A high-quality backlink is a link from a page people read, on a site that covers your subject, placed because an editor chose to. That is the whole definition; the rest of this post is how to check it. Every link seller quotes domain rating, domain authority, or a similar number instead. It is the easiest thing to quote, it fits in a spreadsheet, and it is the least useful thing you can check. This post explains why, then gives the method we use on every prospect before we write to them. It takes about ten minutes per site and you can run it on any link report you have been sent.
Why domain metrics mislead
Domain-level authority metrics from third-party tools estimate the strength of a site’s link profile. They measure links pointing at a site, not whether anyone reads it. That distinction is the whole problem.
A site built to sell links optimises for exactly this number. Its operator buys or builds links to the domain, publishes anything a paying client sends, and quotes the metric in the sales pitch. From the outside, in a spreadsheet, it looks like a strong site. From the inside, it has no editor, no readers, and a hundred outbound links a week to unrelated businesses. Google’s spam policies name this pattern explicitly, its systems are built to find it, and when they do, the sites linking from it lose value and sometimes more.
Metrics are not useless. A very low score on a site that claims to be a major publication is a warning. But a high score tells you nothing about whether the link will help, and it is the number that link shops have spent a decade learning to inflate. Use it as a filter for obvious nonsense, then ignore it.
The five checks
Run these in order. A site that fails the first or fourth is out, regardless of how it scores on the rest.
Topical fit
Does the site cover your subject as a matter of course, or would your link be the only one on its topic? Open the homepage and the category the link would sit in. A B2B software company’s link belongs on a site that covers software, business operations, or the industry it sells into. It does not belong on a general “business blog” whose last ten posts cover mortgages, CBD, crypto, and wedding venues. Google’s guidance on link schemes and its spam policies both focus on relevance, and topical mismatch is the most visible sign that a link was placed rather than earned. Two minutes.
Real traffic
Does the site have organic search visitors in a third-party tool, and does the specific section or page rank for anything? A site with a high authority score and near-zero organic traffic is a site whose links Google has decided not to trust. Check the trend too: a site that lost most of its traffic in a single month has usually been hit by a spam or core update. Two minutes.
Indexing
Search Google for site:example.com and compare the result count with the number of pages the site publishes. Then search for the exact title of a post from last week. A site that publishes thirty guest posts a week and has a few hundred pages indexed is publishing faster than Google is willing to index, which means most of its links live on pages Google has never seen. One minute.
Editorial standard
Is there a named editor or a masthead? Do contributor articles show signs of review (consistent formatting, edited prose, relevant images) or are they clearly pasted in? Do articles have real authors with real profiles, or one “admin” account? Is there a contact page with a person or a company behind it? A site with an editorial process rejects some submissions. A site with no editorial process publishes everything, and that is what a link farm is. Three minutes.
Outbound link pattern
This is the fastest single tell. Open five recent articles and look at where their outbound links go. On a real publication, outbound links point at sources, related coverage, and occasionally a relevant company. On a link shop, they point at casinos, payday lenders, essay-writing services, and CBD retailers, often in the same article as a B2B software review. If you see that pattern, the site is for sale and everyone who links from it is standing next to the casino. Two minutes.
What a purchased profile looks like from the outside
Two signals separate an earned link profile from a purchased one, and both are visible in any backlink tool.
Anchor text distribution. Earned links are mostly branded (the company name), naked URLs, and descriptive phrases that vary because different editors wrote them. Purchased links skew toward exact-match commercial anchors (“best accounting software for startups”) because the buyer specified them. A profile where a large share of anchors are exact-match commercial phrases has been bought, and it looks bought to anyone who checks.
Link velocity. Earned links arrive unevenly: a burst when something newsworthy happens, a trickle otherwise, roughly proportional to how much the company is talked about. Purchased links arrive at a flat rate, twenty a month every month, because that is the package. A steady monthly count uncorrelated with anything the company did is a schedule, not a reputation.
Neither signal is proof. Both are what a reviewer, human or algorithmic, looks at first.
How to use this on a vendor’s report
If you have a link building report in front of you, take five placements at random and run the five checks on each. If more than one fails topical fit or editorial standard, ask the vendor how the sites were selected. If the answer is a metric threshold, you have your answer.
Then ask for the prospect list before the next month’s outreach begins. Approval before spend is the only real lever a buyer has. A vendor who will not show you where they intend to place links is telling you that you would not approve.
Finally, look at the anchor text column. If most anchors are the commercial phrases you would type into Google to find your own product, the vendor is building a profile that will need cleaning up.
A worked example. A report lists a link from a site with a domain rating in the sixties, described as a “business and technology publication”. Topical fit: the homepage carries articles on cryptocurrency, home renovation, and CBD alongside two on SaaS. Fail. Real traffic: a third-party tool estimates a few hundred organic visits a month, down from tens of thousands a year ago. Fail. Indexing: the site has published roughly forty articles this week and Google has indexed a handful of them. Fail. Editorial standard: every article is by “Editorial Team” and the contact page is a form with no company name. Fail. Outbound pattern: the article that links to you also links to an online casino and an essay-writing service. Fail. Five checks, five fails, and a domain rating that made the spreadsheet look fine. That is the link you are paying for when you buy by metric.
Also check the link attributes. A rel="sponsored" or rel="nofollow" link on a real publication is honest and still useful; Google’s guidance on qualifying outbound links explains what each attribute tells it. A followed link on a site that sells followed links is the one that carries risk, because it is the one pretending to be something it is not.
Running the checks on our own work
This is the checklist we use before outreach at our link building service, and clients see the prospect list with fit and traffic notes before we send anything. We publish it because a buyer who can check our work is the buyer we want, and because the checks are not secret; they are what any careful reviewer does.
If you inherited a link profile you are unsure about, a link profile audit is part of the audit we run, alongside the technical SEO audit checklist. We recommend using Google’s disavow tool only when there is a clear manipulative pattern, because over-disavowing does its own damage.