product sense · hard
"Design a jobs product for Facebook"
Design a jobs product for Facebook.
The interviewer is not asking you to build LinkedIn. They are testing whether you can identify what Facebook actually has that LinkedIn does not, commit to a specific user segment, and explain why this product would be worth returning to rather than a one-time utility. The failure mode is designing a job board. The strong answer designs a trust layer built on a social graph that no competitor can replicate.
One piece of context that separates prepared candidates: Facebook launched a Jobs product in 2017, expanded it to 40+ countries, then shut it down in 2022-2023 after settling civil rights lawsuits over discriminatory ad targeting in hiring. Meta relaunched it in the US in October 2025, explicitly targeting entry-level, trade, and service positions. Naming this history before the interviewer brings it up changes the entire tone of the conversation.
Structure a strong answer
Open with a scoping question, then commit to a segment fast. The segment choice is where the answer wins or loses.
strong
"Before I go wide, a quick scope question: are we building from scratch or layering onto the October 2025 relaunch? And is there a segment you want me to optimize for, or should I find the biggest unmet opportunity? I'll assume the latter. The biggest gap is hourly and trade workers: electricians, warehouse staff, restaurant workers, home health aides. 80 million-plus in the US. They are on Facebook every day. LinkedIn does not serve them. Indeed is high-friction (separate account, resume required, often no response). Their job to be done is local, schedule-fit work within the next two to four weeks, from an employer they can vet without taking a flyer on a stranger. That vetting problem is where Facebook's social graph becomes the core mechanic, not a feature. A user sees a job posting at a local restaurant. Their friend Maria works there and her profile is visible on the employer's page. Maria can confirm the place is worth working at. That trusted signal reduces adverse selection on both sides: the worker doesn't waste a commute, the employer gets fewer ghost applications. No job board can replicate this because no job board has the social graph. That is the actual moat. Three prioritized features for MVP: first, social-proof job cards, listings that surface visible mutual connections at that employer with an opt-in Messenger nudge to ask a connection for a reference. Build this first because it is the differentiated feature and without it you have built Indeed. Second, AI-generated job posts from Business Pages: an SMB owner can generate a legally-compliant job description from their existing Facebook Business Page in two taps, solving the cold-start supply problem. In 2026, this costs almost nothing to build. Third, conversational apply via Messenger: a five-question structured conversation covering availability, location, and basic fit, in the channel the applicant already uses daily. No resume required, because most hourly workers don't have a polished one. The employer gets a summary card. On anti-discrimination: Facebook was sued for this in 2019 and shut the product down. That is a first-class design decision, not a legal footnote. MVP ships with standardized screening questions (same questions to every applicant), no age or photo in initial match, and an algorithmic audit for disparate impact before any targeting feature goes live. North star metric: verified hires per week, confirmed by employer within 30 days of first Messenger contact. Leading indicators: share of applications that include a mutual-connection assist (quality signal), Messenger conversation completion rate (intent signal). Guardrail: discrimination complaint rate versus the 2023 baseline. The viable question in 2026 is whether SMBs will pay for promoted listings. They already buy Facebook Ads to reach customers; the sales motion is the same account manager and the same budget line. Lovable means the job seeker never feels like they left Facebook to use a job board. The trust layer, the Messenger flow, and the social proof all live inside the product they already use."
weak
"Facebook's mission is to bring the world closer together. The users are job seekers, employers, recruiters, students, and career changers. I'll focus on job seekers because they have the highest need. I'd build a job listings feed with an easy-apply button, in-app messaging to contact employers, and profile integration so your work history is pre-filled. Metrics would be number of jobs posted and number of applications submitted." This answer describes LinkedIn in 2015. There is no social graph mechanic doing anything specific, no awareness that Facebook tried this and was sued for discrimination violations, no explanation of why an hourly worker would use this over Indeed, and no metric that distinguishes between a submitted application and an actual hire. An interviewer at Meta ends this conversation in their head about three sentences in.
Why the segment choice matters
The intuitive answer is “job seekers broadly.” The strong answer names a segment whose pain is acute, whose existing alternatives are genuinely poor, and whose behavior already fits the platform. Hourly and trade workers clear all three tests. They are on Facebook daily (often more than once). They do not have polished LinkedIn profiles and LinkedIn’s product is not built for them. Indeed serves them somewhat but requires a separate account and produces a low-signal application experience for the employer. The segment also has a clear viable story: SMB employers in service industries already have Facebook Business Pages and Facebook ad accounts. The incremental cost to them of trying a promoted job listing is close to zero.
The segment to avoid anchoring on: knowledge workers and recent college graduates. These users have LinkedIn. The competitive dynamic is harder, Facebook’s social graph is less differentiated for professional credentialing, and the product would need to win on features, not trust, which is a much harder fight.
The discrimination constraint as a design decision
Most candidates mention discrimination as a risk. Strong candidates name specific product decisions that address it. The 2019 EEOC settlement (where Facebook settled with civil rights organizations over ad-targeting that let employers exclude applicants by protected characteristics) was a product design failure, not a policy failure. The fix is not a disclaimer; it is removing the targeting levers that allowed exclusion. Standardized screening questions prevent anyone from designing around protected characteristics. Equal-access enforcement means a job post reaches all eligible users in a geographic radius, not a filtered subset. Algorithmic audits catch disparate impact before it compounds. Naming these specifically signals that the candidate has thought about the product’s actual failure mode, not just the legal footnote.
North star metric and the leading indicators
“Number of jobs posted” and “number of applications submitted” are vanity metrics. They measure supply and activity, not outcomes. The north star is a confirmed hire: employer verifies in the product that they hired someone within 30 days of the first Messenger contact. This is hard to game and directly measures whether the product is solving the actual problem.
The leading indicators matter because confirmed hires are a lagging signal. Mutual-connection-assisted applications as a share of total applications tells you whether the social graph mechanic is doing anything. If this number is near zero, the differentiating feature is not working and you have built a job board. Messenger conversation completion rate tells you whether job seekers are engaging or dropping off before the employer even sees a card. If completion rate is low, the five-question flow is too long or the wrong questions are being asked.
The guardrail that most candidates omit: discrimination complaint rate. Given the product’s history, this needs to be a monitored metric with a defined response threshold, not a retrospective audit.
The 2026 reframe
Feasibility is near-free here. An LLM can generate a job description from a Business Page in seconds, screen applicants conversationally via Messenger, and match on nuanced fit signals at a cost that rounds to zero at the scale of Meta’s infrastructure. That eliminates the engineering constraint that made this product hard to justify building carefully in 2018. The real question is viable and lovable. Viable: there is an 80 million-plus US hourly workforce with urgent recurring need, and their employers already have Facebook ad accounts. Lovable: a job-finding experience that does not make the worker feel like they left their social world to fill out a form. The social proof layer, the Messenger-native apply, and the mutual-connection trust signal are what make it worth returning to rather than a one-time utility. The answer is not “build a job board with Facebook login.” It is “use Facebook’s trust infrastructure to reduce the information asymmetry that makes hourly hiring unreliable for small businesses and stressful for workers.”
For more on how Meta evaluates product sense, see the Meta interview process. On the viable/lovable lens that applies across all Meta product questions, see feasibility is free and lovable, not just usable.