prioritization · hard
Prioritize Facebook Live features: a worked PM interview answer
You are the PM for Facebook Live. What features do you prioritize?
The question tests one thing: can you prioritize in a specific direction and defend why, or do you just list features? Most candidates list features. The interviewer at Meta is watching for structural judgment, not creativity.
What makes this question hard
Facebook Live runs on the social graph, not a follow/subscribe graph. That single architectural fact changes everything. On TikTok LIVE or Twitch, a new creator can reach a topic-interested audience immediately. On Facebook Live, your audience is friends and friends-of-friends of the streamer. That caps the organic ceiling for anyone without an existing social audience and creates the ghost-stream problem: a meaningful share of Facebook Live sessions have single-digit concurrent viewers. The creator goes live, nobody shows up, and they never go live again.
In 2026, creator supply is no longer the bottleneck. Meta’s Creator Fast-Track Program (launched January 2026) provides guaranteed monthly payments to qualifying creators. The bottleneck is viewer-side quality and discovery. Any prioritization that leads with more creator tools is optimizing for the top 5% of creators who already have audiences and ignoring the 95% who churn after one stream.
The north star to name (and defend)
Concurrent viewers per stream session, not total minutes watched per day. Total minutes is dominated by a handful of large streams and masks the ghost-stream problem entirely. A healthier north star for the current state of the product: what share of live sessions have at least 10 concurrent viewers? A secondary metric worth tracking is creator 30-day return rate: did they go live again within a month?
Structure a strong answer
strong
"Before I prioritize features, I want to name the constraint: Facebook Live's social-graph architecture means discovery is structurally harder than on TikTok or Twitch. Meta's Fast-Track Program has solved creator supply. The problem now is the ghost-stream: most sessions have very few viewers, so new creators don't return. My north star is concurrent viewers per session, not total minutes watched.
Given that, here is how I would prioritize. First: topic-signal discovery injection. Surface live content to relevant non-friend audiences based on content signals, not social proximity. This is the highest-impact intervention because it addresses root cause, not symptoms. Second: Top Fan live alerts. Facebook already has engagement-rank data. Notifying a creator's most engaged followers the moment they go live, before the stream goes cold, costs almost nothing to ship and directly improves minimum viable audience size. Third: one-tap Reels clip generation. Meta deprecated 30-day VOD storage in February 2025 as a signal that live-to-short-form is the strategic direction. A frictionless highlight clip flow gives creators a reason to go live even if the live audience is small, because the content lives on. Fourth: Stars onboarding in the pre-live setup flow. Stars and Subscriptions are already in market. The problem is likely setup completion, not feature availability.
What I would explicitly deprioritize: AI auto-captions, background blur, and real-time translation. These are table-stakes infrastructure items in 2026, not PM priorities. Live With (multi-guest co-streaming) already exists. Adding a 15th creative filter does nothing for a creator with three viewers."
weak
"I'd use RICE to score: filters and AR effects (high reach, high impact), multi-guest streaming (medium reach), Stars tipping (high impact). I'd also add auto-captions for accessibility." This lists features that already exist, treats AI captions as a differentiating build in 2026, skips the ghost-stream problem entirely, and never names why Facebook Live's discovery constraint is structurally different from competitors.
The PM judgment
The interviewer is testing whether you understand the structural constraint before reaching for the feature list. Two sharper tests they will run on any answer:
- “Why isn’t total minutes watched per day the right north star?” If you named it, you need to defend why it misleads. If you didn’t name it, you need to explain why concurrent viewers per session is better.
- “Features 1 and 2 are platform-level distribution work, not creator tools. How do you get that prioritized?” The correct answer is that they have higher expected value than creator tools for the ghost-stream problem, and you’d support that case with data on session abandonment rates and creator return rates.
The tradeoff to make explicit: distribution-side fixes feel like platform infrastructure, which makes them politically harder to ship than another creative tool. A strong candidate names this tension and argues for the harder path. A weak candidate defaults to the easier path because it sounds like product work.