Where we stand
Quality is a judgement, the number is an input
No API can say whether a link is good for your page. It can show who links, how many distinct domains, and how a domain scores on a 0 to 100 scale. Someone still has to look.
- Read referring domains before rows of backlinks
- Compare against peers, not against a magic threshold
- Keep a person in the loop for anything you act on
Seven myths, seven corrections
| Myth | Reality | What to do in the pipeline |
|---|
| A DR of 40 is a pass, 39 is a fail | DR is a log scale from our own formula, round(100 · log(ref_domains+1) / log(2·10^7)). One point is not a verdict | Use bands, such as under 20, 20 to 50, above 50 |
|---|
| The graph shows my link today | The index refreshes roughly once a month | Store the check date, and use Search Console for same-day proof |
|---|
| Our DR is the same as Ahrefs DR | It is comparable in spirit, not the same number, like Moz DA | Never mix two vendors' scores in one column |
|---|
| Low score means block the publish | A new or niche domain can score low and be perfectly fine | Warn and tag, do not block |
|---|
| More rows means more truth | A call is billed once up to 1,000 rows, but rows from one domain repeat | Count referring domains, not backlinks |
|---|
| Webhooks fire exactly once | Some senders retry, so a handler can run twice | Make the check idempotent per entry version |
|---|
| Free calls need no design | Timeouts and failures still hit your editors | Set a short timeout and fail open |
|---|
Reading the first and the fifth myth
Take the DR cutoff. A team picks 40 because it sounds round. Then a niche but respectable site with 600 referring domains lands at 38 and gets flagged for no good reason.
The formula explains why. At 600 referring domains the score is 38, at 1,000 it is 41, at 5,000 it is 51. Going from 600 to 5,000 more than eight-fold buys only 13 points, so the scale compresses fast at the top.
Now the fifth. Backlink rows are cheap, but ten thousand rows from three domains tell you less than 300 rows from 300 domains. Sort by referring domain first, then drop into rows when a rule fires.
Where Google fits
Google has said for years that it ignores most spammy links rather than punishing them. SpamBrain, the system behind the December 2022 link spam update, works in that direction.
That is why a pipeline that blocks on a low number does more harm than good. Your signal is weak. Google's is strong, so use ours to prioritise reviews, then check the Links report in Search Console for what Google itself reports.
Related questions
Is there a free way to try this on one page?
Yes. The free domain and overview actions need no key, and the DR checker on the overview takes a domain directly. Run it on your own site before you decide on any rule.
Should a low score block publishing?
No. Warn the editor and tag the entry. A person can judge a partner link in seconds, while a gate on a monthly-refreshed score will misfire on new domains. See the post-mortem for an invented example of that.
How often should the check run?
Run a free overview on publish and a bulk_dr sweep nightly or weekly. Because the index moves about monthly, checking the same domain every hour tells you nothing new and only adds noise to your logs.