How backlinks actually work, and what one is worth

Almost everything written about backlinks is written about acquiring them. That is the wrong place to start. If you do not understand what a link is doing inside a search engine, you cannot tell a valuable one from a worthless one, and you will spend months chasing links that were never going to move anything.

This is the mechanism piece. No outreach templates, no tactics. Just: what a link signals, why some links count for far more than others, what the various attributes actually change, and how to decide whether one specific link is worth your time or money. If you need the plain definition first, what a backlink is in plain terms covers the vocabulary.

Why the idea worked in the first place

Before link-based ranking, search engines mostly read the page itself — the words on it, how often they appeared, what the title said. That is trivially easy to game. Anyone could write "best plumber in Leeds" forty times and outrank an actual plumber.

The insight behind link analysis was that the web already contains a distributed, unforgeable-ish opinion about which pages matter: the pages people chose to link to. A link is a small act of editorial cost. Someone had to decide this page was worth sending a reader to. Aggregate millions of those decisions and you get a rough map of what the web considers important — built from evidence outside the page, which the page's owner does not control.

Two properties made this powerful:

It is external. You can write anything about yourself. You cannot easily make thousands of other people write about you.

It is recursive. A link from a page that is itself widely linked-to means more than a link from a page nobody references. Importance flows through the graph. This is why "how many links" was never really the question — the original models were about where links came from, not how many there were.

Both properties also explain the whole subsequent history of the industry. The moment link-based ranking worked, the incentive to manufacture links appeared, and search engines have spent every year since separating links that represent a real editorial decision from links that were bought, swapped, injected or automated.

Not all links are equal, and the differences are large

Treat link value as a distribution, not a count. Most links in most profiles do close to nothing. A small number do nearly all the work. Here is what plausibly drives the difference — "plausibly" because Google does not publish the weightings, and anyone who tells you they know the exact formula is selling something.

The linking page's own standing

A link inherits from its source. A page that is itself well-referenced passes more than an orphaned page nobody visits or links to. Note the unit: the *page*, not the domain. A link buried on a dead subpage of a famous site is not the same as a link on that site's actively-referenced pages, even though a domain-level metric will score them identically.

Topical relevance

A link from a page about drainage to a page about drainage carries context. A link from a page about cryptocurrency to a page about drainage carries almost none, no matter how "authoritative" the crypto page is. Relevance operates at several levels at once — the site's overall subject, the specific page's subject, and the surrounding sentence.

Placement on the page

A link inside the body of an article, in a sentence that gives a reason to click, is a genuine editorial reference. The same URL in a site-wide footer, a sidebar widget or an author bio block is a template, repeated automatically across every page. Templated links are cheap to acquire and search engines have had decades to learn what they look like.

Whether it was editorially given

This is the axis that matters most and the one people most want to argue about. A link somebody chose to add because the destination was useful is the thing the whole system was built to measure. A link you paid for, swapped for, or placed yourself in a directory submission form is a link where nobody made that judgement. Google's spam policies are explicit that buying or selling links for ranking purposes counts as link spam. Whether a given paid link gets caught is a separate question from whether it is against the guidelines — it plainly is.

How many other links share the page

A page linking to four things distributes its standing differently from a page linking to four hundred. This is why sprawling link directories and "top 500 resources" round-ups do so little.

Attributes: what rel actually changes

You will see links described as "dofollow" or "nofollow". There is no dofollow attribute — it is just a plain link. What exists is a set of rel values a publisher can add, documented in Google's guidance on qualifying outbound links:

| Attribute | What the publisher is saying | Practical effect | | --- | --- | --- | | (none) | This is a normal editorial link | Fully counted, subject to everything above | | rel="nofollow" | I do not vouch for this destination | Google treats it as a hint, not a hard rule | | rel="sponsored" | This link was paid for or is part of an ad | Declares the commercial relationship | | rel="ugc" | User-generated: a comment, forum post, profile | Declares the link was not editorially placed |

Two things people routinely get wrong here.

First, nofollow is no longer the absolute block it was originally. Google's documentation describes these attributes as hints it may take into account, rather than instructions it obeys. So a nofollowed link from a heavily-read publication is not automatically worthless — but you cannot count on it either.

Second, and more usefully: a link's *non-search* value does not depend on the attribute at all. A nofollowed link on a page your actual customers read still sends you actual customers. If you are judging every link by whether it passes ranking signal, you are ignoring the half of link value that is completely reliable.

Why "how many links do I need" has no answer

This is the most-asked question and it is genuinely unanswerable, for three separate reasons.

Ranking is competitive, not absolute. There is no threshold. You need enough to outweigh whoever currently occupies the positions you want, for the specific queries you want, and that number changes when they change. The mechanics of that competition are covered in how Google ranking works without the jargon.

Links are one input among many. A site with a weak link profile but exactly the right page for a query can outrank a site with a strong profile and a mediocre page. Links do not override relevance; they help decide between pages that are otherwise comparably relevant.

The counts you are comparing are wrong. Every backlink checker shows a different number for the same site, because each has crawled a different subset of the web. None of them see what Google sees. Comparing your "847 backlinks" to a competitor's "3,200" is comparing two estimates from a tool that cannot see the actual graph.

The honest reformulation is: *is this specific link worth what it costs me?* That question has an answer. The count question does not.

About those authority scores

Domain Authority, Domain Rating, Trust Flow, Authority Score — these are all inventions of the companies that sell backlink tools. They are not Google numbers. Google does not publish a public per-site authority score, and has said repeatedly that the toolbar PageRank figure people once cited has been gone for years.

That does not make third-party metrics useless. They are reasonable proxies for "how much of the link graph points at this domain", and as a rough sorting device between fifty candidate sites they save time. But three habits are worth breaking:

• Do not treat a threshold ("we only accept DR 50+") as a quality standard. It is a filter that any link seller can engineer against, and the ones selling links have engineered against it hardest.
• Do not compare scores across tools. They measure different crawls on different scales.
• Do not let a domain-level score decide a page-level question. The page matters more than the domain, and no domain score tells you anything about the page.

A worked judgement: two candidate links

Say you run a two-person plumbing business covering one town and its surroundings. Two opportunities land in the same week.

Link A. A national home-improvement site with a large audience and a high third-party authority score offers a listing in their "Find a tradesperson" directory. Your entry would be one of roughly 900 on a page that is four clicks from the homepage, has no inbound links of its own, and lists businesses alphabetically across the whole country. The link is rel="sponsored". Cost: £180 a year.

Link B. The local town website — run by the parish council, modest traffic, unremarkable authority score — is publishing a page on what to do when a pipe bursts in a cold snap. They want a paragraph of advice from a local plumber and will credit you with a link in the body text, in a sentence naming your business and your town. Free, costs you about an hour of writing.

Most people pick A, because the numbers on A are bigger. B is worth more, on almost every axis that matters:

| | Link A (national directory) | Link B (local council page) | | --- | --- | --- | | Source page standing | Deep, unlinked directory page | Modest, but a real page on a locally-referenced site | | Relevance | National, generic trades | Local, specific to your service and area | | Placement | Row in a 900-item list | Body text, in context | | Editorial | None — you paid to appear | Yes — they chose to feature you | | Attribute | sponsored | Plain link | | Non-search value | Near zero; nobody reads page four of a directory | Real; the audience is people in your town |

A costs £180 and buys you a row in a table. B costs an hour and buys you an editorial mention on a page your actual customers might read, on a site local to your service area.

The instructive part is that a tool would rank these the other way round. Every metric that is easy to look up favours A. Every judgement that requires you to actually open the page favours B.

Note also what would change the answer. If the national site were writing an article about emergency callout pricing and quoting you in it, that is a different link from the same domain — and probably better than B. The domain was never the unit.

A quick decision framework

Before pursuing any link, five questions. If you cannot answer them from the page itself in about two minutes, that is itself an answer.

1. Would a real person plausibly click this? If the honest answer is no, the link exists only to influence a search engine — which is the exact category Google's spam policies target.
2. Is the page about something adjacent to what I do? Not the site. The page.
3. Is it in the content, or in a template? Open the source page and check whether the same link block appears on other pages of the site.
4. Did someone choose to include me? If the mechanism was a form, a payment or a swap, nobody chose.
5. What does it cost, including my time? An hour of writing is a real cost. So is £180. Compare against what else that hour or that money could buy.

A link that passes 1–4 is worth pursuing at almost any reasonable cost. A link that fails 1 and 4 is not worth pursuing at zero cost, because it still costs you the time and carries a small risk.

Common mistakes

Chasing volume. Fifty templated directory links will do less than one editorial mention on a relevant page. The distribution is that skewed.

Optimising anchor text. Repeated exact-match anchors across many links is one of the more legible artificial patterns. Real editorial links use varied, natural anchors — your business name, the page title, a phrase from the sentence. If you are choosing your own anchor text on many links, that itself indicates the links are not editorial.

Treating links as a one-time project. Link graphs decay. Pages get deleted, sites shut down, articles get rewritten. A profile built in one push erodes.

Buying at the point of desperation. The people who buy links are almost always people who have been waiting for organic results, got impatient, and want a shortcut. Our honest breakdown of what buying backlinks actually costs and risks works through the arithmetic, including the cases where the answer is genuinely "don't".

Ignoring the referral half. A link that sends you twenty qualified visitors a month is valuable whether or not it passes any ranking signal. Judge links on both axes.

Where this leaves you

If you have no links and nobody has heard of you, the mechanism above tells you something uncomfortable: the links worth having come from people who chose to reference you, and nobody chooses to reference an unknown. That chicken-and-egg problem is real and it is the actual starting difficulty, not a lack of tactics. Getting your first backlinks when nobody has heard of you deals with it directly.

If you have links and no results, the likely explanation is not that you need more. It is that the ones you have are the low-value kind, or that links were never the constraint — the page might not match what people are searching for, or the competition might be doing something you are not.

And if you have concluded that earning editorial links is a slow business that you cannot wait out, that is an accurate conclusion rather than a defeat. Earning links honestly takes months to years, and that is the case whoever is doing it. The realistic version of that timeline is in what a new site's ranking timeline actually looks like.

We sell paid visibility, so we have an obvious interest in you deciding that organic link building is too slow. Which is exactly why we will say the opposite where it is true: if your business depends on ranking for a query you can realistically win, and you have the runway, earning links is the better asset. It compounds and you own it. A marketplace position on our public leaderboard is rented — it lasts as long as you pay for it, it does nothing for your search rankings, and it is honest about being a different thing entirely. We set out those limits in what pay-to-rank cannot do. Buy it if you want visibility now and can price it against what else that budget buys. Do not buy it as a substitute for links, because it is not one.

Questions people ask

Do links from social media count? Most are nofollowed or user-generated, so as ranking signals they are weak at best. As distribution — getting a page in front of people who might then link to it from their own sites — they matter a great deal. The value is indirect.

Can competitors harm me by pointing bad links at my site? Google's documentation states it is generally able to ignore links it does not trust, and negative link attacks rarely work as advertised. The disavow tool exists but is for sites that acquired bad links themselves, usually knowingly. Most site owners should not touch it.

How long before a new link has any effect? The page has to be crawled, the link processed, and any effect has to show up in a ranking that changes for many reasons at once. Weeks is normal. Attributing a ranking movement to one specific link is usually not possible — which is worth remembering when someone shows you a graph.

Is a link from a page with no traffic worthless? Not necessarily. Traffic and link value are different things — a page can be well-referenced by other sites while getting few visitors. But a page with no traffic *and* no inbound links of its own is close to worthless on both axes.

Should I remove old low-quality links? Usually not worth the effort. If they came from ordinary directory submissions years ago, they are almost certainly just being ignored. Spend the time on something that adds rather than subtracts.