What makes a backlink good (and how to tell)

A good backlink is one a real person could plausibly click. That single test does most of the work, and it explains almost everything the tools measure in a roundabout way. The rest of this article is just a way of making that test usable on fifty prospects on a Monday morning.

The grid, two minutes per link

Six criteria, two points each. None of them is disqualifying on its own; the total is what decides.

Nine points or more: a link you would be pleased to have earned. Between five and eight: fine, nothing more, and not worth three follow-up emails. Below five: your outreach time is better spent elsewhere.

There is nothing scientific about this grid and it does not pretend otherwise. Its value is that it forces you to look at the page rather than at a tool's score, and it makes two prospects comparable that your instinct had filed as equivalent.

The four questions worth asking

Could someone actually click it? A link buried in a footer, or on a "partners" page nobody visits, does not get clicked, and a link nobody clicks tells search engines very little. A link inside a paragraph, where it genuinely helps the reader, does.

Is the page about the same thing you are? A link from an article on kitchen renovation to a plumbing supplier makes sense to a reader. The same link from a page about cryptocurrency does not, and that mismatch is visible to anyone, including a machine.

Does the site have a real audience? Not a traffic number: a line. Does it publish regularly, on a subject, for someone in particular? Sites built to host links have a tell: they publish about everything, in several languages, with no apparent reader. Two minutes on their homepage almost always settles it.

How many other links leave that page? A page with four outbound links gives each of them weight and attention. A page with two hundred is a directory, and readers know it as well as search engines do.

What the scores do and do not say

Domain authority, domain rating, trust flow: these are estimates produced by SEO companies from their own crawl of the web. Three of them will give three different numbers for the same site, which should be enough to settle their status.

They are useful for sorting fifty prospects into an order, and they do that well. They are not what Google uses, they do not measure traffic, and a page can score beautifully while being read by nobody: the score describes the domain, the value happens on the page.

Use them the way you would use a credit score: as a first filter, never as the decision.

The red flags, in order of seriousness

• The site sells links openly, price list or not. Whatever the page looks like, the pattern is detectable and the risk is yours, not the seller's.
• A "partners", "friends" or "resources" page with dozens of unrelated entries.
• Articles written around the link rather than the other way round: you can feel it in the first paragraph, and so can a reader.
• A site publishing in six languages with no identifiable translator, on unrelated subjects, at a pace no newsroom could sustain.
• The same anchor text repeated across several sites. See what your anchor text says about you.

None of these is fatal on its own. A pattern of them is. And a bad link is not something you disavow on reflex: here is what a disavow file actually does, and why it is almost always unnecessary.

Three mediocre links that deserve better than their reputation

A strict grid ends up rejecting useful links. Three cases are worth defending.

The serious trade directory. It will not move your ranking, but buyers read it, and it sometimes sends more customers than a well-scored editorial link.

The nofollow link on a very large publication. It passes no authority; it brings readers, recognition and often other links afterwards. Its value is not readable from the attribute: what the rel attribute actually changes is worth understanding before turning down a mention in a national newspaper.

The link from a tiny specialist site. Derisory score, three hundred readers, all of them exactly yours. No tool will rank it well, and it will suit you better than ten generic links.

Where the good links actually come from

A grid that sorts is useless if there is nothing to sort. Links that score nine almost always come from the same four places, and none of them looks like a campaign.

Trade resource pages. Every sector has a few, maintained by someone who keeps them current because it is their subject. Getting on one requires being genuinely useful, and those pages survive redesigns better than anything else.

Journalists and writers looking for a source. They quote whoever answers quickly, with a verifiable figure and a quotable sentence. It is the one channel where your response time matters more than the quality of the site.

What you publish that others need to cite. An original piece of data, a method explained, a comparison table kept up to date. These links arrive without being asked for, and they score highest on the "reason it is there" row.

Relationships that already exist. Clients, suppliers, trade associations, former employers. The link is legitimate because the relationship is, and nobody has to pretend.

What is not on that list: organised link exchanges, marketplaces, and guest posts written to place an anchor. They produce links that rarely score six, and that leave at the first editorial clean-up.

What the grid does not tell you

It does not measure real traffic, it knows nothing about your market, and it completely ignores what the link cost you. A seven-point link earned in ten minutes beats a ten-point link earned over six weeks of follow-ups, and no table will say that for you.

Its only purpose is to make a decision explainable: to yourself in three months, or to a client asking why you passed on a site their tool scored well.

The criterion nobody lists

A good backlink is one that still exists next year. A perfect editorial link that disappears in a redesign six months later has cost you the outreach and returned nothing.

That criterion is reasonably predictable: a link the author found useful survives changes of staff, while a link placed as a favour goes in the first clean-up. It is the same reasoning that explains why some links last and others do not.

Judging a link at acquisition is therefore only half the job: links go away quietly, and the ones worth having are worth watching.