Gerris home
Gerris

Chris Abraham: technical consulting, AI tools, and technical SEO

Home › Guides › Black Hat in Modern Dress

Black hat SEO in modern dress: the tactics that end sites

Practitioners my age buried black hat SEO a decade ago and attended the funeral in good faith. The corpse changed clothes. Every scheme Google killed between 2003 and 2024 has a living descendant running today, rebuilt for answer engines, and clients arrive at Gerris with the descendants already installed. I keep this taxonomy current because diagnosis starts with naming the thing correctly.

Doodle: a black hat banded in red, green, and blue, the old black hat in modern dress

On this page
  1. Tier one: the antiques, still for sale
  2. Tier two: the plausible failures
  3. Tier three: the named offenses
  4. How the descendants get caught
  5. The working alternative

Tier one: the antiques, still for sale

Keyword stuffing survives as entity stuffing: pages salted with entity names and attribute pairs to impress embedding models, prose no human finishes. Doorway pages survive as programmatic location pages, five hundred city variants of one template with the nouns swapped by script. Article spinning survives as paraphrase networks, the same essay rewritten by a model forty ways and posted across forty domains. Comment spam survives as seeded forum accounts that mention a brand with rehearsed casualness. Tier one wastes crawl budget and calendar time. It rarely kills a site. It reliably marks the operator as an amateur in a field that now reads code.

Tier two: the plausible failures

Hidden text survives as hidden instruction. Pages now carry passages written for AI crawlers, directives in markup and metadata telling answer engines how to describe the business, invisible to the customer and legible to the machine. Forged expertise survives beside it: author personas with generated portraits, invented credentials, and biographies assembled to simulate the experience signals Google calls E-E-A-T. Citation stacking rents mentions in listicles built for engines rather than readers. Review velocity gets managed the way link velocity once was, in bursts timed to look like weather. Tier two fails slowly and then publicly. The patterns are statistical, the engines are statisticians, and a fabricated record contradicts itself somewhere crawlable.

Tier three: the named offenses

Google wrote this tier into policy in March 2024, and practitioners should use the official names. Scaled content abuse: content generated at volume to manipulate rankings, human or machine authorship irrelevant. Site reputation abuse: third party pages published on a trusted domain to exploit its signals, the coupon directory and the casino review squatting on a newspaper's authority. Expired domain abuse: a lapsed domain with a history, purchased and refilled to inherit trust it never earned. Cloaking completes the set, now aimed at AI crawlers as often as at Googlebot, one text for machines and another for people. Tier three ends in manual actions and removal. Recovery takes quarters, and my guide to diagnosing a traffic drop starts from the wreckage these tactics leave.

How the descendants get caught

Clients ask why the modern versions fail when the fluent output reads so well. The answer is arithmetic. Schemes must scale to pay, scale leaves distributions, and distributions confess. A paraphrase network shares sentence skeletons across domains. A persona stable shares a biography template, a headshot generator, and a posting rhythm. A parasite section shares a template, a monetization path, and a landlord. Detection no longer reads pages one at a time looking for guilt; it clusters, and the cluster is the evidence. The practitioner who installs a descendant inherits the whole cluster's record, including the parts other operators contributed. That inheritance is the part the sales pitch leaves out.

The working alternative

E-E-A-T compounds; schemes depreciate. A named expert publishing first hand work on a current site gives the engines evidence they can verify and rivals cannot rent. Currency matters more each quarter, because Google's AI answers from the live index while the chatbots answer from training memory months stale. Fresh, accurate, attributable pages win the live arena outright. The audit trail runs both ways: a clean site's history is visible too, in consistent authorship, corrections made in public, and claims that survived their news cycle. Engines read that history when they choose whose answer to quote. The rebuild Gerris runs after a descendant is found is always the same shape: fewer pages, named authors, current facts, and claims the business can defend in the room. Slow work, honestly labeled, and built to be audited by anyone, including the competition. Retire the antiques, refuse the plausible failures, and treat tier three as a firing offense for any vendor who proposes it.

References