SEO began in the mid-1990s as informal tinkering with meta tags, acquired its name in 1997, and has since become an industry worth more than $100 billion a year, with The Business Research Company putting the global SEO services market at $108.28 billion in 2026. In under thirty years the discipline moved from directory submissions to competing for citations inside AI-generated answers. I am Adrian Dąbrowiecki, and I have worked in this field for seven of those years, which means I arrived after most of the foundational fights were settled and in time for the current one. Knowing the sequence matters more than it sounds, because almost every tactic being sold as new is a variation on something that was tried, rewarded, then penalized.
“SEO history moved from directories to AI in less than 30 years” Landingi, SEO History: Inventors, Key Phases and Popularity Timeline
When Did SEO Become Popular?
SEO became popular in the early 2000s, when Google’s rise and its deal to power Yahoo’s results turned a technical curiosity into a marketing requirement.
Through the mid-1990s the practice had no name and no rules. People called it web positioning or site promotion. Editing meta keywords and getting into the Yahoo directory was most of the job, and almost no business owner understood that visibility could be influenced deliberately.
The phrase “search engine optimization” started appearing in agency decks and conference talks around 1997. A year later Google launched with PageRank, which counted links from other sites rather than keyword repetition, and the economics of the whole discipline shifted.
The tipping point came in 2000, when Yahoo chose Google to power its search results. Millions of people were suddenly using Google’s algorithm through a familiar interface, with a small “powered by Google” line underneath. Between 2000 and 2005 businesses worked out that page one was where customers came from, link building became an industry, and agencies started offering ranking guarantees, a promise that has aged exactly as well as you would expect.
Then in 2003 the Florida update wiped out a large share of the spam tactics that had been working, and ethical optimization became the default position rather than a preference.
Who Invented SEO?
SEO has no single inventor, but the practice is usually credited to Bob Heyman, Leland Harden, Danny Sullivan and Bruce Clay, who were experimenting with rankings in the mid-to-late 1990s.
The origin story most often told involves Jefferson Starship. Working on the band’s web presence around 1995, Heyman and Harden noticed the site ranked better when the band’s name appeared more often on the page, so they added more mentions. That is keyword stuffing, described before anyone had a word for it, and it worked because early engines counted rather than understood.
In 1997 the agency Webstep Marketing used the phrase “search engine optimization” in a pitch, giving the practice a name. Danny Sullivan helped popularize it and then founded Search Engine Watch, which became the reference point for anyone trying to understand how rankings worked. Bruce Clay turned it into a profession, offering SEO as a paid service, training practitioners and defining much of the vocabulary still in use.
Worth noting how recent all of this is. The people who named this discipline are, in many cases, still working in it.
The 5 Phases of SEO Evolution
SEO evolved through five phases: keyword matching in the 1990s, link-based ranking in the 2000s, content quality and experience from 2010, semantic understanding from 2015, and AI-generated answers from 2020 onwards.
1. The early days (1990s)
AltaVista, Lycos and the Yahoo directory tried to catalogue a rapidly growing web. Correct keywords in meta tags plus a directory submission could put you at the top. No tools, no guidelines, no concept of user intent. Everything was trial and error, and the errors were cheap.
2. The Google era (2000s)
PageRank made external links the central ranking factor, and off-page work became as important as on-page. Florida in 2003 ended the easiest manipulation. The industry professionalized: agencies formed, paid search grew alongside organic, and Google Analytics and Webmaster Tools gave practitioners actual data for the first time. Local search appeared with Google Local in 2004.
3. Content quality and experience (2010 to 2014)
Panda in 2011 targeted thin and duplicated content. Penguin in 2012 targeted link schemes. Between them they ended the era where gaming the system was the strategy, and made content that deserved to rank the strategy instead.
Experience entered the picture too: site speed became a ranking factor in 2010, mobile-friendliness began influencing results by 2014, and Google My Business standardized local listings that same year.
4. Mobile and semantic search (2015 to 2019)
Mobile overtook desktop and Google responded with the mobile-friendly update in 2015. The more consequential shift was semantic. Hummingbird in 2013 started interpreting conversational queries, RankBrain in 2015 handled ambiguous ones with machine learning, and BERT in 2019 read nuance well enough that pages began ranking for phrasings they never contained.
This is the phase that produced the discipline I actually practise. Once engines understood meaning rather than strings, topical coverage started outperforming keyword placement, and the work became about entities and relationships instead of phrases.
5. AI and generated answers (2020 to now)
BERT went into nearly all English queries by 2020. ChatGPT arrived in late 2022 and within a year millions of people were using a chat interface for tasks they used to give a search box. Bing integrated GPT-4, Google shipped Bard and then Gemini, and AI Overviews began answering queries above the organic results.
Google also spent this period defending the index against the flood that generative tools enabled. The Helpful Content system arrived in 2022, and the March 2024 policy revision named scaled content abuse, expired domain abuse and site reputation abuse directly, with dedicated spam updates following in December 2024 and August 2025.
| Phase | What defined it | What it rewarded |
|---|---|---|
| 1990s | Meta tags, directories, manual submission | Repetition |
| 2000 to 2009 | PageRank, backlinks, analytics, paid search | Links |
| 2010 to 2014 | Panda, Penguin, speed, mobile | Content quality |
| 2015 to 2019 | RankBrain, BERT, mobile-first indexing | Meaning and intent |
| 2020 onwards | Generative AI, AI Overviews, spam policy expansion | Demonstrable expertise |
How Did AI Change SEO?
AI changed SEO by letting search engines interpret what people mean rather than match what they typed, and by changing what a results page looks like.
Intent replaced keyword matching
BERT and MUM introduced language processing that reads context. The practical consequence is that keyword density stopped mattering and topical completeness started mattering, which is why a page covering a subject properly ranks for queries it never mentioned.
AI became part of the production process
Writing, outlining, briefing and optimization now involve AI tooling in most teams, mine included. It accelerates the parts that were always mechanical. It does not supply expertise, and pages where it clearly did are exactly what the Helpful Content system was built to demote.
Research became predictive
Ahrefs and Semrush use machine learning to surface rising trends and forecast difficulty, which means you can build for what people are about to search rather than only for what they searched last quarter.
Search engines started fighting AI spam
The flood of generated content forced a response. Google’s position is now explicit: the issue is not how content was produced but whether it was produced primarily to manipulate rankings. That distinction is the one most people misquote in both directions.
Results became answers
This is the biggest change and the one still unfolding. AI Overviews and AI Mode synthesize responses above the organic listings, so a large share of queries now end without a click. Position one no longer guarantees the visit. Being the source that the summary cites is the new equivalent, and what earns that is clear structure, checkable claims and genuine expertise attached to a named person.
What Has Never Changed?
Three things have held constant across every phase: the engine has to find your page, understand what it is about, and have a reason to prefer it.
Every technique in thirty years has been a variation on one of those. Directory submissions and XML sitemaps solve discovery. Meta keywords and entity markup solve understanding. Links, reviews, brand mentions and AI citations solve preference. The implementations change completely; the three questions do not.
That is genuinely useful when you are evaluating something new. Ask which of the three a tactic addresses and whether it does so honestly. If the answer is that it fakes one of them, history says how that ends, and roughly how long it takes.
When Did Google Become Important in SEO?
Google became central in 1998, when it launched with PageRank and made links the primary measure of authority.
Larry Page and Sergey Brin treated each external link as a vote, and quality of the voting site as the weight of that vote. That single idea invalidated the keyword-stuffing era in one move: you could no longer promote yourself, other sites had to do it for you.
In 2000 the Google Toolbar displayed PageRank scores publicly, which turned an abstract concept into a visible number that everyone could optimize toward. Link building became the centre of the industry more or less overnight, along with every abuse that followed from it.
By the early 2000s Google’s results were relevant and clean enough that it became the default, and from that point optimizing for search meant optimizing for Google. That is still substantially true, with the qualification that a rising share of queries now happens inside assistants that were trained on the same open web.
What Was the First SEO Company?
Webstep Marketing is generally credited as the first company to offer search engine optimization as a named professional service, from 1997.
Their contribution was mostly linguistic and it mattered enormously. Naming the practice made it sellable, and making it sellable created an industry. Bruce Clay’s consultancy and Danny Sullivan’s Search Engine Watch then supplied the two things any new profession needs: someone doing the work commercially, and someone documenting how it works.
What Were the First SEO Tools?
The earliest SEO tools were WebPosition Gold and SubmitWolf, both from 1997, built to track rankings and submit sites to search engines in bulk.
Submission was a real job then. Getting listed was manual and had to be repeated per engine, so the first tools automated exactly that. AddWeb followed in 1998, combining submissions with keyword tracking and link popularity analysis, which was the first hint that this software could inform strategy rather than just save typing.
The modern generation arrived through the 2000s: SEOmoz, now Moz, launched in 2004, and Google’s Webmaster Tools, now Search Console, followed shortly after, giving practitioners feedback directly from the search engine for the first time. Semrush, Ahrefs and the rest built out from there.
The through line is worth noticing. Early tools told you where you ranked. Modern tools tell you why, and Search Console remains the only one reporting what Google itself observes rather than what a third party estimates.
What Does This History Miss?
Almost every account of SEO history, including this one, is written from an English-language and largely American perspective, which leaves out how search developed everywhere else.
Yandex launched in 1997, the same year the term was coined, and has run its own ranking systems in Russian-speaking markets ever since. Baidu launched in 2000 and dominates China with different technical requirements, including a strong preference for in-country hosting. Naver has held substantial share in South Korea with a portal model that ranks its own content ahead of the open web. Seznam still matters in the Czech Republic.
The multilingual side of the story has its own timeline too. Hreflang arrived in 2011, which is remarkably late given that the web was international from the start, and it is still the single most commonly misconfigured element I find on sites operating across markets. Fifteen years of an annotation existing has not made it widely implemented correctly, which tells you something about how much of the industry’s attention has stayed on one language.
Get International SEO Services From Marketer Coffee
The lesson I take from this history is that the surface changes constantly while the underlying job does not, and the same applies geographically. A market you have not entered yet is running the same three requirements, in a language your site does not speak. That is what we do at Marketer Coffee, a Poland-based international SEO agency working with SaaS and ecommerce brands across the US, UK and Europe, at rates 40 to 60% below comparable US and UK agencies. Every engagement opens with a free preliminary audit and a 90-day plan, including the hreflang and architecture decisions that decide whether the next market is straightforward or a rebuild.
See the scope and pricing in international SEO services, or go straight to your market: United States, United Kingdom, Germany. If you would rather see it applied first, the Landingi case study covers 375% organic growth across 29 months and eight language versions.
Written by Adrian Dąbrowiecki, founder of Marketer Coffee and SEO consultant with 7+ years in SaaS and ecommerce. Market data: The Business Research Company, 2026. Historical dates verified against Google Search Central and contemporaneous industry sources. I also co-wrote a guide on this subject for Landingi: SEO History: Inventors, Key Phases and Popularity Timeline.