Black hat vs white hat SEO: the real line, the named violations, and what actually gets a site demoted

White hat SEO improves rankings by making a site genuinely more useful to the person searching, and stays inside the practices Google documents. Black hat SEO improves rankings by deceiving the ranking system, and every major technique in it is named in writing in Google's spam policies for web search. The dividing line is not how aggressive the tactic is. It is whether the tactic depends on the search engine being shown something different from what the reader gets. Cloaking, doorway pages, hidden text, keyword stuffing, link spam, sneaky redirects, scraped content, scaled content abuse, expired domain abuse and site reputation abuse are all on the list by name, and the consequence is a demotion or a removal from results.

This page covers the definitions, the named violations, the consequences and the grey areas. For the tactics that build rankings the durable way, see a guide to on page SEO and one to link building.

The practices Google names

This is the useful part, because the policy document is specific rather than vague. Each of these has its own section in Google's spam policies.

Named practice	What it looks like in the wild
Cloaking	Serving different content to the crawler than to the visitor, by user agent or by IP
Doorway pages	Hundreds of near identical pages, usually one per town or one per keyword, funnelling into the same destination
Hidden text and links	White text on white, text behind an image, a font size of zero, links in a single character
Keyword stuffing	Lists of cities, or the same phrase repeated until the sentence stops being a sentence
Link spam	Buying or selling links for ranking purposes, large scale exchanges, automated link creation, keyword rich guest article campaigns at volume
Sneaky redirects	Sending the crawler to one URL and the human to another, or redirecting mobile visitors somewhere unrelated
Scraped content	Republishing somebody else's pages, with or without a light rewrite
Scaled content abuse	Generating large numbers of pages primarily to rank rather than to help anyone, whether by automation, by hand, or both
Thin affiliate pages	Affiliate templates with the vendor's description and nothing of your own
Expired domain abuse	Buying a lapsed domain for its old links and repurposing it for unrelated content
Site reputation abuse	Renting out a section of a trusted site to third party content that trades on the host's reputation
User generated spam	Comment, forum and profile spam left to accumulate on a site you run
Misleading functionality	A tool or calculator that does not do what the page says it does
Machine generated traffic	Automated queries against Google, including rank scraping at scale
Hacked content and malicious practices	Injected pages and links from a compromise, and malware or unwanted software

Read the policy page itself once. It takes fifteen minutes and it ends most internal arguments about whether something is acceptable.

What white hat actually consists of

White hat is not a softer version of the same tricks. It is a different list of jobs.

• Technical access. Crawlable, indexable, fast, one canonical version of each page, a current sitemap, working redirects.
• Pages that answer a real search better than the pages ranking today, written by someone who knows the subject. Google's own standard for this is set out in its guidance on helpful, reliable, people first content.
• Clear information architecture and internal links, so the important pages are the easiest to reach.
• Earned links and mentions, from original data, tools, press commentary, partnerships and real industry bodies.
• Structured data that describes what is genuinely on the page, which is also where a lot of people drift over the line by marking up reviews or FAQs that no visitor can see.
• Honest labelling, which means rel="sponsored" on paid links and rel="ugc" on user generated ones.

Algorithmic demotion and manual actions are different problems

• An algorithmic demotion is a ranking system treating your pages as low value. Nothing appears in Search Console. Traffic falls, often alongside a documented update. The fix is to change what caused it and wait for re-evaluation, which takes weeks to months.
• A manual action is a human reviewer applying a penalty. It appears in the Manual actions report in Search Console, naming the issue and whether it hits part of the site or all of it. You fix the issue, document what you did, and file a reconsideration request.

Check the Manual actions report first on any traffic collapse. It takes ten seconds and it tells you which of these two conversations you are having.

Three questions that settle any grey area

Most arguments about "grey hat" dissolve under these.

• Does the crawler see something the visitor does not? If yes, it is on the black hat list, whatever the technique is called.
• Would I be comfortable if my best customer read a plain description of what we did? Paid links through a network fail this. A sponsored article labelled as sponsored passes.
• Would this still be worth doing if search engines did not exist? Original research, a free tool, a clearer product page and a faster site all pass. A hundred city pages with the town name swapped do not.

The grey areas, answered plainly

• AI written content. Google's guidance on AI generated content says the issue is content produced primarily to manipulate rankings, not the method of production. Content made at volume to fill a keyword list is scaled content abuse whether a person or a model typed it. An expert using a model to draft faster, then adding what only they know, is fine.
• Guest posting. Writing for a publication with real readers and a real editor is marketing. Buying placements across a network of thin sites is link spam, and the policy names keyword rich article campaigns at scale.
• Programmatic pages. Generating a page per data record is legitimate when each page carries genuinely different and useful information, such as a real listing or a real dataset. It becomes doorway territory when only the town name changes.
• Buying an expired domain. Buying a domain whose brand you are continuing is normal business. Buying it for the link profile and putting unrelated content on it is named in the policy.
• Exact match anchor text. Natural when a writer chose it. A signal of a campaign when four hundred links share the same five words.

black hat vs white hat SEO: what readers write in to ask

Is grey hat SEO a real category?

It is a marketing term rather than a category Google recognises. The policies describe practices, and a practice is either in them or it is not. When somebody describes their work as grey hat, ask them which named practice they are near and why they think they are on the safe side of it.

Is buying backlinks black hat?

Buying links for ranking purposes is named in the link spam section. Paying for a placement and labelling it with rel="sponsored" is advertising, and it is fine. The label is the whole difference.

Does keyword stuffing still work?

No, and it is still one of the most common reasons a page that reads badly also ranks badly. Write the sentence a person would say out loud, then check the page mentions the subject in the title, the first paragraph and one subheading.

How long does recovery take?

A manual action can be lifted within days or a few weeks of a reconsideration request once the issue is genuinely fixed. An algorithmic demotion moves on Google's re-evaluation schedule and is usually a matter of months, so fix the cause thoroughly rather than in instalments.

Can a competitor get my site penalised?

Spam links pointed at a site you do not control are the case Google's systems are built to ignore, so this happens far less than the industry folklore suggests. Watch the Manual actions report, keep your own link building clean, and reserve the disavow tool for a manual action you can trace.

What should I check on a site I have just inherited?

The Manual actions report, the link profile for sudden historical spikes, the page count against the number of pages a human wrote, any text hidden in the markup, the redirect rules, and whether anyone is publishing third party content on a subfolder. That sequence finds almost everything.

Grey areas with a fresh audience

The three questions that settle a grey area are about intent, disclosure and value to the reader. They apply unchanged to writing aimed at assistants. A page stuffed with invented claims for a model to quote is the old trick in a new place, and the policies Google names stay relevant. The clean route is plain pages that answer a question and can be checked by anyone. AI Knows Us writes pages and places businesses on sites the assistants read, and the same three questions apply to what it writes.