What do you actually ask a link seller before you send the money?

Sorry if this is a beginner question for this section.

Background: I've bought four links this year. Two went live and are still there, one went live and vanished after about six weeks, and one guy took $150 and sent me a spreadsheet of "pending placements" that never became anything. I know that's a cheap lesson compared to what some of you have paid, but I'd rather not pay it again.

The thing is I don't really know what to ask. When I say "can you send examples" they send examples, and the examples always look fine, and then it goes wrong anyway. So what's the actual question that separates the two? Assume I'm not going to be able to tell a good site from a bad one by looking.

The examples always look fine because they're the examples. Ask differently.

1. "Send me a placement you did at least nine months ago." Not a sample, a specific old one. Anyone recycling other people's work or churning links off sites they don't control cannot produce this, because their placements don't survive. This one question kills most of them.
2. "Who owns the site?" Not because you'll verify it — because a broker who is honest will say "I don't, I place through the owner" and a broker who is lying will say he owns it and then get the details wrong two messages later.
3. "What happens if it comes down in six months?" Get the answer in writing before money moves. "Replaced free" is a real answer, silence is a real answer too.

I sell domains and hosting and I'd give you the same three questions to point at me.

IME the tell is response latency on a technical question.

Ask whether the link will be in the body content or a sidebar module, and whether the page is in the index. Someone who actually manages the site answers in a sentence. A reseller three layers deep goes quiet for a day and comes back with something vague, because he had to go and ask.

You asked for the question that separates them. I don't think there is one, but there is a sequence, and it's cheap to run. This is what I do before anything above about $200.

1. Establish what you are buying

There are four products sold under the word "link" and the price ranges overlap, which is how people get confused. A guest post on a site the seller owns. A guest post on a site the seller has a relationship with. An insertion into an existing page. A rental. The failure modes are completely different — rentals disappear when you stop paying, which is honest; insertions disappear when the site owner does a content audit, which isn't anyone's fault; owned-site guest posts disappear when the seller sells the site.

Make him name which one it is in writing. Half the disputes I've watched came from the buyer thinking he'd bought permanence and the seller thinking he'd sold twelve months.

2. Check the page, not the domain

Domain metrics tell you what the seller has invested in his own inventory. The page your link goes on is the thing that has to be crawled, indexed, and read by something. Two checks: is the target page in the index today, and does the page have any inbound links of its own. A brand new post on an aged domain, with nothing pointing at it, sitting five clicks from the homepage, is a link into a hole.

3. Ask for the old placement

byte_broker's first question above is the strongest one on this thread and I'd put it before mine. Nine months is a good number because it survives one content audit and one ownership change.

4. Pay in a way that produces a record

Not because you'll get the money back — you won't, and anyone selling you links is not going to be responsive to a dispute process. Because the existence of a record changes behaviour before the fact. Sellers who intend to deliver do not care how you pay. Sellers who don't will steer you, firmly, to the irreversible option, and they'll do it in the first three messages.

5. Buy one

Whatever the volume discount is, it is cheaper than being wrong at scale. Buy one, wait six weeks, see if it's still there and whether the page is still indexed. Then buy the ten.

I'd flag that I mostly buy in two niches and my sample is a few dozen placements a year, not hundreds. Someone doing this at agency volume may have a better read on the pricing questions than I do.

Short version: make him prove an old placement, make him name the product, and never buy the batch first.

All correct, and worth saying that none of it is new. I was running the "show me something from last year" test on directory submitters in about 2011 and it worked for the same reason it works now — churn is the one thing you can't fake retroactively.

The difference is that the sellers used to be bad at it and now some of them keep a small stable of genuinely old, genuinely fine placements specifically to pass this test. Doesn't mean don't run it. Means don't stop there.

This is more than I expected, thank you. byte_broker, the nine month question is the one I'm taking — I asked the guy who took my $150 for an old placement just now out of curiosity and he's already given me two reasons why he can't.

hexforge, the four products thing explains the one that vanished at six weeks. I think I bought a rental without knowing it was a rental. PipelineRat, thank you as well.

That's the likeliest explanation, and for what it's worth a rental at that price isn't a scam, it's a badly explained product.

One addition to my own list since hexforge said buy one first — also buy one cheap one first. New seller, smallest thing he sells. You're not buying a link, you're buying information about whether he does what he says.