What BrightonSEO talks reveal: AI changed the target, not the job

The claim you have heard for three years is that AI is transforming how SEO work gets done. BrightonSEO has published its agenda 15 times since 2019, and across 1,804 talks it says something different: the amount of programming about AI doing the work has not moved at all. What changed is what the work is aimed at.

Splitting AI into two senses is the whole method, because the single word hides the finding. AI as a tool is a talk about automating SEO labour: generating content, scaling audits, writing scripts. AI as an interface is a talk about being visible inside an answer engine: optimizing for ChatGPT, AI Overviews, LLM citation. Counted separately, they move in completely different ways.

Tool talk never grew. It sits between 2.6% and 9.1% on every single event, mean 4.8%, with no trend in either direction. Sep 2019 had 5 tool talks; Oct 2026 had 4, on a longer agenda. Interface talk went from 1 talk in 635 through 2022 to 21.8% of the programme.

The turn was one programming cycle, not a drift

This is the part a trend line would hide. Interface talk does not ramp up through the ChatGPT era: it stays in the noise, between 0.6% and 2.5%, for six consecutive events after ChatGPT launched. Then it jumps 2.5% to 14.6% between Apr 2025 and Oct 2025, a 12.1-point move in a single event, and keeps climbing to 21.8%.

A conference agenda is a lagging indicator with a lead time: talks are pitched, selected and scheduled months before anyone stands up. A step that size means the programme committee's read of the field changed between two cycles, not gradually. If you want a date for when optimizing for answer engines stopped being a fringe topic and became a track, it is Oct 2025.

What counts as which, so you can disagree

The split is keyword classification on talk titles. It is published here in full because a reader who disagrees with the boundary should be able to see exactly where it was drawn rather than take the chart on trust.

• Tool: automate, automation, script, scale, workflow, efficiency, productivity, and AI or LLM terms appearing alongside them, where the object of the sentence is the SEO's own work.
• Interface: ChatGPT, AI Overviews, SGE, Perplexity, Gemini, answer engine, AI search, LLM visibility, citation, AEO, GEO, where the object is being found inside a generated answer.

Two honest consequences. A title mentioning both lands in both counts, so the two series are not mutually exclusive and do not sum to a total. And a title that names neither lands in neither, which is the next section.

A second corpus, pointing the same way

An agenda is somebody's forecast of what an audience wants. That is a real limitation, so it is worth checking against a record of what the industry actually talked about, and this show has one: every episode since 2009, transcribed and counted year by year in The 18-Year Shift.

The two corpora share no text and no classifier. One is conference titles chosen by a programme committee; the other is spoken transcript from interviews and monologues. They agree on the inflection: AI topics climbed from 4.5% of this agenda in Sep 2019 to 38.5% by Oct 2026, on the same timeline over which the podcast corpus goes from no AI mentions to AI in most episodes. Neither measurement can corroborate itself. Together they are harder to dismiss than either alone.

What a flat line is telling you

The useful read is not "learn AI." It is that two things sold as one thing have completely different evidence behind them.

The skills the agenda has never stopped programming, at a steady 4.8% for seven years, are the ones nobody found a shortcut for. Tooling talk did not collapse because tools got good; it stayed exactly where it was while the field found something else to argue about. If you invested in technical craft, the agenda says that investment did not get obsoleted.

What genuinely arrived, on a datable step, is the surface you are optimizing for. That is a change in the target, not in the job. Anyone telling you the work itself has been transformed is making a claim this agenda has had 15 chances to support and has not.

What this data cannot tell you

Published in full rather than footnoted, because a number that turns out to be overstated and undisclosed costs more than the citation was worth.

• Roughly a quarter of talks match no theme at all, and that rate falls over time (38.2% on Sep 2019, 17.3% on Oct 2026). Because the denominator gets cleaner while the numerator grows, every rising share here is marginally overstated. The direction survives; the exact percentage is soft.
• Two eras, two instruments. 8 events were read from archived pages that carry abstracts; 7 from live listings that do not. Abstracts cut the no-theme rate sharply, so mixing them would compare two different classifiers. Every figure on this page is title-only on both sides to keep one instrument throughout, which also means a talk whose title hides its subject is invisible to this count.
• 1 harvested event is excluded as non-reportable: Oct 2020 (n=1). A share computed on that few talks reads like a data point and is noise. 16 events were harvested; 15 are reported.
• One conference. This is BrightonSEO, not the industry. It is the largest SEO conference in Europe and it programmes twice a year, which is what makes a seven-year series possible at all, but a single programme committee's taste is inside every number.
• Titles are marketing. Speakers name talks to fill rooms. "AI" in a title in 2026 may be doing promotional work that the content does not back, which if anything inflates the interface line.

The data

Every number on this page is computed at build time from the dataset below. Nothing is typed in by hand, so the page and the file cannot disagree.

Download the dataset (JSON) · 1,804 talks · 15 reportable events · Sep 2019 to Oct 2026 · CC BY 4.0.

Free to reuse with attribution. Cite it as: The Best SEO Podcast, “The Conference Agenda Topic Shift,” October 2026, https://www.bestseopodcast.com/conference-agenda-topic-shift.