execution · standard
How would you measure the success of Facebook Marketplace?
How would you measure the success of Facebook Marketplace?
This question sorts candidates who understand two-sided marketplaces from those who paste a generic funnel onto a product they half-remember. The tell is usually three-fold: leading with DAU, listing 10 metrics with no hierarchy, and ignoring how Meta actually makes money from Marketplace.
Clarify first, out loud
Before naming a single metric, state the product stage: “Marketplace has over a billion monthly active users globally, so this is a retention and monetization problem, not an acquisition problem. I’ll frame the metrics accordingly. Does that match the scope you have in mind?” That one move signals product stage awareness and stops the amateur reflex of treating every metrics question as a growth problem.
Marketplace’s job: make local commerce frictionless enough that buyers and sellers choose it over Craigslist, OfferUp, or eBay, while Meta monetizes that attention through ads and, secondarily, shipping fees.
Structure a strong answer
strong
North star: successful transactions per active user per month. A transaction requires a willing buyer and a willing seller, so this single number captures both sides. It correlates directly to GMV and to ad revenue from listing views along the funnel. DAU and MAU tell you nothing Marketplace-specific; those reflect Facebook's health, not Marketplace's.
Seller-side primary metrics: median time-to-sale by category; active listing rate (sellers with at least one live listing divided by sellers who listed in the past 90 days); re-listing rate (a seller who re-lists trusts that the platform will eventually clear their inventory).
Buyer-side primary metrics: search-to-inquiry conversion rate; inquiry-to-transaction conversion rate; buyer repeat transaction rate in 30/60/90-day cohorts.
Marketplace health: listing-to-sale ratio (inventory liquidity); median response time between buyer and seller. That last one changed in March 2026 when Meta launched AI auto-replies: sellers can now configure Meta AI to respond to buyer messages from listing details. The metric still matters, but you need to split it by AI-handled vs. manually handled responses and watch whether AI replies actually close transactions or just delay them.
Secondary / diagnostic: AI-assist adoption rate (share of listings created with the AI image-to-listing tool); AI auto-reply-to-transaction conversion vs. manual reply; geographic liquidity ratio (active buyers to active sellers within a 25-mile radius, because sparse markets have a supply problem, not a product problem); GMV concentration by category to flag fragility.
Guardrail metrics: scam and fraud report rate per 1,000 transactions; false-positive listing removal rate; transaction dispute rate. Financial scams on Marketplace surged 340% in Q2 2025. That makes the scam rate a ceiling constraint on every other metric: if it rises, listing volume and GMV numbers are meaningless because the product is burning trust faster than it is generating transactions.
Business metrics (what Meta finance tracks): ad revenue per Marketplace session (the real monetization lever, since Marketplace is primarily an advertising play); shipping label attach rate (the fee-based revenue layer); 90-day seller retention cohort.
Trade-off to name explicitly: time-to-sale and inquiry volume can conflict. A seller who closes in two hours via one message generates less ad exposure than one fielding 20 inquiries over three days. The product has a strategic choice between marketplace efficiency and ad inventory depth. Name that tension; most candidates miss it entirely.
weak
"I'd track DAU, MAU, number of listings, and NPS." This answer has no north star, no two-sided framing, no connection to how Meta makes money, and no guardrail category despite trust being the product's most material risk in 2025 and 2026. Listing 10 metrics with no hierarchy reads as brainstorming, not measurement thinking. Treating Marketplace as a simple e-commerce product ignores that supply (sellers) and demand (buyers) need to be managed independently, and that Meta's financial model is ad-CPM-first, not transaction-fee-first.
The 2026 layer
The viability question for Marketplace has shifted. It is no longer “can we get enough buyers and sellers?” It is “can we make local commerce trusted enough and AI-frictionless enough to hold GMV at scale against category-specific competitors?” Viable means proving that the GMV generated justifies Meta’s investment when the ad business would monetize that same attention regardless of whether a transaction ever completes. Truly usable means a seller uploads a photo, Meta AI generates the listing with a competitive price suggestion, and AI handles buyer inquiries. The seller shows up to hand off the item. A buyer finds what they need locally, sees a verified seller profile, and completes a transaction without anxiety about being scammed.
The leading indicators that reflect this: AI-assist adoption rate, AI-reply-to-transaction conversion, scam rate per 1,000 transactions, and fulfilled-transaction satisfaction score. The old “listing count and DAU” dashboard is a lagging view of a product that has already moved on.