No. Facebook does not publish a public, buyer-facing Facebook Marketplace API for deal alerts, saved-search webhooks, or listing pushes. Every serious watcher polls Marketplace somehow, then filters and notifies you outside Facebook's own system.
If you run a sourcing desk (many profiles, many cities, first message wins), you are not shopping for JSON. You are shopping for scheduled detection plus filters. An "official Marketplace API" pitch for car alerts is a yellow flag until someone shows a public Facebook doc you can open.
What searchers usually want
"Facebook Marketplace API" almost always means one of three jobs:
- Deal alerts: interrupt me when a matching listing appears
- Data dumps: titles, prices, and links into a sheet or warehouse
- Auto-contact: message sellers without opening Facebook
For resellers, job 1 is the money path. Job 2 is research. Job 3 is a different product and a different risk. Crawlbench automates detection and filtering only. You still send on Messenger.
What Facebook actually exposes
Facebook's public developer surface is built around Pages, ads, and business messaging, not a consumer Marketplace inventory feed for arbitrary buyers. There is no documented endpoint that says "new private-seller Tacomas under $28k in my cities every few minutes."
That gap filled with scrapers, desktop agents, browser extensions, hosted "Marketplace APIs" (scrapers with nicer docs), and deal monitors. None of them have a blessing Facebook published for buyer alerts. They watch on a schedule.
Facebook also publishes no SLA for third-party watchers. Plan around uncertainty, not magic.
Option A: DIY script or GitHub "API wrapper"
A script fetches Marketplace results and returns rows. It feels like an API because your code calls a function.
Fits: raw data, you can own breakage, missed hours are fine.
Fails: income depends on first-hour replies. Sessions and layouts break without notice. Keeping the pipe open becomes a part-time job (crawler maintenance). Full tradeoff: scraper vs monitor.
Option B: Generic scraping APIs
Third-party services sell HTTP access to pages they fetch. You point them at Marketplace URLs and parse JSON-ish output. Vendor blogs will call this "the Marketplace API."
Fits: occasional research pulls, price studies, one-off comps.
Fails: a live desk. You still need pass/fail filters (make, model, year, exclusions), dedup, recency, and a phone interrupt. Per-request pricing also hurts when you check many profiles often.
Option C: A Marketplace-specific monitor
A monitor owns the reseller workflow: scheduled checks, structured filters, match reasons, alerts on Telegram, email, or an in-app queue. Crawlbench works this way. Fan-out runs about every 10 minutes for due profiles, with coverage rotating across anchor cities, so one metro is not promised every cycle.
Fits: time-to-first-message on inventory you actually buy.
Fails: bulk historical dumps Facebook never offered, or a requirement to self-host every request.
Tool shapes: monitoring tools list. Labor vs software when volume grows: desk operator vs monitor.
How to evaluate any "Marketplace API" pitch
Ask in plain language:
- Public Facebook doc URL, or your own polling layer?
- Stated check interval for my cities, and does coverage rotate?
- Filters before the alert, or every new row?
- Who fixes breakage when Marketplace changes?
- Auto-message sellers? (Default answer for a deal desk: no.)
Safe automation stops at detection and filtering. Messaging stays human (bot safety).
A one-week desk test
Pick one hot model across the cities you actually cover. For seven days, log:
- When you first saw the listing (or when a tool alerted you)
- Post time, if Facebook shows one
- Whether you would have messaged it
You do not need unpublished ranking rules. You need your latency baseline. If native Notify Me is late on that model, an "API" that cannot state a check interval will not save the desk.
Who should still build a scrape pipe
Build when your product is data: pricing models, research dumps, internal catalogs. Maintenance is cost of goods.
Do not build when your product is "message the seller first across dozens of searches." That desk needs a stated schedule, filters that kill junk, and an interrupt people answer. Confusing those jobs is how engineering weekends eat sourcing weeks.
Bottom line
There is no buyer-facing Facebook Marketplace API for deal alerts. Buy the outcome: raw data you maintain, or filtered alerts someone else keeps alive. Crawlbench is the second path (match reasons and scheduled coverage). Your time stays on Messenger, not on pretending an official feed exists.