that we identified through our interviews, we were able to empathize with our target audience and dive into the project with their needs and wants in mind.
We wanted to get the data to the user as quick as possible, and therefore we chose to make user flow as straightforward as possible. In terms of the user interface, we also believed a minimalistic, modern design would help build our website's credibility. With the minimalist design and a straightforward user flow, we aimed to create a clean user experience that flowed with each step of the experience.
For example, initial mockups included a navigation bar at the top of each page. However, we eventually decided to remove it because it cluttered our design and offered little benefit, giving that our site navigation was already straightforward.
Early on, we had to make several design sacrifices due to technical constraints. For example, we strongly believed that making an iOS app for iPhones would have been ideal, because our user interviews showed that not all of iPhone users had a Mac computer that they also texted from. However, Apple’s iOS infrastructure made it nearly impossible to access iMessage conversations from a third-party app without jailbreaking. The only plausible way we could quickly access the texts was through a "chat.db" file that Mac computers store when the user is logged into Messages onto their computer.
Another one of the main constraints that we worked within was the upload process for the 'chat.db' file on Apple computers. Because of how the file is stored, we needed to provide clear instructions to the user during the upload process. We brainstormed and tested several onboarding methods, including a slideshow tutorial process, a short GIF outlining the process, and simple text directions.
A week before our project was due, we released a prototype to the rest of our class and sought feedback. Suggestions included...
From the feedback, we knew which parts we had to improve before launch day. We added moments of 'delight' to our project, including fade in animations and a revamped loading screen. We also iterated on several different versions for our upload instructions, eventually settling on simple text commands after further user testing.
One of the biggest things that we took away from our prototype stage was the fact that our users wanted more than just quantitative data. Our biggest addition was a sentiment analysis system, in which we looked for keywords that indicate happiness, anger, stress, love, and outgoingness (which we called the 'social' sentiment).
Originally, we just displayed scores for each sentiment in the analytics, but we began to see a larger potential for the new sentiment scores. For example, we created a 'Reconnect' graph that displayed contacts with high love sentiment, but hadn't been in communication for a period of time. We believed that this type of feedback analysis could help our users improve their relationships with others through text messages.
As the number of graphs increased, we needed a more efficient way to organize the analytics. Because our analytics could be grouped based on the type of information they showed, we believed that the tab system implemented above could promote fluid exploration without overwhelming the user.
From the outstanding positive feedback from both our peers and faculty, we knew that we wanted to expand Left On Read to our friends and family, and possibly even the world. With our increased user base, we were able to receive much more user feedback. Since then, we've added several features to make the user process more exciting, including more filter options and the ability to share graphs on social media.
Even from the start, we knew that Left On Read wasn't going to change the world. We were first interested in the project because of the proposed challenge of parsing texts and displaying analytics. But we tapped into a service that catches almost everyone's curiosity, and we want to share the feeling of exploration when we first got our data. It was clear that people loved to see data about themselves.
Since the end of spring 2018, we have been working on a design overhaul and preparing for a mass launch to the rest of the world. In the fall of 2018, we were accepted as a partner project for the DALI Lab, and onboarded two new designers and one new developer for the term. I have since taken a step away from the project because of my study abroad in Budapest, but look forward to working on the project when I return! Check out what we're up to at leftonread.me.