Twitter wants to be the place where people find out what's happening about the world. With the growth in Twitter's users and new algorithms that tend to result in more "viral" content, it's easier for people to feel lost in the conversation. Our objective was to make notifications and engagements more intuitive at scale, to convert and measure more follower relationships and sustained interaction between users.
I led the development of a series of feature fixes on Tweets.
Notifications are key to the Twitter platform: it's how engagement spreads. Recent Twitter developments trend through this, such as with algorithmic feeds that recommend tweets from the interactions of our followed user's own followings.
After being tasked to identify an area of improvement from the user profile page, we thought intentionally about core interactions becoming a spectacle. In contrast to desired goals of larger reach, inhibiting large conversations closes people on to follower loops. We believe there's space for notifications as they are today to grow to a place for stronger connections, and not just meaningless numbers.
Due to the timeframe (less than three days!) and with the given scope, we relied on quick user interviews, conversations with current engineers, and friends-and-family input to gather frustrations. We penned these down on boards and saw a common theme around interactions:
There are three problem areas we wanted to tackle:
Interestingly, many users sharing their frustrations gave comparisons—they weren't looking for new groundbreaking features or algorithmic changes, they wanted quality of life fixes—referencing functionalities present in other apps. These smaller gestures and usability pieces stuck out to us as our focus.
We developed two key personas that helped guide our thinking in people's approaches to these problem spaces.
Checks on Twitter multiple times a day (more when procrastinating) as a nineteen-year-old university student up-to-date on pop culture. Twitter is not just for news, more like tidbits, first joke source. and group DMs. Interacts mostly with friends with a <700 followers, most are mutuals. Follows celebrities, pop culture, and comedy accounts on Twitter. Engages liberally through likes and retweets and isn't afraid to tweet their mind or have conversations run long in threads.
Checks Twitter a few times a week, especially when prompted by the news (e.g. following breaking events, hearing their President tweeted something) as a middle-aged, responsible user. Not enough time to be online without intent. Not concerned with profile completion, but excited at conversation and discoverability. Likes tweets liberally, low follower count (<100). Mostly engages through replies and "lurks" on the application.
From this series of observations and with our personas in mind, we defined goals based on Twitter's recent product releases and opportunity areas in our problem scope:
We want Cane to not feel overwhelmed when he goes viral, or be able to join in the conversation even after a few hours. Mary wants to sort through replies faster because they have little time, and feel like their inputs are being seen when responding to larger figureheads—they may also benefit from more followers and interaction.
Repurposing existing components was also a key design decision. There are QoL fixes and shouldn't take too much sprints to put out; we wanted something measurable that wouldn't take too much engineering time away from larger features: so to keep the design true and build on many underexplored portions of Twitter's functionality, everything is reused.
Better engagements introduces an array of quality of life fixes in the tweet and timeline engagements area, here's a brief overview of the spaces that we're touching:
Users want to see who's participating in the conversations around with them.
At the moment, viewing the list of likes or retweets doesn't offer the list of users engaged in any sort of order.
To make tweet engagements more navigable and clear, we decided to redesign modal views for likes and retweets on the details page.
In the new modal, it's now possible to make sense of accounts engaged with a tweet. Reverse-chronological ordering and categories empower Twiter users to know their audiences better.
Categorized views let users filter through mutuals (people you follow & people who follow you back), people you follow, and others. When looking at engagements with viral tweets, it's easier to see how your network has connected with it.
A search bar allows users to specifically filter through @s and account names to look at whether a user has interacted with material previously. When this information is public on profiles, it makes sense to have this easily viewable on Twitter's other spaces, too.
The best way to captivate audiences is to engage directly with them. With these new engagement views, we saw an opportunity to allow brands to appear at the top of the appropriate category as "promoted" given they've genuinely interacted with the user's Tweet.
Allowing promoted accounts in search encourages brand interaction and exciting new lead generation opportunities beyond the feed: this is a step forward to let brands tell users they're here, and they want to listen.
In our suite of interface changes, we also decided to be intentional about interface changes to promote user safety. Because these movements remove the veil under each user's engagements, it was critical that we propped up Twitter's existing security measures: such as muting or hiding notifications from users who don't meet certain standards of verification to prevent harassment or abuse.
To the left is the new modal view with the search bar with tweets categorized in collapsible categories.
We saw a huge gap with quote tweets, which are growing as such a critical part of twitter conversations. The only way to view all of a Tweet' quoted replies is to copy its permalink and paste it into search. It doesn't make sense to obscure this part of the conversation.
Adding a tangible quote number and building on a modal for transparency opens us up to more diverse and sustained conversations.
We decided to modify the current tweet details view to host information on Quoted Tweets.
Similar in appearance to the Like/Retweets modal, the most recent quoted tweets are displayed with the original tweet at the bottom—mimicking its original display. It's easy to see how repetitive (yet, still funny) jokes made through quote retweets can get at times.
As we were assessing the profile view, we noticed a lesser used feature of Twitter: lists.
Lists on Twitter are one of the most important spaces for users to connect with one another. They allow users to curate interests, build feeds to share with others, and express connection. Lists are available in the tabs at the left—but we saw underutilized space on the user profile to directly show Twitter users of their presence on this community platform.
I added a new space below the "You might like" to display the lists that users are a part of. Highlighting these groupings of users is also a key discovery tool—special in that it is user-curated.
This case study brought to our attention a lot of facets of Twitter interaction: what algorithmic choices at one endpoint can do in influencing user desires and actions, the necessity of putting up safety measures when increasing visibility, and also valuing certain metrics over others—in our case, follow relationships.
Navigating engagement and performance is tricky. But what makes Twitter incredible is all these untapped features: DMs turning global conversation into intimate connections, to recognizing the jokes and also controversies from quote tweets. Taking the timeline to another level was interesting when recognizing that conversation is more than just in its 280-character purity: simple clarifications on the interface and understanding of user behaviors has the potential to create huge ripples in engagement.
Better Engagements is also a take on control over communities. The tweets we engage with are very telling—it's a common joke for a reason that "to know someone truly you should look through their Twitter likes."