Selling Backlinks in 2026: What Publishers Actually Earn, What Google Really Penalises, and the Safer Way to Monetise Authority

There is a large, liquid, mostly invisible market for backlinks. Every week, publishers get emails offering a few hundred dollars to add a link to an existing post, and most of those emails are sent by someone working on behalf of a brand that would never admit to it in a case study.

Plenty of articles will tell you how to get in on this. Very few will tell you what the transaction actually looks like from both sides, what the current market rate really is, or what happens to the site that sells the link when Google notices.

We work on the buying side of link acquisition for clients every day. Here is the honest version.

How does it work when you sell backlinks?

The mechanics are simpler than the surrounding mystique suggests. A publisher lists a site – or gets found – and a buyer pays for one of two products.

Guest posts. The buyer supplies or commissions an article, the publisher runs it with a contextual link pointing at the buyer’s page. This is the bulk of the market: PressWhizz’s analysis of 22,703 completed placements found guest posts accounted for roughly 64% of orders and 75% of revenue.

Link insertions (niche edits). The buyer pays to add a link into an article the publisher already published. Cheaper, faster, and no content to review. FatGrid’s study of 1.3 million offers found insertions run about 39% below guest post prices on average, though not always – in 13% of cases the insertion actually costs more, usually when the target article already ranks.

The transaction itself follows a predictable shape:

1. Contact. An outreach email, a marketplace order, or an enquiry through a “write for us” page.
2. Vetting, in one direction only. The buyer checks the publisher’s metrics. The publisher rarely checks the buyer at all — which is where most of the seller’s risk originates.
3. Price agreement. Usually a flat fee per placement. Some publishers charge extra for a followed link, extra for a restricted niche, extra for more than one link in the article.
4. Content. Supplied by the buyer, written by the publisher for a fee, or produced by whatever tool the buyer is using that month.
5. Publication and payment. Typically PayPal or bank transfer, often paid on publication, sometimes 50% up front.
6. The link-permanence clause. Increasingly common: the buyer requires the link stays live for 12 months or longer. This matters more than it looks – it converts a one-off sale into a long-term liability on your domain that you have contractually agreed not to remove.

Almost all of this gets labelled “sponsored content,” “partner content,” or nothing at all.

What publishers actually earn

This is where most articles on the subject become fiction. Numbers get quoted from screenshots of income reports that are half a decade old, or from surveys of agencies quoting their own retail rates.

The most useful data comes from marketplaces that publish their own transaction records. One analysis of more than 22,700 completed link placements found an average price of about $161 and a median of $112 – meaning half of all placements sold for less than that, before any agency markup. Guest posts made up roughly 64% of orders at an average of $164, and the strongest predictor of price was not Domain Rating but real organic traffic.

Studies based on rate cards rather than transactions land much higher. Adsy’s review of over 52,000 websites puts the average guest post at $459 against $225 for a link insertion, up from $427 the year before. BuzzStream’s analysis found guest post links averaging around $365, with higher-quality placements in the $692–$957 range before markup.

Both sets of numbers are true. They measure different things: asking price versus clearing price. A publisher’s rate card says $450; the marketplace fills the order at $112 because supply is enormous.

What that means in practice. A mid-tier site in a non-competitive niche selling ten placements a month at market clearing rates is looking at roughly $1,100–$1,600 in gross monthly revenue. To reach the “$3k a month” figure the industry likes to advertise, you are either sitting on genuine authority and real traffic in an expensive vertical — finance, gambling, and sports command the highest prices — or you are selling volume at a rate that will eventually get the site classified as what it has become.

What Google’s policy actually says

Here is the part the “how to sell links ethically” articles skip.

Google’s spam policies name buying or selling links for ranking purposes as link spam. That covers exchanging money for links or for posts containing links, exchanging goods or services for links, and sending someone a product in exchange for a write-up with a link. It explicitly names advertorials and native advertising where payment is received for articles containing links that pass ranking credit.

The policy applies to both sides. The site receiving the link and the site selling it are each in violation.

There is a legitimate version, and Google states it plainly: paid and sponsored links are a normal part of the web’s advertising economy and are not a violation, provided the link is qualified with rel=”sponsored” or rel=”nofollow”.

That single attribute is the entire dividing line. A disclosed, attributed sponsorship is advertising. The same placement with a followed link is a policy violation, and the reason buyers pay a premium for “dofollow” is precisely because they are paying for the violation.

Anyone selling you an “ethical, white-hat, Google-safe paid link” is selling you a category that does not exist.

Is selling backlinks illegal?

One thing worth stating plainly, because it comes up constantly: selling backlinks is not illegal. No law prohibits taking money to publish a link. What it does is breach the terms of a private company whose traffic your business may depend on, and — as the disclosure section below covers — it can breach consumer protection rules when the payment isn’t disclosed. Those are two separate exposures, and neither is criminal. People conflate “against Google’s rules” with “illegal” and then reason their way into ignoring both. They’re different risks with different consequences, and the first one is the one that will cost you money.

The risk sellers don’t hear about

Buyers worry about penalties. Sellers usually assume the risk sits entirely with the other party. It doesn’t.

Unnatural outbound links. This is a specific manual action, issued when Google’s reviewers determine a site is selling links that pass ranking credit. It appears in Search Console under Security & Manual Actions. Recovery means removing or correctly attributing the links, then submitting a reconsideration request – an process with no guaranteed outcome and no fixed timeline.

Algorithmic devaluation. Far more common and much harder to diagnose, because there is no notification. SpamBrain has been refined continuously through the 2025 and 2026 spam updates. Google’s own documentation is blunt on this point: once the ranking benefit of spammy links is neutralised, it does not come back. Traffic drops, nothing appears in Search Console, and the publisher spends months looking for a technical cause that isn’t there.

Do paid links still work in 2026?

That last point is also the answer to the question buyers ask most often. Yes, sometimes, for a while. That’s precisely what makes them a bad purchase. A tactic that works until it is detected and then retroactively yields nothing is not an investment; it is a lease with an unknown expiry date and no refund clause. The ranking gain disappears, the money is already spent, and the recovery path is to rebuild the authority you were renting.

The compounding problem. Link revenue is a decay function disguised as a growth function. Each sale slightly degrades the editorial quality that made the site worth buying from. Rates fall as quality falls. The publisher sells more placements to hold revenue flat. Eighteen months later the site is inventory, not a publication, and the underlying business – the audience, the email list, the ad rates, the thing that had actual enterprise value – is gone.

The disclosure problem nobody budgets for

Search penalties are not the only exposure. In the United States, the FTC’s Endorsement Guides require clear and conspicuous disclosure of any material connection between an endorser and an advertiser. Material connection covers cash, free products, discounts, affiliate commissions, and access – anything a reader would want to know when weighing the recommendation.

The standard is that a disclosure must be prominent, unambiguous, and hard to miss. Vague labels don’t satisfy it. And the FTC’s 2023 revision expanded what counts as an endorsement to include tags, reposts, and likes, and applied the same rules to AI-generated personas.

Civil penalties run to five figures per violation. For a publisher running undisclosed paid content at volume, that is a materially worse downside than a ranking drop.

The link farm economy, and why it’s a bad business now

A meaningful share of “publisher inventory” is manufactured. The playbook is consistent: buy an expired domain with existing backlink history, install a template, generate content at scale targeting high-volume low-difficulty keywords, and let third-party tools report the site as a trafficked authority.

The traffic is engineered to look real to a metrics tool rather than to be real. Ranking on page two for celebrity net worth or a branded login query produces an impressive number in Ahrefs and no human beings.

Two things have changed since this tactic was profitable:

1. Expired domain abuse and scaled content abuse became named policy categories in March 2024, which means detection is systematised rather than opportunistic.
2. Buyers got better at screening. Any competent agency now checks traffic distribution against topical relevance, outbound link velocity, and whether the site’s rankings are commercial or vanity queries. Inventory that only looks good in a metrics tool is being priced out.

The economics still work for the operators, briefly. They do not work for the brand paying $400 for a placement on a site that will be devalued before the campaign report is written.

Why links still matter for AI search – just not the way you think

Here is the argument you’ll hear from anyone selling link packages in 2026: AI search is the new battleground, links build authority, therefore buy more links.

The research does not support that chain of reasoning. It supports something more specific, and understanding the difference is worth real money.

The correlation data

Ahrefs analysed 75,000 brands to find which signals actually predict whether a brand appears in Google’s AI Overviews. The ordering upended a decade of link-first strategy:

Brand mentions correlate roughly three times more strongly than raw backlink count. And critically, that mention metric counts linked and unlinked mentions alike – the link is not what the correlation is measuring.

That detail is the most exploitable finding in the whole dataset. An unlinked mention passes no PageRank, so under traditional SEO logic it is worth nothing and nobody bids on it. Under the AI visibility layer it appears to be worth more than the link. Which means unlinked mentions are systematically underpriced relative to what they now do – and the standard industry practice of chasing a publisher to convert a mention into a link is spending outreach effort to bolt the weaker signal onto the stronger one you already had.

Seer Interactive reached a compatible conclusion by a different route. They ran 10,000 brand-triggering questions through GPT-4o and joined the results against SERP data. Page-one Google rankings correlated around 0.65 with LLM brand mentions. Backlinks and domain rank came out weak, in places effectively neutral.

The mechanism, and why it explains the numbers

Seer proposed a two-step model based on 541,213 LLM responses across 20 brands and six platforms: the system generates its answer first, choosing which brands to name from what it already holds, then runs a retrieval step to find sources supporting those choices. Citations are the bibliography, not the brainstorm.

In their data, a brand’s citation rate was 53.1% when the brand was already named in the response and 10.6% when it wasn’t.

Seer themselves flag this as inference rather than proven fact – token generation isn’t directly observable. But it fits the correlation data cleanly. If the model decides who to talk about from accumulated exposure across the web, then the number of times your brand name appears in credible contexts matters far more than the number of <a href> tags pointing at your domain.

A bought link on a site nobody cites achieves neither. It doesn’t generate the branded mention in a context models trust, and the anchor text is usually a commercial keyword rather than your brand name – which is the signal correlating at 0.527.

Where links do the work: getting you onto citable pages

Backlinks haven’t stopped mattering. They’ve moved upstream.

Ranking well still gets you retrieved. Ahrefs’ first study of 1.9 million AI Overview citations found 76.10% of cited pages ranked in Google’s top 10. Their updated analysis tells a more complicated story: 37.10% of cited URLs now rank top 10, 26.20% sit at positions 11–100, and 36.70% don’t rank in the top 100 at all.

Read those two studies together and the strategic implication is clear. Ranking is still a strong path into the citation set, but it is no longer the only path – a third of citations now come from pages that don’t rank at all. Query fan-out means AI systems are pulling from a much wider pool than the ten blue links.

This is where link building earns its place, indirectly: links get your brand onto the pages AI systems prefer to cite from. Seer’s analysis of 76.7 million AI Overviews found roughly a 0.70 correlation between being mentioned on heavily-linked pages and AI Overview visibility. You want to be inside the authoritative roundup, not just linked from it.

The citation surface is not where most agencies are buying links

Ahrefs’ Brand Radar tracks the most-cited domains in AI Overviews. As of July 2026:

Not one of those is a site you can buy a guest post on.

The pattern is consistent across platforms. Peec AI’s analysis of 30 million cited sources ranked Reddit first across ChatGPT, AI Mode, Gemini, Perplexity, and AI Overviews. Wix’s 2026 review found listicle content – “best X for Y” roundups – accounting for around 21.9% of all AI citations, the single highest-converting format.

So consider what a typical $400 paid guest post buys you against that backdrop. A followed link on a mid-tier blog with modest traffic, no community, no video presence, no meaningful chance of being retrieved as a source. It may move a ranking. It contributes nothing to the mention layer that increasingly decides whether you appear in an AI answer at all.

A worked example

Two B2B SaaS companies, same budget, same twelve months.

Company A spends $24,000 on 60 paid placements at $400 each. Result: 60 followed links, mostly from sites in the same marketplace inventory their competitors are buying from. Some ranking movement on mid-tail terms. Zero unlinked brand mentions generated. Zero presence on Reddit, YouTube, or in comparison roundups. When a buyer asks ChatGPT “what’s the best [category] tool for [use case],” the model has no accumulated exposure to draw on.

Company B spends the same $24,000 on an original benchmark study, a digital PR push to get it covered, placement in six industry “best tools” roundups, a modest YouTube presence, and genuine participation in two relevant subreddits. Result: fewer followed links, but dozens of branded mentions across exactly the source types AI systems retrieve from, plus the earned links the study attracts on its own.

Company A bought the signal that correlates at 0.218. Company B built the one that correlates at 0.664.

Two things not to waste budget on

llms.txt. Ahrefs studied 137,000 domains and found only about 28% publish the file – and among those that did, 97% received no requests for it from any bot in the month studied. SE Ranking’s analysis of roughly 300,000 domains found no measurable relationship between having the file and AI citation frequency. Google’s John Mueller has compared it directly to the old keywords meta tag: a claim a site makes about itself that nobody verifies.

Stale content. Analyses of AI-cited URLs consistently find a recency bias – cited content runs meaningfully fresher than the organic top 10, and the majority of AI crawler hits land on material published within the past year. Updating your best pages is cheaper than a link and demonstrably more relevant to this channel.

The honest caveat

Every figure above is a correlation observed in a sample, not a demonstrated cause. Brands with lots of web mentions are usually well-known brands, and well-known brands get cited for reasons that have nothing to do with the mentions themselves. The mechanism Seer proposes is a hypothesis they explicitly label as such.

Treat this as directional evidence about where to point your budget, not as a formula. Anyone quoting these numbers as settled science – including anyone selling you an “AI visibility package” on the strength of them – is overreaching.

How much does a backlink actually cost?

This is the question everyone actually wants answered, and it is the one the industry answers worst – usually with a single average lifted from whichever study flatters the writer’s business model.

What does a backlink cost by Domain Rating?

DR is the metric the market prices on, even though it probably shouldn’t, but also prices vary based on a link building georgraphics. Here is what the two largest listing datasets report:

Two findings should change how you read that table.

First, the market is far weaker than the tiers suggest. BuzzStream found the average DR across the entire guest-post marketplace is just 24, roughly 80% of sites sit under DR 40, and 39.6% have zero average monthly organic traffic. They classified 96.2% of guest post sites as low quality.

Second, DR is not what actually drives price, and it is not what should drive yours. PressWhizz found the single strongest predictor of price was organic traffic, not Domain Rating. That arbitrariness is the publisher’s opportunity. If you have real traffic and a real audience, the market is systematically underpricing you relative to DR-farmed inventory. Sell on traffic and relevance, and refuse to be benchmarked against a DR 60 expired domain with no readers.

What does a backlink cost by niche?

Niche premiums are real but smaller than folklore suggests. Adsy’s listing data (against their $459 average):

What does a backlink cost by country?

This is where the data gets genuinely surprising, and where we found a contradiction nobody appears to have written about.

Median guest post price across 64 countries, filtered to markets with at least 100 publishers above Authority Score 10:

What this means if you’re on the buying side

Most people who read an article about selling links are actually thinking about buying them. So, briefly:

The market clearing price for a link is around $112. If someone quotes you $400 for a placement, roughly $290 of that is agency margin – which is fine if the agency is doing genuine vetting and outreach, and pure waste if they are reselling from the same marketplace you could access directly.

More importantly: a purchased link is a rented ranking. It carries no compounding value, it can be neutralised in any update, and the money is gone either way. As the section above sets out, it also buys you the weakest available signal for the channel where discovery is shifting fastest.

The same budget put into original research, a genuinely useful tool, or a digital PR campaign that earns coverage produces links that don’t need to be re-bought next quarter – and produces the branded mentions, roundup inclusions, and community presence that increasingly determine whether you show up in an AI-generated answer at all.

That is a slower answer than a marketplace order form. It is also the only one we are willing to put our name on.