How to vet a link seller before you pay them
I vetted three link sellers on spam score. 16%, 26%, 44%. Then out of curiosity I ran the same check on my own domain.
48%.
My site scored worse than every seller I’d been sizing up, including the one I’d nearly walked away from. That number changed how I use these tools, so it’s where I’ll start.
The authority score is the wrong first question
Ask a seller for numbers and the first thing they send is an authority score, some 0 to 100 figure from whichever tool they subscribe to. It’s the least useful number in the conversation.
I have three purchases in front of me with ranks of 307, 210 and 464. The best of those belonged to the placement that took $150 and never published. The worst shipped clean, dofollow, correct anchor, from a site with nearly 8,000 referring domains.
The score predicted nothing. It’s also the number most easily inflated, which is exactly why it leads every pitch.
The four numbers worth pulling
You can get all of these from a backlinks API for roughly $0.05 a domain:
- referring domains (distinct sites linking to them)
- total backlinks (a different, much larger number)
- spam score of their own backlink profile
- what topics those referring domains sit in
Five cents. I labour the point because people skip vetting on cost, and the cost is nothing. What they’re actually skipping is the ten minutes of thinking afterwards.
What a spam score measures
It estimates how much of what points at them looks automated, low-quality or bought. That’s the entire scope. People read it as a verdict on the seller’s honesty, or as a penalty-risk score, and it’s neither.
A high number could mean they bought links badly years ago. It could mean they’re a large site that automatically attracts scrapers and directory clones, which happens to everyone at scale. Or it could mean they run a link farm.
The number can’t distinguish between those, and nobody mentions that when they quote it at you.
The 44% I bought anyway
One came back at 44%, comfortably the dirtiest of the three, and going in it felt like the safest purchase on the list. Real company, real product, an audience I wanted, a name you’d recognise.
I bought it. I’d make the same call again.
What I wouldn’t do is make it without seeing the number. Treat vetting as pricing information rather than as a gate. A 40% site should cost you less than a 15% site. That’s the correct response. Refusing to transact above an arbitrary threshold just means you buy fewer links at worse prices from sellers who’ve learned to game the metric.
And if I’d applied a 30% cutoff, I’d have rejected myself at 48%, along with two of the three placements that turned out fine.
My own number was high because that site has been online long enough to accumulate the usual sediment of scraper sites, aggregator clones and directory spam. I hadn’t bought a single bad link. It just gathered.
Referring domains versus backlink count
A seller will quote you 677,100 backlinks and it sounds enormous. It usually isn’t.
That particular site had 7,800 referring domains behind those 677,100 links. Normal ratio. One site links to you from many pages, so raw link count inflates fast and means very little.
Referring domains is the number with signal. If someone leads with total backlinks and gets vague when you ask for referring domains, the gap is your answer.
The relevance check no tool does
No API will tell you whether a link makes sense. Open the page they’re proposing and read it as a visitor rather than as a buyer. Would somebody reading this genuinely want what you sell? If the answer needs a paragraph of justification, it’s a no.
My two that shipped went onto a page about automation tooling and a page about IP lookups. I sell mobile proxies. I didn’t have to construct an argument for either.
Relevance has survived every algorithm update anyone’s written about, and it’s the factor vendors quietly ignore, because their inventory is their inventory and they’ll place you wherever there’s a slot this week. Name the constraint in the order and be willing to wait.
Read the page’s outbound links
Before buying onto a page, look at what else it already links out to.
If it carries eleven dofollow links to gambling sites, crypto exchanges and essay mills, you’re buying a spot in a link farm regardless of domain metrics. One commercial link surrounded by editorial content is a completely different purchase from the twelfth commercial link in a row.
Takes a minute. No tool required.
Payment terms are a signal
I sell placements as well as buy them, so I see both sides.
On my sell side right now: one buyer paid within two days by invoice and the link went live that week. Another has been “awaiting transaction hash” for several weeks after asking for crypto terms. A third requested a quote in June and has never sent back a target URL.
Read those in reverse as a buyer. A seller insisting on crypto up front, no invoice, no company name, has removed every mechanism you’d use to recover money. That doesn’t make them dishonest. It does mean delivery risk is entirely yours, and you should price accordingly.
Because sometimes it doesn’t arrive. 39% of one of my purchase batches was money paid for a listing that still wasn’t live eight days later.
So ask what happens if it never publishes. You want a refund window or a replacement of equivalent value. You’ll usually get a friendly reassurance that this never happens, which isn’t a term and can’t be enforced. Small publishers often have no refund process because they’ve never needed one. Avoid them over that and you’d cut out most of the good inventory, so keep individual orders small until someone has delivered for you once.
I now treat the first purchase from any new seller as a test order regardless of how good the metrics look. Smaller placement, lower price, no expectations.
The terms to write down before money moves
Vetting the seller is half of it. The other half is vetting the deal, and that’s the part done over chat and then forgotten.
When I sell, the terms I put in writing are the ones I’d want as a buyer: how long the link stays up, whether the article can be rewritten or reordered in that time, where in the piece the link sits, and what the anchor says.
A real example from my sell side. One buyer paid $250 for a dofollow placement on a comparison article. Terms were twelve months, no rotation, and a specific position in the ranking, second place, in both the list and the summary table. That last clause matters more than people realise, because comparison articles get rewritten as the market moves, and without it a paid position quietly drifts to seventh on the next refresh.
A seller who won’t put duration, rewrite policy and position in an email is telling you the placement is disposable.
The first ten days after paying
Set a reminder about ten days out and check three things.
The page exists at the promised URL. Your link carries no nofollow or sponsored attribute, which takes thirty seconds in the page source. And the anchor matches what you ordered, because when an agency’s writer builds the article around your link, the anchor is what most often comes back as your brand name or something bland instead.
Raise any of those immediately, while the writer still has the draft open. A month later you’re asking a stranger to edit a published page as a favour.
What I’ve stopped bothering with
I no longer care about the authority score beyond a sanity glance. I don’t ask for traffic screenshots, because they’re trivially faked and everyone knows it. And I stopped negotiating hard on price with small publishers, because the ones worth buying from aren’t short of buyers and the discount isn’t worth the relationship.
What I always do is the five-cent data pull, the one-minute look at the page’s existing outbound links, and the renewal email.
More on what those purchases actually returned is here.