What is a PBN?

The name sounds harmless. There is nothing illegitimate about a private blog, and nothing wrong with running several websites either. What turns a group of domains into a network of this kind is the intent behind its existence: the sites are not published to win readers of their own, but to send links to one main website.

The line between this and a legitimate group of websites lies in whether each property has a reason to exist on its own. A company with a shop, a corporate blog and a careers portal runs three sites with different audiences, content and goals; the links between them describe a real relationship and can be explained out loud. A publishing house with several titles works the same way. In a PBN it is the other way round: without the destination site, none of the properties would have an audience or a commercial purpose.

Google does not use the acronym PBN in its policies, but it does describe the behaviour. Its link spam policy prohibits creating links whose primary purpose is to manipulate rankings, and it expressly includes creating low-value content for the purpose of manipulating linking and ranking signals. A private blog network fits that description whatever trade name it is sold under, whether as authority links, content seeding, digital press or an owned media network.

The mechanism is easy to describe. Someone gathers several domains, often expired ones that still carry old links from their previous life, publishes filler articles on them and places links from there to the website that is meant to rank. The bet is that the search engine will read those links as independent recommendations. Google also treats expired domain abuse as a violation in its own right, where a domain is bought back to exploit its history for content that has nothing to do with what came before.

For the buyer, the useful question is not how Google detects it, but how to recognise the offer before signing. Almost all of it can be checked in a browser. Who publishes each site, and whether a name, a legal notice and a verifiable contact appear there. If the provider refuses to show the domains until after payment, that refusal is already the answer. If the articles sound the same across every site, if none of them has branded traffic, comments, a newsletter or any presence beyond its own domain, then those sites have no readers. A real publication can be verified from outside: third parties cite it, it shows up when you search its name, and it publishes things that serve none of its advertisers.

On the search engine side, detection combines automated systems with human review. When the quality team confirms a pattern of unnatural links, the consequence is recorded in Search Console under the manual action category “Unnatural links to your site”, and the notice goes to the domain receiving the links. Not to the one that sold them.

The decision that hangs on this is a very concrete one: accept or refuse a link offer. Almost none of them arrive labelled as a PBN. They arrive as a package of placements on topical media, with third-party authority metrics, a delivery date and a price per link.

The risk is badly distributed. If Google acts, the manual action lands on the client domain, not on the network that sold the links. The provider carries on trading and the advertiser is left with a link profile to clean up. Getting out of it is not automatic: you have to identify the links, try to have them removed, disavow the ones you cannot get taken down and request a review, with no guaranteed response time. In the meantime, lost visibility cannot be made up with media budget.

There is a second cost, less visible. Even where the search engine applies no manual action at all, it can simply ignore those links. In that case the money bought no positions: it bought nothing, and the link report still shows a profile that any later audit will flag as suspicious. For a company about to sell, raise funding or go through technical due diligence, that history weighs more than it cost to create.

• Publish something people have a reason to cite. Your own data, a study with an open methodology, a calculator or a guide that solves one concrete task will attract links without anyone having to buy them.
• Do digital PR with real publications. Taking a story to sector journalists is slower than buying a package, but the resulting link comes with readers and never has to be removed later.
• Write bylined pieces for industry publications. Identifiable collaborations, signed and with no payment in exchange for the link, that hold up if anyone looks closely.
• Recover what already exists. Unlinked brand mentions, broken links pointing at old URLs, and supplier or association pages where the link is missing usually yield more than a buying campaign.
• Label paid distribution as paid. When you pay for a placement, the sponsored or nofollow attribute on the link is the documented route and removes the problem at the root.
• Audit the link profile before investing. Knowing what was inherited from previous agencies avoids paying to repair the same damage twice.

• Taking a high authority metric as proof of legitimacy. Third-party scores describe links, not readers, and an expired domain can hold a high score for a long time with nothing behind it.
• Assuming the seller carries the risk. The manual action notice goes to the site receiving the links, and the agency contract almost never covers the loss of visibility.
• Trusting that disavowing afterwards fixes it. Disavowal is part of the way out, not an insurance policy; the review that follows is done by a person and comes with no committed deadline.
• Confusing a group of owned websites with a private blog network. Several brands or several countries under one owner are normal; what defines the problem is sites existing only in order to link.
• Accepting a proposal without seeing the domains. Buying blind rules out the one check that matters before payment, which is whether those sites have a life of their own.