When a Marketplace listing needs a sold ping, not a scrape

The job is a card you already have in the file. The price just moved, or the listing is gone. Someone still hears it from a screenshot in a group chat after the window has closed.

Teams ask for “Marketplace data.” That is a pipeline. This loop pings when a watched item sells, disappears, or crosses a price you already wrote down, then stops.

What the automation actually does

It reads snapshots that are already landing. It compares last seen, sold status, and price on the items you named. It sends one note with the listing, the old figure, and the new one. It does not open a new keyword search.

The Facebook Marketplace case study is that series: first seen, last seen, sold flag, and price history per item. AI automation is the compare-and-ping. We will not treat a buying bot or a Marketplace inbox agent as this job.

This is not deciding whether you should scrape Marketplace. Coverage is keywords, cadence, and a grain you can query. A sold ping assumes that file already updates. It is also not a property listing match. A villa brief is a new card. This job is a change on a card you already hold.

When we leave the alert with a person

If two desks still argue what “sold” means — last seen, a status flag, or a guess — software will page the wrong shift. If a person with a search tab already answers one category, an alert is early. If the useful signal is a WhatsApp voice note from a picker, write the item id first.

Kuwait and GCC desks use Marketplace for cars, furniture, and phones. Name the category on the watch list. Do not mix a Hawally furniture keyword with a US car lot.

Bring the ten listings you already refresh by hand and last week’s “it sold before we saw it” misses. Book thirty minutes or write hello@jamilglobal.com.