Link Farm Inclusion

What a link farm is, and what inclusion means

A link farm is a group of web pages or sites that exist mainly to link to each other and to paying members, generating inbound links at volume with no editorial reason behind any of them. The family includes several related structures worth naming because readers meet them under different labels:

• The link wheel, where sites are arranged so that link equity passes around a ring rather than being exchanged directly between two sites, which was designed to evade reciprocal-link detection.
• The private blog network, or PBN — expired or purchased domains rebuilt as fake independent blogs whose only function is to link to whichever money sites are paying.
• The expired directory network — dormant or abandoned web directories bought in bulk, resurrected, and repopulated with outbound links.

Link farm inclusion as a negative SEO attack means adding a target's URL to one of these structures without the target's knowledge or consent. Most farms and directories accept submissions with no verification of any kind, and many are fully automated, so putting another party's site inside one costs the attacker close to nothing. The theory is that Google will treat membership in a recognized network as evidence that the target is running or paying for it, and demote or penalize them for it.

The signal being aimed at is the one Penguin was built to find: a link profile — every page on the web that points at a given site — dominated by machine-built links with no editorial judgment behind them. The flaw in the theory is that being easy to recognize is precisely why link farms became worthless, and a class of link that search engines recognize as worthless does not become dangerous by being pointed somewhere new.

This page names no network, directory, vendor or service, and describes no submission procedure.

18 September 2014: when Google acted, it acted on the network

Most negative SEO questions have to be argued from what Google says. This one can be argued from what Google did, at scale, in public, and the record is the strongest single piece of evidence on any inbound link page on this site.

On 18 September 2014, Google issued manual actions — penalties applied by human reviewers and shown in writing to site owners in Search Console — across a large population of private blog network sites. Contemporaneous reporting by Search Engine Land is explicit on the point that decides this page: the penalties hit the network sites themselves, not the sites they linked to.

Sit with what that means. Google conducted its largest documented enforcement action against link networks. The beneficiaries of those networks were not, for the most part, innocent — they were paying customers who had deliberately bought network links to improve their own rankings. If ever there was a population Google had cause to penalize for receiving farm links, it was that one. Google penalized the farms.

A negative SEO victim is in a strictly weaker position than those buyers, in the sense that matters: they did not pay, did not know, and did not benefit. If membership in a network was not treated as proof of guilt for people who actually joined it, the case that it will be treated as proof of guilt for someone who was added to it without knowing is very thin indeed.

Two honest qualifications, because this evidence is load-bearing and I would rather state its limits than have you find them later. It rests on trade reporting rather than on a Google statement: Google declined to comment on the specifics at the time, on its standing policy of not discussing particular manual actions. And whether any recipient sites were also actioned in that wave is not established either way by the sources I can find. What the record shows is where the enforcement was aimed. That is enough to carry the argument, and I will not inflate it into a Google confirmation.

The decade in which the networks were dismantled

The 2013 answer to this question and the 2026 answer are genuinely different, and the change happened in three moves.

March 2012. Google deindexed a wave of paid blog networks — removed them from search results outright — and began issuing unnatural-links notices at scale. The commercial product being sold by those networks stopped working because the sellers stopped existing in the index.

24 April 2012: Penguin 1.0. Google launched Penguin, its link-spam ranking system, affecting a reported 3.1% of queries and explicitly targeting link schemes including link farms. In this first form Penguin applied a site-level demotion. That architecture is the entire reason this attack ever sounded credible: if bad links could drag down a whole site, aiming bad links at a rival was at least a coherent plan.

23 September 2016: Penguin 4.0. Google folded Penguin into its core algorithm, made it run in real time, and changed the handling of spam links from site-level demotion to granular devaluation: the link is discounted and stops counting, rather than the site being marked down.

That third move is where this attack lost its mechanism. Once the algorithmic response to a farm link is to discard the link, inclusion in a farm becomes an event with no output. Not a small effect. No effect, by design.

SpamBrain's definition of a farm is a definition of the seller

The most recent development points the same direction as everything before it.

SpamBrain is Google's machine-learning spam detection system. In the December 2022 link spam update, announced on 14 December 2022 and completed on 12 January 2023, Google extended it so that, in wording quoted from Google in trade coverage of the announcement, it "can now detect both sites buying links, and sites used for the purpose of passing outgoing links."

That second phrase — sites used for the purpose of passing outgoing links — is a description of a link farm. Google now runs a classifier whose explicit job is to identify these networks. And the described outcome is again neutralization rather than punishment of whoever they point at: rankings "may change as spammy links are neutralized and any credit passed by these unnatural links are lost."

Look at the pattern across the whole decade. March 2012: deindex the networks. September 2014: penalize the network sites. September 2016: discount the links rather than demote the recipients. December 2022: build a classifier to identify the networks and strip what they pass. Every documented enforcement action and every documented algorithmic change points at the farm. Not one of them points at the site a farm links to.

The counterweight, stated so you have it: Google's assurances about third-party harm are phrased as effort — the documentation says Google works very hard to make sure actions on other sites do not negatively affect a website — and effort is not a guarantee. There is also a narrow residual case, where a site has almost no legitimate link acquisition and farm links are effectively everything Google has recently seen about it. That is plausible in principle and has never been demonstrated in a disclosed experiment on a site whose farm links were placed by an attacker.

The one-minute check most people never run

Take a distinctive sentence from one of the farm pages linking to you, put it in quotation marks, and search Google for it.

If nothing comes back, that page is not in Google's index. A page that is not indexed passes nothing, to anyone, ever. Many resurrected directory networks were deindexed years before your URL was added to them, which means the links that just appeared in your backlink tool were dead on arrival. This check frequently ends the entire investigation in about sixty seconds, and almost nobody runs it, because the panic arrives through a backlink tool and the backlink tool has no reason to tell you the page it found is invisible to Google.

The backlink tools index pages independently of Google and keep them long after Google has dropped them. That is a legitimate design choice for a link database and a terrible basis for an emergency.

Recognizing a farm injection

Farm inclusion is the most visually obvious of the inbound link attacks, which is convenient for a defender.

• Search Console, Links, Top linking sites. Dozens or hundreds of domains appearing at once, with near-identical templates, near-identical outbound link counts, and often sequential or thematically identical domain names. Check Top linking pages too: farms typically link from a single "links" or "directory" page rather than from content.
• Search Console, Manual actions. The decisive check and the one most site owners skip. "No issues detected" means no human at Google has penalized the site, no reconsideration request is available or appropriate, and the ranking change has another cause.
• Outbound link density. Open five of the linking pages. A page carrying hundreds of unrelated outbound links to unrelated commercial sites is a farm page by definition, and it is as obvious to a search engine as it is to you.
• Shared infrastructure. Reverse-IP and nameserver lookups across a sample of the linking domains. Farms cluster on the same hosts and the same DNS, and this is also the evidence a legal claim or an abuse report would need.
• First-seen dates and anchor text. Genuine directory listings accrete over years; an injection appears in one cluster. Farm submissions use whatever the submitter typed once, so the anchor text is uniform across hundreds of links — a pattern no organic profile produces.

What this gets mistaken for: a core or spam update landing the same week; directory submissions the business genuinely made in 2008 and forgot about; syndicated press releases; scraper sites that copy your content and your links along with it; and legitimate aggregator or comparison networks that are templated because that is how aggregators work. Rule those out before concluding sabotage.

What to do about it

1. Do nothing, deliberately. For the overwhelming majority of cases this is the correct answer. Google's disavow guidance puts it in Google's own words: "in most cases, Google can assess which links to trust without additional guidance, so most sites will not need to use this tool."
2. Check in the right order. Manual actions, then the published update calendar, then technical health, then the link profile. Run those in the wrong sequence and the cleanup routinely does more harm than the injection did.
3. Check whether the farm pages are indexed, as above. Unindexed pages need no response at all.
4. Preserve evidence before anything is removed if a legal claim or abuse report is in prospect: dated Search Console exports, archived copies of the farm pages, reverse-IP and registration records. Networks churn domains quickly and the record goes with them.
5. Disavow only under Google's stated condition — a considerable number of spammy, artificial or low-quality links that have caused, or likely will cause, a manual action. Where an action does exist, domain level is the right granularity for farm links, a documented removal attempt has to come first, and Google warns against blindly adding all backlinks to a disavow file and describes the tool as an advanced feature that can potentially harm a site's performance in search if used incorrectly.
6. Reconsideration request only against an outstanding manual action. With a clean report there is nothing to reconsider.

What does not help: sending removal requests to networks with no working contact address; a link-detox subscription; disavowing entire top-level domains or country domains; adding noindex; blocking Googlebot; rebuilding the site; or buying good links to improve the ratio.

Where the recourse actually is: hosts and registrars

Link farms are unusual among negative SEO vectors in that the most effective complaint is not to Google and not to a court.

Because farms concentrate on a small number of hosting providers and registrars — which is exactly what the reverse-IP lookup above reveals — they can be reported as a bulk terms-of-service violation under the host's own standards, without needing to establish anything about search rankings or about who submitted your URL. A host acting on its own acceptable-use policy moves faster than any other route available here, and it removes the pages rather than merely discounting them. That is the one place effort on this vector reliably produces a result, and it is worth the hour.

Everything else is thinner. Google's spam report process targets the farm, returns no notification to the reporter and is not a remedy; file it as a civic act if you like. Lanham Act section 43(a), 15 U.S.C. section 1125(a), covering false advertising and unfair competition, is the federal theory counsel looks at first where a competitor is behind the placement and the farm pages make false statements about the business or misuse its marks — and note what it turns on: the content published on the farm pages, not the ranking effect. Proving a ranking loss would mean showing what Google's systems did with the links, and no private plaintiff can obtain that. Proving what a farm page published needs nothing but a browser and an archiving service. State-law tortious interference is the conventional alternative and needs a nameable defendant.

I found no decided case establishing liability on a link-farm fact pattern, and no court has held that farm inclusion moved a ranking. Attribution is the practical wall: farm operators are typically offshore, pseudonymous and judgment-proof, and whoever submitted your URL is usually untraceable even when the network is not.

The disavow mistake that does the real damage

• Disavowing legitimate directories in the same sweep. This is the specific damage I see most often on this vector. Trade association listings, chamber of commerce directories, supplier registers and genuine industry directories are real editorial links that a business earned, and a pattern-matching disavow file aimed at "directories" destroys them alongside the junk. The recovery is slow and uncertain.
• Mass disavowing on suspicion generally, which turns a non-event into a real loss.
• Filing reconsideration requests with no manual action outstanding, then reading the silence as Google ignoring them.
• Treating a vendor Toxic Score or Spam Score as a Google signal. Google publishes no such metric and never has. Those scores do move when a farm injection lands, which is a genuine commercial nuisance when an agency or an acquirer runs the report, but it is not a ranking event.
• Paying a monthly fee for protection. Anyone selling you ongoing defense against being added to a link farm is selling protection against a mechanism that Google's spam policies describe as a practice of creating links, and that Google's own enforcement history aims at the network rather than at you. That is a subscription against somebody else's risk.
• Naming a competitor publicly, which trades a search problem that was probably imaginary for a libel problem that is not.

Frequently asked questions