Link spam updates
that changed everything.

Google's Link Spam Update and SpamBrain: What Every Link Builder Needs to Know

Google updates its algorithm thousands of times a year, most of them minor and barely noticed. The December 2022 Link Spam Update was different. It was a direct, explicitly announced assault on the economics of the paid link market — powered by an AI system called SpamBrain — and it had concrete consequences for anyone whose link building programme relied on the kinds of sites that have proliferated across the web over the past decade.

This guide explains exactly what the update did, how SpamBrain detects link spam, what it means for your link building strategy going forward, and where legitimate questions about its scope and precision remain unanswered.

What the Link Spam Update Actually Did

On December 14th 2022, Google announced a significant change to how its algorithm handles paid and manipulative links. The core of the announcement was a statement about SpamBrain — Google's AI-based spam prevention system — and two new capabilities it had developed:

• The ability to detect sites that are buying links
• The ability to detect sites being used specifically to pass outgoing links to others

Translated into practical terms: Google's system can now more reliably identify both sides of a paid link transaction — the site paying for links and the site being paid to publish them — and neutralise the ranking value those links were providing.

What the update does and does not do:

The distinction between neutralising and penalising matters. This is an algorithmic update, not a manual action campaign. Sites affected by it will not see a notification in Google Search Console, will not receive a manual action, and will not be told which links triggered the devaluation. The practical effect is simply that links Google has identified as spam stop contributing to rankings — silently, without recourse, and potentially retrospectively.

SpamBrain: How Google's AI Detects Link Spam

SpamBrain has been part of Google's anti-spam infrastructure for several years, but the 2022 update significantly expanded its capability specifically in the link detection domain. Understanding how it works helps explain both why the update is significant and why the low-quality link market has become increasingly risky.

The patterns SpamBrain is designed to detect are not particularly subtle. Any reasonably well-trained machine learning system working on link data would have no difficulty spotting most of them. The red flags that flag a site as a link seller include:

Site-level signals:

• Expired or repurposed domains relaunched with thin content to generate outbound links
• Sites publishing content across multiple unrelated categories with no coherent editorial identity
• A disproportionate ratio of outbound links to inbound links
• Low-quality content with no consistent readership or engagement signals
• "Write for us" or "sponsored post" pages with minimal editorial standards
• Site designs and templates common to known link networks

Link-level signals:

• A pattern of exact-match commercial anchor text pointing to service and money pages
• Links going consistently to the same commercial niches — legal, finance, casino, insurance — from sites with no natural editorial reason to cover those topics
• Outbound links clustered around a small number of recurring commercial domains
• Link placement that reads as paid rather than editorial — abrupt, contextually irrelevant, often in the middle of unrelated content

Network-level signals:

• Sites appearing in the same link neighbourhoods as known link sellers
• Overlapping IP ranges, hosting patterns, or ownership footprints across networks of sites
• Sites that receive a sudden influx of outbound links after a period of no publishing activity

Google almost certainly maintains internal classifications of sites it has determined are link sellers with high confidence, distinguishing between definite offenders (which it devalues entirely) and borderline cases (where the AI is less certain and the response may be more nuanced).

The Helpful Content Update: The Other Half of the Picture

The timing of the Link Spam Update was not coincidental. It arrived nine days after Google rolled out the Helpful Content Update on December 5th 2022 — and the two updates together represent a coordinated attack on the dominant forms of manipulative SEO at that time.

The Helpful Content Update targeted a different problem: content produced primarily to rank rather than to serve a genuine reader. This includes heavily template-driven content, articles optimised for keyword density with little original insight, and content produced at scale by AI writing tools with minimal human oversight. Google's guidance specifically described it as favouring content "written by people for people."

The combined effect of both updates:

A site that had been propped up by both manipulative link building and thin SEO content could have been hit by both simultaneously — making it particularly difficult to isolate which update caused a given traffic drop. This is one reason Google's standard advice after any update is to avoid knee-jerk reactions: when multiple updates roll out in close succession, attributing traffic changes to a single cause is genuinely unreliable.

How to Respond to the Update

The most important thing most sites should do in response to the Link Spam Update is: nothing immediate. The update rolls out over several weeks, attributing traffic changes to it during that window is difficult, and reactive decisions — particularly around disavowing links — can cause more harm than the update itself.

On disavowing:

John Mueller of Google addressed this directly. He noted that Google is now very effective at simply ignoring spam links, and that disavowing is no longer necessary in most cases where links are merely low-quality rather than actively manipulative. The recommendation to disavow applies in narrow circumstances:

• You have received a manual action specifically related to your link profile
• You have verifiable evidence that a concentrated negative SEO attack is causing ranking harm
• You built a PBN or large-scale manipulative link campaign and want to clean up before a manual review

For most sites that have simply accumulated some low-quality links over time — whether from past SEO providers, organic spam, or minor paid placements — Google's preference is to ignore those links rather than to penalise around them. Filing an aggressive disavow in response to this update risks accidentally removing links that are contributing positively, which can suppress rankings just as effectively as the update itself.

What Links to Build Going Forward

The practical consequence of the update is clear: high-quality links from legitimate sites with real editorial oversight became more valuable, and low-quality paid placements on obvious link farms became effectively worthless — while still costing money. The update does not change what good link building looks like, but it does sharpen the penalty for cutting corners.

Links worth building:

The last point deserves elaboration. The update targets sites whose primary or dominant function is link selling. A well-established publication with genuine organic traffic, strong editorial standards, and a history of original content is not going to be devalued simply because it accepts occasional sponsored posts. Google's system is calibrated to identify sites built around link selling, not to penalise every site that has ever accepted payment for any content.

Links to avoid:

Assessing Whether You Have Been Affected

Because the update is algorithmic rather than manual, there is no direct notification when it affects a site. Diagnosing impact requires inference from available data.

How to assess whether the update affected your rankings:

• Check Google Search Console for traffic drops in the weeks following mid-December 2022, filtering specifically by organic search
• Cross-reference the timing against the Helpful Content Update rollout (early December) — if your traffic was already declining before December 14th, the Link Spam Update may not be the primary cause
• Use Ahrefs or SEMrush to examine your backlink profile for domains that are clearly link farms — high outbound link counts, thin content, no organic traffic, obvious commercial link patterns
• Compare your Domain Rating or Authority Score trend against the link volume trend — if DR has stalled or dropped despite ongoing link building, some links may have been devalued
• Look for pages that had previously benefited most from link building and check whether their specific ranking positions declined around the update window

One important limitation: the update neutralises value rather than introducing a penalty, meaning there is no negative signal to detect directly. The effect is simply the absence of positive ranking impact from links that were previously contributing. Distinguishing that from other ranking volatility requires careful timeline analysis.

Where Legitimate Questions Remain

The update's announcement was clear about what Google intends. Some practical questions about how the system operates in ambiguous cases are genuinely unresolved — and honest analysis requires acknowledging that.

The grey area problem: SpamBrain is designed to classify sites with high confidence before devaluing their links. But the link selling market is not binary. Many sites sell occasional links while also running genuine editorial operations. At what threshold does occasional link selling trigger devaluation? A site that sells five links a year is different from one that sells five hundred, but the line between them is not publicly defined. Google's AI presumably applies a confidence threshold and ignores links from sites it cannot classify clearly, but the specifics of that threshold are unknown.

The false positive risk: Any pattern-matching system trained on known spam sites will produce some false positives — legitimate sites with characteristics that overlap with spam signals. A genuine editorial site that publishes across multiple topics, has a high outbound link count from a resource page, and happens to be in a niche commonly associated with link buying could theoretically exhibit several red flags without selling a single link. Whether Google's system is sophisticated enough to avoid these misclassifications at scale is not verifiable from the outside.

The detection lag: Google has confirmed it can detect link selling, but the announcement frames it as a new capability — implying that the detection was less reliable before December 2022. Links that were purchased and counted before the update may have contributed to rankings that are now permanent in ways that predate the devaluation. Sites that benefited from historical link buying may retain those gains unless the devaluation is applied retroactively and retrospectively to existing link profiles. Google's language suggests it can nullify "previous value" these links were providing, but the degree to which retrospective devaluation is applied in practice remains unclear.

The Broader Trend: What SpamBrain Signals About the Direction of SEO

The Link Spam Update is best understood not as an isolated event but as a data point in a long-term trajectory. Google has been systematically improving its ability to detect and discount manipulative link signals for over a decade, from the original Penguin update in 2012 through successive iterations and into the AI-powered detection capabilities SpamBrain now represents.

Each improvement in detection narrows the gap between the link building strategies that work and those that merely appear to work temporarily. The strategies most vulnerable to each successive update have been those that rely on volume and automation rather than genuine editorial relationships: link farms, PBNs, bulk directory submissions, and now AI-detectable paid link networks.

The strategies that have remained durable across every update share a common characteristic: they are built on real relationships, real content, and genuine editorial decisions by third parties who are not simply being paid to publish a link. Guest posts on publications that would reject poor pitches. HARO responses that earn coverage through the quality of expert insight. Broken link building that fixes a real problem on the linking site. Linkable assets that attract citations because they contain information people actually find useful.

None of this is new advice. But the practical effect of each SpamBrain update is to make the cost-benefit calculation for low-quality link building progressively worse — more cost, less benefit — while the economics of high-quality link building remain intact.

Ready to Build Links That Survive Algorithm Updates?

If the Link Spam Update has called into question the quality of your current link profile, or if you want to build a link acquisition programme that is genuinely durable rather than algorithmically vulnerable, get in touch at [email protected]. We are happy to review what you are working with and discuss an approach focused entirely on links that will still be contributing to your rankings in five years.