Three things to gather first
Start with the address itself, copied whole rather than retyped from memory. Long addresses hide their important parts in the middle, and a single wrong character turns a working link into a dead one before you have begun. Copy the whole string, including the section that looks like padding.
Second, the date you received or saved it. The age of a link decides how much checking it needs, and a date you cannot remember is a date you cannot use when something goes wrong later. A date turns a claim into a record, and records are what disagreements run on.
Third, what you paid, if anything, and to whom. Money changes which part of this site matters to you: a paid order takes a different route from a link somebody pasted into a chat group. Sort yourself into one of those two groups before reading anything else here.
Four moves, in order
Move one: identify what you hold. A wrapped address carries an item ID and a source marker. A plain marketplace address carries the item ID in its own form and no wrapper at all. Say which one you have before doing anything to it.
Move two: rebuild. The legacy link resolver takes the addresses you paste and returns the source address behind each one, which is a better use of an afternoon than maintaining a formula by hand. Work in small batches, three or four addresses at a time, and read each result as it comes back.
Move three: check the rebuilt address before you rely on it, and write the date next to the result. Move four: decide, using the platform ledger to see which platforms are recorded as taking orders and which fields on that row are marked unverified.
The order matters more than the speed. Rebuilding before identifying gets you a page you cannot read. Deciding before checking puts weight on a figure nobody confirmed. Checking before recording leaves you with a result you cannot show anyone tomorrow, which is the same as having no result at all.
How you know it worked
A rebuilt address that answers is a page you can open and read: the item, its description, its size table and its price. That is the working state, and it is the only one worth calling finished. Half a page does not count, and neither does a landing page about something else entirely.
If the address returns nothing, you are holding a failed route rather than proof of a missing item. If it lands somewhere unrelated to the item ID, stop and re-check the address you pasted before drawing any conclusion from it.
Working also means the row is written down with today's date beside it. A result you cannot reproduce tomorrow is a memory, and replacing memories with dated readings is the whole reason this site exists. Write the address you used as well, so the next check starts from the same place.
The mistakes that cost the most
First mistake: pasting a wrapped address into a platform that takes rebuilt ones. The wrapper is the layer that failed, so an order box is the worst place to discover that. Test the address in a checker first and the mistake costs you a minute instead of an order.
Second mistake: reading a status word as a stock note. Resolves means an address answered on a date. It never meant the item was available, and it does not mean it will be available tomorrow. Those are three different questions, and only one of them was ever asked.
Third mistake: moving a whole sheet before testing a few rows. Rebuild three or four addresses first, confirm each one answers, and only then commit the rest of the list to a new platform. A sample you tested yourself is worth more than a list you hoped about.
Fourth mistake: keeping no record of what you saw. Date, address, result — three short entries per row, written while the page is still open in front of you. That habit turns a guess into something you can still defend months later, when nobody remembers which version was on screen.