How to verify an ad network’s publisher claims before you sign
A network’s pitch deck is not evidence, and neither is its published seller list on its own. But both sides of that list are public, which means anyone can test whether a network’s claimed roster is real. Here is the method, and the results.
1 August 2026 · 8 min read
Every ad network shows prospective publishers a list of impressive sites it works with. Some of those relationships are current and central; some ended two years ago; some were never what the slide implied. From the outside these look identical.
They are not identical in the public record, and the check takes about twenty minutes.
The principle
Both sides of every advertising relationship are obliged to publish. The network declares the sellers it represents in its sellers.json; the publisher declares who may sell its inventory in its ads.txt. Either document alone is one party’s assertion. Together they are an agreement, because two independent parties had to take an action for both to say the same thing.
So the question to ask about any claimed relationship is not “does the network say so” but “does the publisher say so too”. If you only remember one thing from this page, that is it. The mechanics of both files are covered in ads.txt and sellers.json explained.
The check, step by step
- Get the network’s roster. Open
thenetwork.com/sellers.json. Search it for the marquee domains you were shown. If a site the network told you about is not in its own seller file, ask why before anything else. - Note the seller id. Each entry carries the account id that publisher holds with the network. This is what you will match on.
- Check the publisher’s side. Open
thatpublisher.com/ads.txtand look for that account id. Present means both parties agree. Absent means only the network is claiming it. - Look at the page. Load the publisher’s homepage and check what ad code it actually loads. This is present-tense evidence where the files are paperwork.
- Repeat across five or six names, not just the most famous one. A single confirmed relationship tells you a network has one real publisher. The pattern across several tells you whether its roster is broadly real.
The confirmation rate for every network we track
| Network | Listed | Confirmed | Claim only | Confirmed % |
|---|---|---|---|---|
| Raptive | 6,979 | 6,663 | 150 | 95% |
| Monumetric | 668 | 637 | 18 | 95% |
| Mediavine | 8,638 | 7,853 | 191 | 91% |
| Ezoic | 7,683 | 6,896 | 451 | 90% |
| AdPushup | 997 | 814 | 177 | 82% |
| Infolinks | 882 | 627 | 233 | 71% |
| SHE Media | 1,591 | 1,110 | 474 | 70% |
| Venatus | 627 | 406 | 219 | 65% |
| Newor Media | 173 | 111 | 61 | 64% |
| Freestar | 637 | 366 | 255 | 57% |
| Primis | 192 | 104 | 79 | 54% |
| NitroPay | 759 | 400 | 355 | 53% |
| PubGalaxy | 2,599 | 1,303 | 1,271 | 50% |
| Publift | 2,090 | 1,044 | 1,034 | 50% |
| Gourmet Ads | 972 | 463 | 462 | 48% |
| Playwire | 968 | 461 | 496 | 48% |
| Adnimation | 564 | 265 | 273 | 47% |
| Snigel | 282 | 131 | 137 | 46% |
| Setupad | 876 | 403 | 466 | 46% |
| Vidazoo | 443 | 126 | 266 | 28% |
A low rate is not automatically an accusation
This matters, and articles that publish tables like the one above rarely say it. There are legitimate reasons a network’s published roster goes unconfirmed:
- Reselling. A network that sells through upstream partners may never appear under its own name in a publisher’s file. The relationship is entirely real and structurally invisible from the publisher side.
- Shared seats. Agencies and networks sometimes place several publishers under one account, so the id in the publisher’s file does not match the one published for that specific domain.
- Unreadable publisher files. Some sites serve no ads.txt, or serve it in a way crawlers cannot read. Absence of evidence, not evidence of absence.
- Stale entries. The publisher left and the network has not pruned the file. This one is a problem, and it is the most common.
Which is why the table is a prompt for a question rather than a verdict. A network at the low end is not necessarily misrepresenting anything — but it is a network whose published roster you should not take at face value, and it is entirely fair to ask it which of the four explanations applies.
What to ask a network directly
Three questions, all reasonable, and the quality of the answers is informative regardless of the numbers:
- “Are the publishers on your deck currently live?” Ask for a date. Vagueness here is the clearest signal available.
- “Will my site appear in your sellers.json under my own domain?” A yes is a commitment to a public, checkable record of the relationship. A no may be legitimate — confidential sellers are permitted — but you should understand why.
- “How quickly do you remove publishers who leave?” A network that prunes promptly is one whose claims about anything else are more likely to be accurate.
Why this check is worth doing at all
Because the roster is a proxy for the two things you actually care about and cannot see. A network with real, current, comparable publishers has demand relationships that work for inventory like yours. A network whose published roster is substantially aspirational may still be perfectly good — but you have just learned that its public statements need verifying, which is a useful thing to know before signing a contract.
Every network we track has a page here with its full indexed publisher list and the evidence behind each match — start from all networks. The rules we apply are on the coverage & method page.
Common questions
How can I check if an ad network really works with a publisher?
Open the network's sellers.json and find the publisher's domain, note the seller id, then open that publisher's ads.txt and look for the same id. Both files present and agreeing means two independent parties declared the relationship. Only one side means it is a claim rather than a confirmed relationship.
Does a low confirmation rate mean a network is lying?
Not by itself. Resellers legitimately never appear under their own name in a publisher's file, agencies use shared seats, and some publishers serve no readable ads.txt at all. The one genuinely bad explanation is stale entries — publishers who left and were never removed — and it is also the most common. A low rate is a reason to ask which applies, not a verdict.
Which ad networks have the best-kept publisher records?
Raptive leads the networks we track at 95% of its published relationships confirmed by the publisher side. A high rate indicates a network that lists real publishers and removes those who leave, which is a reasonable proxy for how carefully it maintains its other public claims.
What should I ask an ad network before signing?
Whether the publishers on its deck are currently live and as of what date; whether your site will appear in its sellers.json under your own domain; and how quickly it removes publishers who leave. All three are reasonable, all three are checkable afterwards, and the willingness to answer them precisely is itself informative.
“Confirmed” means the publisher’s own ads.txt or homepage evidences the network. Unconfirmed relationships may still be real; see the four explanations above. Every figure on this page is read from our own index when the page is built, and refreshes on a weekly cycle — so the numbers move as the data does. Method and evidence rules: coverage & method.