SEO Bandwagon · Reference
36 SEO terms, defined and dated. A glossary usually tells you what a term means. This one also tells you whether anyone still uses it — we measured every entry across 25 years of evidence, and 12 of them are effectively obsolete.
Dates come from our study History of SEO Terms on Wikipedia, which measured every term below across 6,083 article-year snapshots. Peak is the year a term was used most; retained is how much of that peak frequency survives today.
The text alternative on an image, read aloud by screen readers and shown when an image fails to load. It is an accessibility requirement first and an SEO signal second — the practice endures even though the phrase itself has dropped out of the study corpus, last seen there in 2023.
The percentage of a page’s words that are a given keyword. Once treated as a number to hit — guidance circulated for years suggesting 2–3% — it was never a published ranking factor, and modern retrieval scores relevance by meaning rather than repetition. Writing to a density target today produces worse copy and no ranking benefit.
An HTML meta tag listing terms a page wanted to rank for. Because publishers filled it with whatever they liked, it became worthless as a signal; Google confirmed in 2009 that it does not use it for ranking. It is the clearest example of a tag that survives in page source long after it stopped meaning anything.
Elements in a page’s <head> describing it to browsers and crawlers. Some still matter — the description influences what appears in results, robots directives control indexing — but the phrase is a relic of an era when filling in tags was most of what on-page optimisation meant.
The HTML element giving a page its title, shown in the browser tab and used as the basis for the headline in search results. Still one of the few on-page elements with direct, observable influence — though since 2021 Google frequently rewrites it when it judges the page’s own title unhelpful.
The visible, clickable words of a link. It tells both readers and crawlers what the destination is about, which made it a target for manipulation — and is why exact-match anchors at scale now read as a spam signal rather than a relevance one.
Links from other sites to yours. Still among the strongest signals in search, and still the hardest to obtain honestly, which is why almost every abuse term in this glossary is a scheme for manufacturing them.
The practice of acquiring backlinks deliberately. The legitimate version is publicity by another name — producing something worth citing and telling the right people about it. The illegitimate version is buying links, and the distinction is enforced.
Informal term for the ranking value passed along a link. Never official terminology — it is the trade’s own metaphor for PageRank flow, and its rise and fall tracks how much of the industry’s thinking was once organised around plumbing.
A link attribute introduced by Google, Yahoo and Microsoft in 2005 to let publishers disclaim endorsement, principally to defuse comment spam. Since 2019 it is a hint rather than a directive, and it has been joined by the more specific rel="sponsored" and rel="ugc".
Unpaid results, earned by relevance rather than bought. The distinction between organic and paid is older than SEO itself and remains the cleanest line in the discipline.
The generic spelling, used loosely to mean “how well a page ranks”. Worth separating from PageRank the algorithm: the two-word form is a description of an outcome, the one-word form is a named method. The trade largely abandoned the generic spelling once the toolbar score vanished.
The algorithm named for Larry Page that scored a page by the number and quality of links pointing at it, treating a link as a vote weighted by the voter’s own standing. It is still part of how Google ranks; what disappeared was the public toolbar score, retired in 2016.
A page’s position for a given query. The phrase dates from when a ranking was a single stable number; personalisation, localisation and AI surfaces have since made “your ranking” a distribution rather than a value.
What a search engine returns for a query. The plain-English form of SERP, and the phrasing that outlasted the acronym in general writing.
Search engine results page — the page of results returned for a query. Once a simple list of ten links, now a composite of ads, local packs, featured snippets, video, and AI-generated summaries, which is why “position one” means far less than it used to.
The plural, used as shorthand for the results landscape generally. The plural form was heavier trade jargon than the singular and has faded faster — the singular survived by becoming an ordinary noun.
Techniques that violate search engine guidelines. The hat metaphor is borrowed from security research and, like most such borrowings, flattens a spectrum into a binary that practitioners then argue about endlessly.
Serving different content to crawlers than to human visitors. A direct policy violation and grounds for removal from the index, because it attacks the assumption that what a search engine indexes is what a reader receives.
Pages built to rank for a query and funnel visitors elsewhere, offering nothing themselves. The tactic survives as mass-generated location pages, which is why the modern policy language targets scaled content abuse rather than the original term.
Keywords hidden from readers but visible to crawlers — classically white text on a white background. The technique described in Wikipedia’s 2002 spamdexing article, and the crudest form of keyword stuffing.
Loading a page with repeated keywords to force relevance. Among the oldest recognised abuses, explicitly named in search engine spam policies, and reliably counterproductive since at least the mid-2000s.
A group of sites created solely to link to each other and inflate apparent authority. One of the earliest recognised link schemes, and among the first things search engines learned to detect and penalise.
Paying a search engine to index or list a page. Once an openly sold product — Yahoo and others ran formal programmes — and now essentially extinct in general web search, which is why the term has all but vanished.
Links bought for ranking rather than traffic. A direct violation of search engine guidelines unless disclosed with rel="sponsored", and the reason the nofollow attribute family exists at all.
A mutual exchange — you link to me, I link to you. Common enough in the early 2000s to have dedicated directories and software; now understood as a scheme when done at scale, and simply unremarkable when it happens naturally.
Techniques that comply with search engine guidelines. Defined entirely by its opposite, which is why the term is far less useful than it sounds — it describes an absence of misconduct rather than a method.
Systems performing tasks that once required human judgement. In search it now describes both how results are ranked and how they are presented, through AI Overviews and conversational search surfaces.
Publishing material valuable enough to earn attention and links on its own merit. Its rise in the 2010s marks the point where the industry’s answer to “how do we get links” shifted from acquisition schemes to publishing.
Google’s named page-experience metrics: Largest Contentful Paint for loading, Interaction to Next Paint for responsiveness (which replaced First Input Delay in March 2024), and Cumulative Layout Shift for visual stability. A genuine ranking input, though a modest one against relevance.
Models trained on vast text corpora to generate and summarise language. They now sit between a query and its answer, which changes the object of optimisation from a ranked link to a cited source.
Systems that improve from data rather than explicit rules. Present in ranking since RankBrain in 2015, and the mechanism behind almost every ranking change the industry has since tried to reverse-engineer.
Indexing the mobile version of a page as the canonical one. Announced in 2016 and completed across the web by 2023, it turned a design preference into an indexing fact: if content is absent on mobile, it is effectively absent.
Machine-readable markup — usually schema.org in JSON-LD — that states explicitly what a page is about: a product, an article, a business, its author. It does not raise rankings directly, but it is how a page becomes eligible for rich results and how AI systems resolve entities and attribute quotations.
How usable and satisfying a page is in practice. Long dismissed in SEO as unmeasurable, then partially operationalised through Core Web Vitals — an example of the field absorbing a term from an adjacent discipline once it acquired numbers.
We measured 5,901,821 words of Wikipedia’s revision history to date the rise and fall of every term on this page — including the exact hour the industry’s own name was accepted.
Read the research
Networked publishing platforms. The single most frequent term in our corpus by an order of magnitude — not because it is an SEO technique, but because it reshaped what the surrounding articles were about.
First seen 2006 · Peaked 2022 · Retains 76% of peak