Is SEO still worth teaching?

A professor rewriting a digital marketing module asked me this recently, and it is the reasonable version of a question the industry keeps asking in a louder voice: is SEO dead? If AI assistants now answer the questions that used to return ten blue links, should the SEO weeks give way to AI weeks?

I have been doing SEO since 2010, which means I have heard that it is dying for my entire working life. The short answer: yes, SEO is still worth teaching, but the 2015 version is not. Students need to understand how organic visibility is produced, how AI answers reuse and reassemble web sources, why clicks are declining, and how success now includes citations and influence without a visit. SEO and GEO belong in the same teaching unit. The rest of this piece is the case for that answer, and it changed what our own AI Marketing course teaches.

For anyone meeting the acronym cold: GEO is generative engine optimisation, the practice of increasing the likelihood that a brand or its content appears, and is cited, in the answers produced by AI systems such as ChatGPT, Perplexity and Google's AI Overviews. The term is barely two years old and the vocabulary is still unstable, which this article treats as a feature of the field rather than a reason to wait. A short glossary sits at the end.

How many times has SEO died now?

At least five times by headline, on my count. ClickZ ran “SEO Is Dead. Long Live, er, the Other SEO” in 2007, when universal search was the killer. In 2008 the killer was social media, in a widely circulated post arguing SEO had no future. By 2012, after the Panda and Penguin updates, ClickZ was running “SEO Is Dead.… No, It Isn’t.… Yes, It Is.…”, which at least had the honesty to argue with itself in the headline. In 2014 mobile was the cause of death. And Forbes has published two fresh obituaries within the past year, both naming AI.

None of those obituaries culminated in the extinction it predicted. Google now says it sees more than five trillion searches a year, and independent panel data measured Google search growing over 20% in 2024 alone (SparkToro/Datos, 2025). What actually dies in each cycle is a mechanic: a tactic, a page format, a way of counting. The behaviour underneath, a customer with a question and an intermediary with an answer, remains enormous and has spread across new intermediaries. The habit worth teaching students is to read every “death of search” headline as a signal that a mechanic is changing, and to ask which one. That is a more useful skill than either believing the obituaries or ignoring them.

What does the evidence say about SEO in 2026?

This cycle deserves more respect than its predecessors, because parts of it are real. Start, though, with the honest mess: nobody agrees how big AI search actually is, and the disagreement is itself worth teaching. Ahrefs reclassified ChatGPT’s usage categories and estimated it handles query volume equal to roughly 12% of Google’s (2026). SparkToro and Datos, working from a clickstream panel, measured Google at roughly 373 times ChatGPT’s search volume (2025), which puts ChatGPT under 1%. Both are serious attempts by capable researchers.

They disagree because they count different things, and “what exactly counts as a search?” is precisely the question a marketing course should train students to ask of any vendor’s number. Two firmer reference points help bound the argument: OpenAI’s own research had ChatGPT above 2.5 billion messages a day by mid-2025, while Ahrefs’ panel of 76,000 websites (2026) found Google still sending roughly 190 times more traffic to the open web than ChatGPT.

The click evidence is firmer. Pew Research Center (2025) tracked 68,879 real Google searches from 900 US adults and found that when an AI summary appeared, users clicked a traditional result on 8% of visits, against 15% without one. Ahrefs’ study across 300,000 keywords (2026) puts the top organic position’s clickthrough loss at roughly 58% year on year where AI Overviews appear. SparkToro’s 2024 study found about six in ten US searches already ended without a click to the open web. In Similarweb’s e-commerce data (2026), the AI-referred visits that do arrive look markedly higher-intent: roughly 11.4% converted, against 5.3% for ordinary organic search, with around double the engagement.

The strongest evidence for the obituary deserves its place in the lecture too. Reuters Institute and Chartbeat measured Google organic traffic to more than 2,500 news sites falling 33% globally, and 38% in the US, in the year to November 2025. For publishers the click decline has already happened. Gartner, by contrast, predicted in early 2024 that traditional search volume would fall 25% by 2026, a forecast rather than a measurement, and Apple testified in 2025 that Safari searches had declined for the first time in over two decades. If a course teaches SEO as a traffic-harvesting machine for informational content, the observed evidence against that course is accumulating. Whether that indicts SEO, or one narrow idea of what SEO is for, is exactly the seminar debate worth staging.

Why can't you teach GEO without SEO?

Start with an observation from practice rather than a study. Inside Google, the boundary between SEO and GEO has for most working purposes already collapsed. The same results page now carries classic rankings, an AI Overview assembled from them, and AI Mode one tab away; the same crawler, the same index and largely the same signals feed all of it.

Nobody optimising a site in 2026 can cleanly separate the SEO work from the GEO work on that page. It is one job with more surfaces than it used to have. There is precedent, too: the results page has been absorbing new result types since universal search in 2007, the occasion of the first funeral above, and AI answers are simply the latest absorption. The SEO-versus-GEO distinction only earns its keep beyond Google, where assistants like ChatGPT and Perplexity retrieve and cite by their own habits.

The measurement studies back the observation, though they look contradictory until you notice they measure different layers and periods of the same system. seoClarity’s analysis of 432,000 keywords (2025) found 97% of AI Overviews cite at least one source from the top 20 organic results, drawing five URLs from them on average. BrightEdge (2025) measures the share of citations that also rank organically at around 54%. And Ahrefs’ March 2026 update, across 863,000 SERPs, found 38% of cited URLs came from the same query’s top ten, down from roughly 76% a year earlier.

Taken cautiously, those studies suggest a direction rather than a settled fact, and the direction is the useful part. Google’s AI answers still draw heavily on organically discoverable content, but ranking for the exact query appears to be becoming a weaker predictor of citation on that query, because the system fans the question out into related sub-queries and cites what ranks across them. If that reading holds, the reward is shifting toward broad, genuine topical authority and away from chasing one position for one keyword: SEO’s durable layer mattering more, its most mechanical habit mattering less. For assistants beyond Google the mechanisms are less documented, which is worth saying plainly in a classroom; the concepts of retrieval, authority and citation still give students the working model.

There is a pedagogical version of the point as well. In developing our companion text we found that GEO is easiest to understand after a student has grasped the SEO logic of organic visibility, authority and technical structure, and a 2026 paper in Marketing Education Review makes the same argument for integrating GEO on top of SEO rather than replacing it. Teach GEO first, or alone, and it degrades into incantations: be quotable, be structured, be cited, with no model of why.

How should SEO be taught in the AI era?

It helps to be concrete about what the standing model is, because it is remarkably uniform. Read the public outlines of university SEO teaching, from extension programmes to MBA electives, and the same unit repeats: how search engines work, keyword research, on-page optimisation, off-page and link building, technical SEO, an audit assignment, often a platform certification to finish. That is the technical, content and backlinks model, and it was a fair map of the discipline in 2015. It is not wrong today. It is incomplete in ways that now matter more than anything it includes.

The 2026 unit keeps that spine and widens every pillar. Content becomes quality as Google formally assesses it: E-E-A-T (experience, expertise, authoritativeness, trust), with visible authorship rather than anonymous copy. Backlinks become earned authority in the round: digital PR, expert commentary, reviews and mentions, the same material AI systems weigh when they choose what to cite; the history of manufactured authority, and what the platform did to businesses built on it, stays on the reading list as a warning. Technical work extends from crawlability to being retrievable and quotable by machines.

And the arena widens from one results page to search everywhere: YouTube, TikTok, Reddit, Amazon and the AI assistants are all surfaces where a customer's query can now end, each with its own ranking logic.

The older habits worth keeping get sharpened rather than dropped. Reading a ranking position as information rather than a score. Relevance before volume, with the evidence that brand familiarity affects organic click behaviour (Baye, De los Santos and Wildenbeest, 2016).

Topic coverage over keyword chasing, which the fan-out evidence has just made examinable. And the new measurement reality: presence inside AI answers, influence without the click, and the contested numbers above, taught as contested, because evaluating vendor measurement claims is now part of the job.

In our AI marketing course (syllabus here) this all lives in one deliberately paired week: organic discovery, SEO and GEO together. Students audit a set of prompts across two AI systems, compare what the assistants cite with what traditionally ranks, and then choose one SEO intervention and one GEO intervention for the same brand. The pairing is the point. Separate weeks would imply separate disciplines, and the whole argument of the companion text is that they are one discipline with two surfaces.

So the answer from the top of this piece stands: yes, SEO is still worth teaching, and the case now depends on teaching a different version of it. The 2026 version is measured against zero-click behaviour and machine-written answers, pairs SEO with GEO in one unit, and keeps its own funerals on the syllabus as evidence. If you would like the week-by-week version, the full fifteen-week AI marketing syllabus and the companion text chapters are free to any professor who emails: clark@novela.academy.

A short glossary

Generative engine optimisation (GEO). The practice of increasing the likelihood that a brand or its content appears, and is cited, in answers produced by generative AI systems such as ChatGPT, Perplexity and Google's AI Overviews.

AI Overviews. Google's machine-written summaries that appear above the classic results, assembled from ranking pages and citing a few of them.

AI Mode. Google's conversational search experience, which answers in dialogue rather than returning a results list.

Zero-click search. A search that ends without a visit to any website, because the results page or answer resolved the question in place.

Query fan-out. Expanding one question into several related sub-queries and drawing citations from pages that rank across them, rather than only from the original query's results.

E-E-A-T. Experience, Expertise, Authoritativeness and Trust: the quality framework in Google's published guidelines for its human search evaluators.