What Is Black Hat SEO?
This page explains what each risky practice is and why it is a risk. It does not explain how any of them is done. That is deliberate. Almost nobody arrives here planning to cheat. They arrive because something they were offered sounded odd, else because they took over a site and want to know what is in it.
Why You Are Reading This
Most people who look this up are in one of two situations. Somebody has proposed work that sounded slightly wrong, else a site has changed hands and nobody is certain what the previous arrangement involved.
The first situation. A proposal you cannot quite assess.
An agency describes something in terms you do not recognise, promises a result faster than seems reasonable, else becomes vague when asked how it works. That instinct is worth taking seriously. Vagueness about method is the single most reliable warning available to a buyer.
The second situation. An inheritance.
You bought a business, changed agency or took over a site somebody else built. Something in it looks odd and nobody who made the decision is still around to explain it.
What this page is written to do. Help you recognise, rather than catalogue.
It names each practice, explains what makes it a risk and stops there. You will not find instructions here, because a page that told you how to do these things would be useful to exactly the wrong reader.
What it is not. An accusation.
Finding one of these in your own site is common and it is rarely anybody's fault who is still involved. Block twelve deals with what to do about it.
What Makes A Practice Black Hat
Lists of banned techniques date badly, because new ones appear constantly. A test does not. Almost everything described on this page fails one of two questions.
The first test. Is this showing search engines something different from what it shows people.
Anything built so that a machine sees one version and a visitor sees another is deception by design. It does not matter how it is achieved or how sophisticated it is. The intent is what defines it.
The second test. Is this manufacturing a signal that has not been earned.
Search engines infer quality from things that are difficult to fake, such as other people choosing to reference you. Practices that fabricate those signals rather than earning them fail this test, whatever they are called.
Why the test is more useful than a list. It covers what has not been invented yet.
A technique with a new name still has to pass both questions. If you can apply the test, you can assess a proposal nobody has written an article about.
What the test is not. A judgement about difficulty.
Plenty of legitimate work is technical, aggressive and competitive. Being clever is not the problem. Deception and fabrication are.
Showing Different Things To Different Visitors
A site can be arranged so that what a search engine is shown differs from what a person is shown at the same address. The practice is usually called cloaking. It is the clearest example of the first test failing.
What it amounts to. One address, two realities.
Search engines rank what they were shown. Visitors receive something else. Every ranking that follows is based on a page the visitor never sees.
Why it is treated as deception rather than as a technical choice. The gap is the point.
Sites legitimately vary what they show, by language, by device or by whether somebody has logged in. None of that is this. The distinction is whether the variation exists to serve the visitor or to mislead the machine.
What happens to sites doing it. Treated as a policy breach.
It sits squarely within published spam policy rather than in a grey area, which is covered on the Google spam policy page. The consequences fall on the site, not on whoever was paid to arrange it.
How a buyer recognises it. A reluctance to explain.
Nobody proposing this will describe it plainly. If an explanation of why your pages will rank cannot be given in ordinary language, that is the finding.
Text Nobody Is Meant To Read
Words placed on a page so that a search engine encounters them while a visitor does not. The principle is the same as the previous block, applied to content rather than to whole pages.
Why it fails the test. The audience is a machine.
Content written for a visitor and content hidden from a visitor are different things with different purposes. The second exists only to influence what a search engine believes about the page.
How most readers will actually meet it. As an inheritance.
This is an old practice. The sites still carrying it are mostly carrying it by accident. A page built years ago, a template bought from somewhere, an agency arrangement that ended long before the current owner arrived.
Why that matters to how you respond. It changes the question.
An inherited instance is a tidying job rather than a scandal. Nobody currently involved decided to do it. The useful response is to establish what is there and remove it rather than to work out who to blame.
The adjacent thing that is not this. Text for accessibility.
Descriptions provided for screen readers are not hidden text in this sense. They exist to serve people who cannot see the page, which is the opposite intent.
Repeating Words To Signal Relevance
Loading a page with the phrase you want to rank for, on the theory that more mentions signal more relevance. Usually called keyword stuffing. It is the oldest manipulation in the subject.
What people picture. The crude version.
Paragraphs so laden with a phrase that they read as nonsense. That version is largely extinct, which is why businesses assume this no longer applies to them.
What actually persists. Something subtler.
Copy that has been edited to include a phrase more often than a person would naturally write it. Nothing reads as absurd. It simply reads as slightly stiff, slightly repetitive, written for a machine by somebody being careful.
Why writing naturally is the actual answer. Not a stylistic preference.
Search engines have understood related terms and meaning rather than exact repetition for a very long time. Writing the way a knowledgeable person would explain something to a customer is not a softer approach. It is the approach that works.
How to spot it in your own copy. Read it aloud.
Anywhere you stumble, else where a sentence names your service and your town more often than a person would, was written for the wrong reader.
Networks Built To Point At Each Other
Collections of websites that exist principally to link to other sites. Usually described as private blog networks. They fail the second test completely.
What they are for. Manufacturing references.
Search engines treat a link as somebody choosing to point at you. A network removes the choosing, because the same interest controls both ends. Whatever the sites look like individually, the reference was not earned.
Why the risk sits with you rather than with the network. The uncomfortable part.
A business buying into one of these is usually told the network is the thing at risk. In practice a site receiving manufactured links carries the consequence. The network operator moves on to the next client.
Why they are detectable. Patterns are what search engines are built to find.
We publish nothing about how detection works. The relevant point for a buyer is that this is a well studied category rather than a clever secret. The people selling it know that.
Where the fuller treatment sits. The link risk page.
Including what to do about the emails everybody receives, in what is link farming in SEO. How links are legitimately earned belongs in our backlinks material rather than here.
Borrowing Somebody Else's Standing
Publishing on a site that already carries weight, in order to rank using its standing rather than your own. Usually called parasite SEO. It has become considerably more visible in recent years.
What changed. Established publishers began renting out space.
Sections of well known sites have been let to third parties producing commercial content unrelated to the publication. The content ranks on the strength of the host rather than on its own merit.
Why it is a risk to the tenant. Nothing is yours.
Rankings held by somebody else's standing disappear when the arrangement ends. They do end. Whatever was spent bought a tenancy rather than an asset.
Why it is a risk to the host. The larger exposure.
A publication renting its standing puts its own position at stake for a share of somebody else's revenue. Several have discovered that the arrangement was worth less than what it cost them.
Where this is addressed in policy. Under site reputation abuse.
It is named directly rather than being inferred. The attributed account is on the Google spam policy page.
Pages Built Only To Catch A Search
Pages that exist to intercept a query rather than to serve the person who arrives. The textbook example is a set of near-identical pages differing only by the name of a town. This is also the practice sold most often to ordinary businesses.
Why this block is different from the others. We build town pages.
This agency produces location pages for clients. Any page on this subject that avoided saying so would not be worth reading. So here is the line, drawn where we actually draw it.
What makes it a doorway page. The town is the only variable.
Take the page and swap the place name. If nothing else meaningfully changes, the page was built for a search engine. Producing forty of those does not create forty pages. It creates one page and thirty-nine copies.
What makes it a location page. Something that exists nowhere else.
Genuine detail about working in that place: what the housing stock is like, what the common jobs are there, where the coverage boundary actually falls, which nearby areas are included, what is different about serving that area.
The test to apply. One unforgiving question.
Would somebody who lives in that town find this page worth reading. Not tolerable, not adequate. Worth reading. If the answer is no, it is a doorway page whatever it is called on the invoice.
Why the volume matters as much as the content. Scale is assessed too.
Producing large numbers of pages primarily to catch searches is addressed directly in published policy under scaled content abuse. Whether a person or a machine wrote them is explicitly not the point. That is set out with attribution on the Google spam policy page.
Sending People Somewhere They Did Not Choose
Arranging for a visitor to arrive somewhere other than the page they selected. This one needs care, because the underlying mechanism is entirely legitimate and used constantly for good reasons.
The legitimate version. Necessary, also everywhere.
When a site moves, when pages are reorganised, when two pages become one, redirects preserve what was there and take people to the right place. Without them a site rebuild would destroy everything it had earned. That is the whole subject of our site migrations material.
What makes a version of it manipulative. Intent and disclosure.
A redirect exists to serve somebody who asked for a page that moved. The manipulative version sends somebody somewhere they did not ask for, else shows search engines one destination while sending people to another. That returns to the first test in block two.
Why the adjacency is worth stating. Nobody should be nervous of redirects.
Businesses read about this and conclude that redirects are risky in general. They are not. Failing to use them during a rebuild causes far more damage than using them ever has.
The question that separates the two. Would you explain it to the visitor.
If a person arriving would consider the destination reasonable given what they clicked, it is a redirect. If they would feel misled, it is not.
Clicks That Never Happened
Advertising clicks generated by something other than a person genuinely interested in the advert. This belongs to paid advertising rather than to organic search, so it is included briefly for completeness rather than covered in depth.
Why it appears in this cluster at all. People ask.
It gets grouped with organic manipulation because both involve fabricated signals. The mechanisms and the consequences are entirely different. Conflating them causes confusion about what a business is actually exposed to.
The two directions it runs in. Both worth knowing.
Clicks generated against a competitor's advertising to exhaust their budget. And clicks generated by whoever hosts advertising, to increase what they are paid.
What a business should actually do. Rather less than the anxiety suggests.
Advertising platforms filter invalid activity and credit it back, because they have a direct commercial interest in advertisers trusting the numbers. Unusual patterns are worth raising with the platform. They are rarely worth building a strategy around.
Where this will be covered properly. Not here.
This belongs with paid advertising material rather than with organic risk. It is on the list to be revisited when that cluster is built.
Why None Of This Is Worth It
The usual case against these practices is that they get punished. That is true and it is the weaker argument. The stronger one is that they build nothing.
What legitimate work leaves behind. Things you own.
Pages that answer what customers ask. References other people chose to give you. A site that converts. Those persist through algorithm changes, through agency changes and through you deciding to do nothing for six months.
What manipulation leaves behind. Nothing.
Rankings held by a technique last exactly as long as the technique does. Each of these has eventually stopped working, at which point the business is back where it started having paid for the interval.
Why that framing matters more than the risk of penalty. It removes the gamble.
Presented as a risk, this becomes a calculation about whether you will be caught. Presented accurately, there is nothing to weigh. You are choosing between renting a position and owning one.
The version of the question worth asking a supplier. Direct.
If this works, what do we still have in three years when it stops. Anybody who has a good answer will give it immediately. The reasoning behind that question sits in what is grey hat SEO.
What To Do If You Find It On Your Own Site
Discovering one of these in a site you own is a normal Tuesday rather than a crisis. Sites accumulate history, agencies change, then practices that were ordinary a decade ago are not now.
First, establish whether anything has actually happened. Before anything else.
Finding a questionable practice is not the same as having a problem. Some sites carry these things for years with no visible consequence. Start with how to check for a Google penalty, which resolves the definite part in about a minute.
Second, find out what is actually there. Rather than assuming.
People frequently discover one thing, assume the worst about everything else and start removing material that was working. Establish the scope before touching anything.
Third, remove it properly. Partial does not count.
Half-removing something is the commonest reason a recovery attempt fails. If it is going, it goes completely. The reasoning is on the recovery page.
What not to do. Panic.
Mass deletion, wholesale rewriting and sudden structural change during an unexplained situation all destroy your ability to work out what was wrong. Nothing here needs fixing this afternoon.
Who to ask. Somebody without an interest in the answer.
Preferably not whoever installed it, nor somebody selling recovery services before they have established there is anything to recover from. The full series is on the dangerous SEO practices guide.
Somebody
offered you
something odd.
A review that tells you what is actually in your site, what carries risk and what does not, written so you can act on it rather than worry about it. We will tell you plainly if we find nothing, because most sites we look at have nothing worth removing.
What a risk review covers:
A risk review is a one-off piece of work rather than a monthly service, so it is scoped against the size of the site and how much history it carries.
Every guide.
One practice.
Black hat, grey hat, link risk, thin content, the spam policies, negative SEO, manual actions, algorithmic penalties, how to check for one and how to recover from one.
