Bringing casual games to an established communication platform

SUMMARY

Integrated casual games into Discord's voice features, boosting user engagement and ARR. Overcame technical hurdles to launch Activities, paving the way for new partnerships in casual gaming.

ROLE

Product Designer

YEAR

2021 - Present

TEAM

Product Designer, 2 PM, 4 Engineers

RESPONSIBILITIES

Customer interviews, UX Research + data collection, Spec creation/collaboration, Design prioritization and strategy, Running multiple team meetings.

Introduction

Upon joining Discord, our mission was clear and ambitious: to integrate casual games into Discord's voice features and create a two-sided marketplace for developers. The goal was to emulate the success of Discord's bot ecosystem, but with a focus on casual gaming. This is the story of how our team navigated this transformative journey, from initial research to global launch.

Goals

We set ambitious goals to guide our path:

Research

Before diving into design, our team conducted extensive user research. Collaborating with industry experts, we held weekly design sessions, using data as our guiding star. Our collective efforts revealed that over 50% of Discord users were willing to pay between $2 and $5 for the new feature, validating our direction and giving us the green light to proceed.

Challenges and Opportunities

Even with really positive feedback from our research and initial tests it was no easy feat navigating through technical and operational hurdles to seamlessly integrate it into synchronous voice calls.

Challenges & Solutions

Opportunities

Hypotheses

1. Integrating Activities will increase user engagement and retention for both the voice channel feature and the app.

2. The willingness of users to pay for Activities suggests a new, potentially lucrative revenue stream.

3. Activities will deepen user connections for our key user groups and expand Discord's community.

4. First Party Activities will draw more developers to Discord, creating a vibrant the developer ecosystem, similarly to bots.

The Design Process

Early Explorations

In the early stages of this project, our first task was to immerse ourselves in Discord's existing design language. Fortunately, the organization had already invested in robust Design Systems and Sticker Sheets, enabling us to prototype at a higher fidelity than is often possible. Over the first few weeks, we focused on conceptualizing how Discord's platform could evolve to accommodate activities, particularly within the framework of a two-sided marketplace for both developers and end-users.

To ensure we were on the right track, we held feedback sessions involving product managers, engineers, and our design team. These meetings served as pivotal checkpoints, allowing us to present our ideas to leadership for validation and to identify areas requiring iteration. Interestingly, our initial designs were considered too ambitious, prompting us to pivot towards a more iterative and component-driven approach. This shift led us to explore specific questions that would guide our subsequent design efforts. All of this exploration led to some more structural questions we wanted to dive into - here are a couple of examples:

What are some of the best-case scenario primitives for how we integrate Activities into the wider product?

Like "how might we's" around notifications, Active now, and how we can represent ongoing activity presence in places like someone's profile.

Explorations around an Activities Home, Notifications inside the chat, and how a mobile activity might look.
Explorations around an Activities Home, Notifications inside the chat, and how a mobile activity might look.

What can we ship quickly to test the waters on how users will receive this change?

A huge piece of this exploration work was to make a call on how we might start testing the hypothesis that activities are A. Wanted and B. will deepen connections as folks use them. Luckily we were working with users from our previous startup, Backyard, to create a coalition of folks to come try this new thing we were working on.

Explorations around an Activities Home, Notifications inside the chat, and how a mobile activity might look.
What might some longer term primatives look like?

Initial Testing

To rigorously assess user engagement and willingness to pay, we launched targeted alpha and beta tests. These tests were deployed to 15% of total servers, specifically those with fewer than 100 members. While the tests may seem straightforward, they laid the essential groundwork for the feature set and provided critical data for future decision-making.

The Alpha: Baseline Engagement Bundle (BEB)

Our alpha test, known as the Baseline Engagement Bundle (BEB), was a resounding success. We rolled out basic functionalities and a bundle of games, focusing primarily on Voice Chat (VC) usage and secondarily on user adoption and retention rates. This really hammered home that there was a willingness to play and it could be an engagement and experience lever!

Updates to the VC and RTC Controls: We leveraged Discord's existing RTC panel, a relic from its gamer-focused days, as a strategic entry point for activities. This decision paid off: more than two-thirds of all activity starts originated from this panel. We also added an entry point in the video controls to guide new users, and both entry points have been retained due to their effectiveness.

Explorations around an Activities Home, Notifications inside the chat, and how a mobile activity might look.
The simplified activity shelf launched from control UI
Explorations around an Activities Home, Notifications inside the chat, and how a mobile activity might look.
The same simplified shelf launched from the RTC panel in early versions

Updates to the Video Grid: We updated the video grid to offer a focused view, similar to how screen sharing works, allowing users to immerse themselves in the activity while still interacting with friends. This required significant cross-functional collaboration to ensure alignment with Discord's core values.

Explorations around an Activities Home, Notifications inside the chat, and how a mobile activity might look.
The modified grid with an entry point into already started activities.
Explorations around an Activities Home, Notifications inside the chat, and how a mobile activity might look.
The earliest version of the focused iframe integration

Alpha Results

The alpha test led to a 35% adoption rate, surpassing our 20% target, and a 30% week-one retention rate among servers that engaged with the new features. Additionally, 65% of surveyed users indicated a strong likelihood to continue using Activities. Importantly, we observed no negative impact on key performance indicators at the end of the first month. These data led us to move to phase 2 where we would focus in on the monetary themes we'd been exploring. We made only a few specific UI changes from here to make sure we were keeping that consistency of expectation, while zeroing in on the new data we needed to understand.

The Beta: Willingness to Pay or Boost Gating

The beta phase aimed to test users' willingness to pay for Activities. We introduced server-level perks, allowing one person's purchase to grant access to the entire server. While this approach limited the total addressable market (TAM), it facilitated rapid testing because the infrastructure was already there to integrate a feature set into the Boosting package at a server level. Ultimately, this would prove out our hypotheses that we were driving sellable value to users!

A Revenue Focused Activity Menu: Building on insights from the alpha, we designed a revenue-focused activity menu with clear labeling and messaging. We also introduced a "Rotating Free Activity" feature, accessible to all users regardless of subscription status.

Explorations around an Activities Home, Notifications inside the chat, and how a mobile activity might look.
The simplified activity shelf launched from control UI

Upsells and Quality of Life Improvements: We integrated the existing boosting payment flow and developed new upsell strategies and promotional surfaces to test willingness to pay.

Purchasing and upsell modals and coachmarks.
Purchasing and upsell modals and coachmarks.
Surveys, Server Statuses, and Happening Now. Oh My.
Surveys, Server Statuses, and Happening Now.

Beta Results

Although the beta test was generally successful, it raised some strategic questions for the global launch. We observed a 12% increase in Level 1 boosting among servers but also noted some negative impact on other metrics, such as Messages Sent. However, the test did result in a low-six-figure bump in Monthly Recurring Revenue (MRR) - which aligned with our goals... but made us curious if we could do more.

Global Launch and beyond

Leveraging the insights and successes from our alpha and beta tests, we implemented targeted design solutions to address remaining challenges and enhance user experience. We continue to use these insights to improve the platform with updates to add more content, expanding the Activities to both Mobile and G/DM surfaces, and eventually filling out some of that Viral Loops work as well.

The Evolution of the List-Based Shelf: Our team oversaw the design evolution from a small menu to a comprehensive Activity Shelf, significantly increasing user engagement. We also introduced a Mini Shelf, a streamlined, hover-activated version featuring promotional space and a recommendation engine.

The mini shelf and the activity shelf.
The Activity Shelf with a Promo spot and the updated activity shelf.

Higher Visibility Promotional Surfaces: Recognizing the untapped potential of empty states within the Voice Channel UI, we transformed these spaces into high-visibility promotional surfaces. This not only filled a visual gap but also served as a strategic location for upsells and user engagement.

Purchasing and upsell modals and coachmarks.
Empty state upsell with before friends show up..
Surveys, Server Statuses, and Happening Now. Oh My.
Empty state upsell for groups of 2 - 4 people.

From Boosting to Nitro: Recognizing the limitations of our server-based pricing strategy, we shifted to a user-based model, opening up new revenue streams and challenges that we successfully navigated.

Purchasing and upsell modals and coachmarks.
New layouts and processes for purchasing access to Activities.
Surveys, Server Statuses, and Happening Now. Oh My.
Empty state upsell for groups of 2 - 4 people.

Overall Results

Discord Activities
TechRadar Feature
Discord Activities might be its most ingenious addition yet

Read more about how Discord is innovating with its new Activities feature. This might be a game-changer!

Learnings

This project was a marathon, not a sprint. I was able to use and hone skills in stakeholder management, data-driven decision-making, and agile methodologies. The most valuable lesson was the importance of agility and strong partnerships. Collaborating closely with a like-minded engineering partner was instrumental in our success.


More of My Work

Discord: Wumpus' Eighth Birthday

Backyard: Social Gaming for Distant Friends

Envoy: Touchless Sign in and the Early Days of the Pandemic

Treehouse: Techdegree Office Hours

Mealthy: Building Software Tools for Discerning Home Cooks