Discord PM interview: community trust at IPO scale

Discord filed a confidential S-1 with the SEC in January 2026, with Goldman Sachs and JPMorgan Chase as underwriters. That single fact reframes what the PM interview is measuring. The company has $561M in 2025 revenue, 7.3M Nitro subscribers, and 4 billion server conversations happening every day. It also has a community of users who will leave the moment the platform feels extractive. A Discord PM sits precisely at the intersection of viable and lovable: viable means proving that Quests, Orbs, and in-game commerce can scale without a user revolt; lovable means preserving the feeling that Discord is yours, your server, your community, your identity. Candidates who treat this like a generic social app exercise fail. Candidates who understand the three-sided community dynamic (server owner, member, platform) and can design revenue features that strengthen rather than extract from that dynamic get offers.

The seven conversations

Discord’s loop spans seven conversations.

Recruiter screen (30 min). Standard background and motivation pass. The recruiter is checking for genuine familiarity with Discord’s product surface, not just that you use it. Know the difference between a server, a channel, and a thread. Have a specific answer for “why Discord” that references the actual business model, not just “I love the product.”

Hiring manager screen (45-60 min). Conversational but directional. Expect questions about your experience with community products, how you think about competing user needs within a single surface, and what autonomy looks like in your work. Discord values mastery and purpose alongside compassion and playfulness. Candidates who describe themselves as process-heavy program managers get screened out here.

Data and analytics round. Discord is metrics-first, but the right metrics are not DAU or MAU by default. Discord’s meaningful engagement signals live at the server level: active member retention week-over-week within a server, voice session duration in gaming-context servers, message depth within a channel relative to broadcast announcements. Expect a metric decomposition question anchored to a real surface: Nitro trial-to-paid conversion, server retention by size cohort, or Quest completion rates by game genre. A weak candidate recites the HEART framework. A strong candidate names why “engagement” means something different for a 50-member server versus a 5,000-member esports community, and connects that to Discord’s business outcome.

Product design round. Tests user research instincts and how you scope a feature. Discord’s design signal: they want candidates who design for the three-sided community dynamic (server owner, member, lurker or mod) rather than for the individual user. A feature that improves individual DM search without touching community dynamics is a Slack feature, not a Discord feature.

Product sense round. The core interview. Covered in detail below with a worked example.

Leadership and behavioral round. Discord’s stated values are autonomy, mastery, purpose, and compassion. Playfulness and empathy are consistent cultural signals across the company. The behavioral round tests whether you operate well in a high-autonomy environment, have a genuine point of view on product decisions, and have navigated a moment where community need conflicted with a short-term business metric. STAR delivery is fine; pre-packaged answers are not.

Case study presentation. A take-home or in-session case presented to a panel. The prompt is usually an open-ended product or strategy question about Discord’s business. Preparation on current product bets (Quests, Orbs, Social SDK, in-game commerce) pays off concretely here. Interviewers will probe whether your answer changes when they add the constraint: “this has to work for an esports server in Brazil and a study group in Japan, not just English-speaking gaming communities.”

The distinctive signal: viable/lovable at IPO scale

Discord’s challenge is not feasibility. The infrastructure, Social SDK, and developer ecosystem mean almost anything can be built. Unity integration, Meta Horizon account linking, cross-platform voice through a Rust partnership: the technical surface is already there. The hard question a Discord PM must answer in every design exercise is whether a feature serves the server owner and the lurker and the mod, or just the metric the business needs right now.

The product facts worth knowing cold before the interview:

Nitro subscriptions represent 54% of $561M in 2025 revenue (roughly 7.3M subscribers). Nitro’s health is the financial signal interviewers care about most heading into the IPO.
Discord Shop digital items account for $123.47M (22% of revenue). The Orbs system, launched globally in July 2025, drove a 16x increase in first-time Shop purchasers during a 7-week pilot, with 79% of those buyers new to the Shop.
Quests advertising (Sponsored Quests, Video Quests on mobile, Arena Quests launched October 2025) achieves a 96% median ad completion rate. That number is a community-design result: Quests are opt-in, tied to games users are already playing, and the reward is something users want inside Discord. Interviewers will ask you to explain why the completion rate is so high, and the answer needs to go beyond “good targeting.”
In-game commerce launched in late 2025 with Marvel Rivals as the initial partner, letting players buy and gift in-game items directly from chat.
Discord’s creator monetization cut is 10%, versus YouTube’s 30% and Twitch’s up to 50%. This is a deliberate community-trust bet, not a pricing error. A strong candidate can explain the mechanism: lower take rates attract more creators, which deepens server content density, which drives Nitro upgrades. Interviewers notice when candidates name the compounding logic.
93% of Discord users play games; 75% of all active servers are gaming communities. But Discord is actively expanding beyond gaming. A PM candidate must hold both truths: the gaming identity is load-bearing for community trust, and diversification is necessary for the IPO story.

Product sense round: what the question is actually testing

The most common prompt is some version of “Design a new feature for Discord.” Here is what separates the answers.

strong

"I'd anchor on a specific segment where Discord has a real unmet need. Mid-size gaming communities of 500 to 5,000 members are interesting because the server owner is usually a volunteer, not a company, and the mod team is small and unpaid. Their core problem is retention: they invest significant effort building a community and have almost no visibility into whether it is healthy or declining until members are already gone. I'd design a Server Health Dashboard that gives mods week-over-week active member retention, message depth by channel, and a signal for members who are going quiet before they leave. This serves mods directly, increases server stickiness for Discord, and creates a natural upsell surface for Nitro features like enhanced analytics or automated re-engagement messages. Success at three levels: server-level (active member retention week-over-week), platform-level (Nitro conversion from mod users who engage with the dashboard), and community-trust-level (server owner NPS or willingness to share server data voluntarily, which is a revealed-preference signal that they don't feel surveilled). The key tradeoff to name explicitly: mods get power, but Discord cannot make them feel like they are working for Discord. The dashboard must feel like a tool for their community, not a retention lever for the platform. And the feature needs to work for an esports server in Brazil and a study-group server in Japan without requiring bespoke localization of every data label, which means the underlying metrics must be universal and the UI must be adaptable from day one."

weak

"I'd improve DM search so users can find old messages faster." This is the most common failure mode. It treats Discord like a messaging app and ignores the community layer entirely. A second failure: proposing engagement features (streaks, notifications, activity scores) without acknowledging Discord's known allergy to dark patterns. The community will migrate if Discord feels manipulative, and interviewers know this from watching it happen to competitors. A third failure: "Add an AI assistant to servers" with no thought about how it interacts with server culture, moderation norms, or the server owner's authority. Mentioning AI without specificity reads as template thinking. Interviewers at Discord have heard the generic AI suggestion dozens of times.

Metrics: what engagement actually means at Discord

Discord’s engagement is not measured the same way as a feed-based social app. The useful signals are server-level first: retention of active members week-over-week within a server, voice session frequency and duration in gaming-context servers, depth of threaded conversation relative to broadcast announcements. Platform-level, Nitro’s 54% share of revenue means subscriber retention and conversion are the financial health indicators heading into the IPO. Shop revenue is growing, and Orbs is the mechanism driving new-buyer acquisition. In a metrics decomposition question, frame your answer at the server level first, then connect it to the platform-level outcome. Candidates who start at DAU/MAU have signaled that they haven’t thought about what makes Discord structurally different from other social platforms.

AI/ML fluency: what Discord actually expects

Discord does not require PM candidates to have engineering backgrounds. But the AI fluency bar is specific. Discord uses ML for content moderation, server recommendations, and Quest targeting. A PM does not own the model logic. A PM owns the input parameters, the edge cases the model cannot resolve, and the feedback loops that correct for systematic errors.

In practice, fluency looks like this: you can describe how a recommendation signal might differ for a new server with 50 members versus an established server with 5,000 active members, and you can name what product change (not model change) would improve the signal. You can explain why a general-purpose AI assistant inside a server might undermine the server owner’s authority if it is not scoped carefully. You can describe what “AI-generated content” means for community dynamics in a platform where identity and belonging are the core product. That level of fluency is required. Model architecture is not.

What clears the bar

Know the product surface at the level of a paying user, not a case study reader. Know the business facts heading into the IPO: revenue breakdown, Nitro subscriber count, Orbs pilot results, Quests completion rate. In the design round, anchor on community dynamics rather than individual user experience: name the tradeoff between the server owner, the mod, the lurker, and the platform’s business need. In the metrics round, name server-level signals before platform-level signals. In the case study, show that “international-first” is a design constraint that changes the feature (language, server norms, regulatory environment, payment infrastructure for in-game commerce), not a platitude added at the end.

In every round, hold the viable/lovable tension: Discord cannot extract revenue from its community without its community’s trust, and the PM’s job is to find the designs where those two things point in the same direction.

For the engagement-versus-revenue metrics tension in practice, see two metrics in conflict: engagement and revenue. For the broader 2026 framing on why lovable is now the hard constraint, see lovable, not just usable. For how to build the viability case inside a product design answer, see proving viability.