I was deliberate and proactive in communicating with senior leadership regarding Pegboard.

After conducting an audit of current and planned creator-owned live interactions, I presented a review of the problem space in a collaborative session with Twitch design leadership. The session was open to any design or product stakeholder interested to learn or provide input.

I kicked off the design process in mid-February with a two-day design sprint at Twitch HQ in San Francisco. In addition to generating a broad range of potential starting points, this sprint would give me an opportunity to recruit design collaborators.

After the exercise, I recruited seven designers to contribute to the project in their spare time. These designers worked in areas of Twitch that were important to Pegboard - two of them had designed the existing watch experience on web, two worked on Chat, two worked on Extensions (live interactive experiences developed by 3Ps), and one was an talented mobile designer.

In addition to a volunteer design team, I reached out and formed a partnership with the Sr. PM leading product development in the "live watch" space. He was enthusiastic about the opportunity and an invaluable collaborative partner over the following months.

To be successful, Pegboard’s design would have to balance competing priorities.

Humans are biased to assume that “what I see is all there is” (Kahneman 2012), so interactive calls-to-action in Pegboard needed to be visible by default and passively communicate the nature of the interaction in order to drive viewer engagement. However, they needed to do so without pulling focus from either the live video or the “Follow” and “Subscribe” buttons that drove Twitch’s business, or overwhelming viewers with simply more calls-to-action on an already busy page.

Pegboard needed to invite viewers to participate, and set expectations that its interactions came from the creator (as opposed to Twitch), while remaining independent of Chat and not bound to Chat visibility.

With my volunteer team in place, I led an IA analysis of Twitch's watch experience on web and mobile, along with a review of comparative “real-time interactions with live video” experiences across large- and small-screen formats. After the team had found their footing, I led a critical review of our first round of design exploration.

Following Phase 1 of design development in late April, my PM partner and I met with Twitch org leaders. We shared an overview of the project, objectives, success measurements, tenets, and some work-in-progress from Phase 1.

The work-in-progress centered around four design themes that had emerged from our process in the prior weeks:

After the meeting, org leaders granted our request to move Pegboard to the "working backward" phase of product development.

After a round of design exploration, I better understood the details and constraints that would help us design a successful system. I synthesized these learnings for the team.

Some of the creator-driven interactions in Pegboard were time-limited to 5 minutes or less, while others were persistently available for the entire stream. For temporary interactions, urgency is a feature, but in order to be appropriately foregrounded these urgent interactions needed to be visually distinct from persistent interactions.

Secondly, we needed to avoid anchoring Pegboard calls-to-action over the video content. Most Twitch content was video game content, and most video games have their own on-screen UI — in practice, it could be difficult to distinguish Twitch UI that viewers could interact with from game UI only usable by the creator. Additionally, over-video UI would occlude ad content in a way that advertisers might find problematic.

Lastly, all creator-sourced interactions should be available to all viewers. Twitch channels are a live, multiplayer environment where viewers can talk to and learn from each other. In this setting, surfacing some interactions to only some viewers would cause more confusion than clarity. In addition, the predominant culture within Twitch communities emphasized equality and the inherent value of everyone present. In this environment, hiding interactions from some users could be negatively perceived as gatekeeping.

Our next round of explorations reflected these learnings, with interactions anchored below the video in a way that was distinct from Follow and Subscribe calls-to-action but also without reducing video size.

As Pegboard was a design-led project, I was not funded for usability research on our developing solutions. However, we did benefit from foundational research insights.

Early on, I initiated a discussion with a senior researcher regarding Pegboard, its problem space, and the kinds of understanding that would most benefit our work. I was most interested in the predominant mental models that viewers shared regarding creator-driven interactions, and any common problems viewers encounter with existing interactions. As it happened, there were two arcs of research that might shed light on these questions, and by late July we had our insights.

These insights seemed to validate of many of our prior design decisions. The choice to ground Pegboard in the context of the live video aligned with viewer mental models of creator-driven interactions as "on-stage" entertainment. Furthermore, the utility of timely creator-driven interactions as stream metadata ("what's happening now") gave us further reason to not filter interaction availability or visibility.

Meanwhile, the way that interaction relevance was dependent on a viewer's relationship to channel revealed a potential opportunity. Instead of filtering interactions, we could use common markers of channel-relationship strength as a predictor of relevance, and weight the way we displayed interactions on a per-viewer basis — essentially personalization via prioritization.

My next round of explorations reflected these learnings, separating interactions by persistence and timeliness and experimenting with personalized weighting.

In mid September, my PM partner and I had our second meeting with Twitch CEO Emmett Shear. We requested approval to move forward with new UX research and potentially live experiments.

In this presentation, we focused on how elements of prioritization could help bridge the gap between states in the Twitch viewer journey. The presentation included a short doc accompanied by high-fidelity prototypes.

Prior to the meeting, I met individually with the Chief Product Officer and org leaders. Everyone was on board and aligned to the following experience.

"Persistence" versus "timeliness" was the primary differentiator between Pegboard interactions across web and mobile modalities. In each case, timely interactions were positioned in proximity to existing interactions (such as Follow/Subscribe and Chat) and to the most physically accessible areas (down and to the right for right-handed mobile users).

Within persistent and timely groupings, I pursued "personalization via prioritization" using four basic tranches of viewer-to-channel relationship: New-to-channel viewers, Followers, Participants, and Supporters

For each viewer, available Pegboard interactions would be prioritized right-to-left in order of relevance as shown below. Right-to-left prioritization aligned with the existing channel IA by placing newer and more relevant interactions closer to other channel-focused interactions, such as Chat.

While I was confident that the Pegboard design would strike the right balance of visibility and subtlety for most viewers, it was not uncommon for a small cadre of vocal users to tilt general customer perception of a Twitch feature. With that in mind, I designed functionality that would allow a disinterested user to minimize Pegboard interactions. Pegboard would be "pinned" open by default, but a user could choose to "unpin" and mostly conceal the interaction surface. Mousing-over the video would reveal Pegboard, and interacting would temporarily re-pin the bar for easy browsing.