Domain Rating vs Press Verification: The Data

We matched 737 Indonesian media domains against the national press council register, then compared the result to each outlet's Ahrefs Domain Rating. Two findings came out of it. One was expected. The second one was not, and it is the more useful of the two.

Expected: authority and verification correlate, but weakly. Verification rates run from 81.8 percent among outlets above DR 70 down to 9.7 percent below DR 20. So DR carries some information, just nowhere near enough to substitute for checking.

Not expected: our own "not verified" results were wrong often enough to matter. We labelled 138 rows incorrectly on the first pass, and the reasons generalise to anyone doing the same lookup. That second finding is why this piece exists.

Method, stated plainly so you can judge it

Source A: 737 Indonesian media domains from our own placement records, deduplicated, one row per domain.

Source B: the Dewan Pers register, cyber media category, pulled from the official portal on 2 September 2026. 988 records, comprising 830 verified factual and 158 verified administrative.

Source C: Ahrefs Domain Rating for every domain in source A.

Matching was done on domain, because that is the only field the two sources appear to share. Straight away, that's the weak point, and we'll come back to it.

Finding one: the correlation is real but shallow

First cut, including every row:

Domain Rating band	Outlets	Found in cyber register	Rate
DR 70 and above	208	70	33.7%
DR 50 to 69	91	31	34.1%
DR 30 to 49	169	50	29.6%
DR 20 to 29	94	24	25.5%
Below DR 20	175	17	9.7%

Look at the top three rows. Basically flat, around 30 percent, all the way from DR 30 to DR 90. If DR were a decent proxy for verification, the top band would be much higher than the middle.

Something was distorting it. That something turned out to be subdomains, and finding it led straight to the second finding.

Finding two: 138 rows were labelled wrong, and here is why

191 of the 737 domains are subdomains. A subdomain inherits its parent's Domain Rating almost entirely, because Ahrefs calculates DR at domain level. So `jakarta.tribunnews.com` shows DR 86 not because that regional edition earned it, but because `tribunnews.com` did.

Meanwhile the register lists `tribunnews.com` and does not list its regional editions separately, because no publisher registers every subdomain. So the subdomain arrives in the data with inherited high authority and an absent register match. It looks like a high-DR unverified outlet. It is neither.

Take the subdomains out and the picture changes completely:

Domain Rating band, root domains only	Outlets	Found in register	Rate
DR 70 and above	55	45	81.8%
DR 50 to 69	68	31	45.6%
DR 30 to 49	154	50	32.5%
DR 20 to 29	94	24	25.5%
Below DR 20	175	17	9.7%

Now it's a proper gradient. 82 percent down to 10 percent, monotonic. So DR does carry signal, and one confounder was hiding it entirely.

But notice what the corrected table also says. Even at DR 70 and above, 18.2 percent of root domains produced no register match. Ten outlets, and several of them are among the largest publishers in Indonesia.

Which raises the obvious question, and this is where we had to stop and be careful.

Why we are not naming those ten

We could list them. We are not going to, and the reason is the point of this article.

We do not have evidence that those publishers are unverified. We have evidence that our domain-based lookup did not find them. Those are different claims, and only the second one is supported.

Three explanations are available, at least two of which we can demonstrate:

Format category. The register is split by media type. A broadcaster verified as television, or a state radio network verified as radio, does not appear in the cyber list. Filter for cyber, and they are absent from your results while being fully verified elsewhere. We found this pattern in our data.

Registered under a different product name. Publishers run several properties. One may be registered while a sister product on a different domain is not listed separately, despite belonging to the same verified company. We found this pattern too.

Extraction gaps. The register's tables load through JavaScript and are paged. Our pull returned 988 cyber records, which is consistent with the 809 verified-factual figure Dewan Pers reported for May 2026 plus subsequent additions. Consistent is not the same as complete, and we cannot prove completeness from inside our own dataset.

So the honest statement is: ten high-authority root domains produced no match, and we do not know which explanation applies to each. Publishing a list of names under a "not verified" heading would convert a data limitation into an accusation. That would be wrong, and it would also be the kind of error that destroys the credibility of everything else in the dataset.

The register has no domain column, and that explains most of it

Here is the structural fact underneath all three failure modes.

The Dewan Pers table records media name, responsible person, address, province, legal entity, editorial phone, editorial email, and status. It does not record a website URL as an authoritative key.

So any attempt to answer "is this domain verified" is really answering "does this domain string appear somewhere in a table that was not designed to be queried by domain". Those are not the same question, and the gap between them is where every false negative lives.

The reliable procedure is the slower one. Take the legal entity name from the outlet's own masthead, search the register for that name, and compare. It's per outlet, it's manual, and it's the only version that produces an answer you can defend.

Two outlets that make the point

Rather than name the ambiguous cases, here are two clear ones from opposite ends.

There are outlets in our data sitting below DR 15 that hold factual verification. Factual is the higher of the two states: it means the press council physically visited the newsroom, checked the address, looked at the editorial workflow, and confirmed regular publication over at least six months. A DR of 12 tells you the site has few backlinks. It tells you nothing about the fact that a state body walked through the office.

And at the other end, there are aggregators and contributor platforms above DR 70 with no register entry of their own. Some of those are legitimate properties of verified publishers. Some are not publishers in any editorial sense at all. DR does not distinguish between the two, and neither does a domain lookup.

What to do with this

Four practical consequences, and the last one is the one that actually changes behaviour.

Limits of this study, stated up front

Sample is our own placement history, not a random sample of Indonesian media. It skews toward outlets that accept paid placement, which is exactly the population a media buyer cares about but is not the population of Indonesian media as a whole.

The register pull is a single snapshot dated 2 September 2026. Status changes: factual certificates run five years, and roughly 300 outlets were removed from the official register between December 2025 and June 2026 for not renewing. Any figure here has a shelf life.

We cannot prove our extraction of the register is complete, only that its scale is consistent with the official figures. And Domain Rating is a third-party metric that changes independently of anything in this analysis.

None of that invalidates the two findings. It does mean the numbers should be read as one dataset's shape rather than a national census.

The underlying data, with both columns kept separate and the check date shown next to each row, is published in the Arfadia media directory.

Frequently Asked Questions

Does Domain Rating indicate whether a media outlet is verified?

Only weakly. In our dataset of 546 root domains, verification rates ran from 81.8 percent above DR 70 down to 9.7 percent below DR 20. So DR carries some signal, but nearly a fifth of the highest-authority outlets produced no register match, and outlets below DR 15 held full factual verification. DR is not a substitute for checking.

Why did subdomains distort the analysis?

Ahrefs calculates Domain Rating at domain level, so a subdomain inherits almost all of its parent's score. A regional edition on a verified national publisher's subdomain therefore shows high authority while having no separate register entry, because publishers do not register every subdomain. Including them flattened a clean gradient of 82 to 10 percent into a meaningless 34 to 30 percent line.

What is the difference between not found in the register and not verified?

The first is a statement about your lookup. The second is a claim about the publisher's legal standing. A lookup only supports the first. In our data, 138 of 427 negative results, close to a third, turned out to have an explanation other than the outlet being unverified.

Why does the press council register not include website domains?

The Dewan Pers table records media name, responsible person, address, province, legal entity, editorial contact details and status. A website URL is not held as an authoritative key. So a domain-based query is really asking whether a string appears in a table that was not designed to be searched that way, which is where false negatives come from.

How should verification actually be checked?

Take the legal entity name from the outlet's own masthead, search the register for that name, and compare the two. Then note which format category the record sits in, because the register is split by cyber, print, television and radio. It is slower and per outlet, and it is the only method that produces a defensible answer.

Can a low Domain Rating outlet still be a legitimate newsroom?

Yes, and our data has clear cases. Outlets below DR 15 hold factual verification, meaning the press council visited the newsroom, checked the registered address, examined the editorial workflow and confirmed publication over at least six months. A low DR means few backlinks, not a weak newsroom.

What are the limits of this study?

The sample is one agency's placement history rather than a random sample of Indonesian media, so it skews toward outlets that accept paid placement. The register pull is a single snapshot from 2 September 2026, and status changes because certificates expire. Completeness of the extraction cannot be proved from inside the dataset, only shown to be consistent with official figures.

Should a media directory show verification status at all, given these limits?

Yes, provided the label is accurate. Showing "found in the cyber register with factual verification" is useful and defensible. Showing "not registered" for everything else is neither, because a third of those negatives have another explanation. The fix is the wording, not removing the column.

---

---