Question.Marketing
Book

About Simon

I published an article on AEO 1,153 days before ChatGPT existed

I'm Simon Young. I've worked in search for thirty years. On 4 October 2019 I published an article titled "AEO is your business ready?", using the term, naming answer engines, and asking whether the reader's agency had raised either. OpenAI released ChatGPT on 30 November 2022, one thousand, one hundred and fifty-three days later. The academic paper that coined "generative engine optimisation" appeared in November 2023, more than four years after that article.

Everything below is dated, linked and checkable. Including the parts I got wrong.

Last updated 31 July 2026

Why this page exists

Most agencies selling answer engine optimisation are younger than ChatGPT. Their founding story is that they spotted the shift early. There's no way to verify that, and no particular reason to believe it.

So rather than tell you I saw this coming, here's the timestamped public record. Read it and decide for yourself.

The receipts

Five published articles, dated and linked

These are long-form articles on LinkedIn, not feed posts, they have permanent public URLs you can open right now. Each publication date below was read from LinkedIn's own structured data on the live page, not estimated from memory. The captures were taken on 31 July 2026 and show the title, the byline and LinkedIn's own "Published" line in frame. The earliest one, September 2018, has no capture yet, the live page carries its date.

Two of them make the AEO call explicitly: “AEO is your business ready?” (4 October 2019), which asks whether your SEO company has even mentioned AEO, and “Questions & Answers in SEO” (5 February 2020), which names Answer Engine Optimisation as the biggest opportunity in search. The other three argue about the SEO industry rather than about answer engines, and are kept here for the record rather than as evidence of the call.

Verifying the dates changed one of them: I had the "Death of the SEO industry" piece filed as 2020, and it is actually November 2019. It is corrected here rather than quietly left as it was.

· 1,534 days before ChatGPT

SEO Companies, thief, fraud or ignorant? Are they worth it?

The earliest published article in the record, and an earlier draft of the argument that became "Death of the SEO industry" fourteen months later, with nearly identical opening text. About the industry's billing model rather than about AEO, so it sets the ground rather than making the AEO call. No capture yet, the live page carries the date.

Date read from LinkedIn's structured data · no capture yet, open the live page

LinkedIn article “AEO is your business ready?” by Simon Young, published 4 October 2019

· 1,153 days before ChatGPT

AEO is your business ready?

Uses the term AEO, names answer engines, and asks whether the reader's agency has raised either. Published four years and one month before the paper that coined "generative engine optimisation".

Date read from LinkedIn's structured data · capture 31 July 2026

LinkedIn article “Death of the SEO industry, it needs a cleanup” by Simon Young, published 20 November 2019

· 1,106 days before ChatGPT

Death of the SEO industry, it needs a cleanup

Earlier than I had it filed. The verified date is November 2019, not 2020. The argument, that the industry's billing model rather than the practice was the problem, is the one I still make.

Date read from LinkedIn's structured data · capture 31 July 2026

LinkedIn article “Questions & Answers in SEO, building brand credibility” by Simon Young, published 5 February 2020

· 1,029 days before ChatGPT

Questions & Answers in SEO, building brand credibility

Questions, not keywords, as the unit of search. Six years later the company is called Question.Marketing, which is either consistency or stubbornness.

Date read from LinkedIn's structured data · capture 31 July 2026

The archive

The feed posts around those articles, oldest first. Every date is derived from LinkedIn's own post identifiers, which encode creation time. Where a permalink is still outstanding, it says so rather than being implied.

21 August 2019, the AEO feed post

1,197 days before ChatGPT, permalink outstanding

“SEO is dying! Do you know what AEO is? It's very quickly taking over from SEO. Has your agency talked to you about answer engines?...”

This is a feed post rather than a published article, and I have not yet recovered its permalink. The wording is almost identical to the published article six weeks later, which asks whether your SEO company has even mentioned AEO to you, so the credential on this site rests on that article, 4 October 2019, 1,153 days, which is fully linked and dated.

LINK TO POST, PERMALINK OUTSTANDING · Same argument, published and dated, in “AEO is your business ready?”, 4 October 2019

This is the one that matters, and not because of "SEO is dying." People have been saying that since 1999, and I'll come back to why that framing was a mistake.

It matters because of what sits next to it: AEO, answer engines, and the question of whether anyone's agency was discussing them.

In August 2019 the answer for almost everybody was no. There was no category. No platforms, no funded startups, no conference track, no Google guidance. The paper that introduced "generative engine optimisation" to the academic literature was still more than four years away.

Right: the term, the concept, and the fact that agencies weren't preparing clients.
Wrong: "dying" was too strong, and I was early enough that nobody could act on it. Being right too soon has a commercial cost.

26 December 2019, search has to change shape

26 December 2019

“The future of search engines isn't as simple as me searching for something online. Search needs to evolve...”

LINK, PERMALINK OUTSTANDING · Written up a month earlier in “Death of the SEO industry, it needs a cleanup”, 20 November 2019

Right: the structural point, that the interface itself would change, not just the ranking factors. Which is exactly what happened when answers replaced lists.

8 February 2020, questions as the unit of search

8 February 2020

A post about a blog article “all about Q&A and how it helps with SEO... the biggest thing in SEO this year.”

LINK, PERMALINK OUTSTANDING · The article it points to is “Questions & Answers in SEO, building brand credibility”, 5 February 2020

Right: that the question, not the keyword, was becoming the unit of search. Six years later my company is called Question.Marketing, which is either consistency or stubbornness.

12 October 2020, the five-year call

12 October 2020

“SEO will be dead and forgotten in 5 years... Thoughts?”

I got this wrong, and it's the most useful entry on the page.

LINK · SCREENSHOT

Five years from October 2020 was October 2025. SEO is neither dead nor forgotten. Google's search revenue rose 19% year-on-year in Q1 2026 with query volume at an all-time high, and Google still holds around 90% of traditional search. I overshot badly.

But look at what did happen by that deadline. SparkToro's June 2026 research found fewer than one third of Google searches now send a click, an 8 percentage point fall in a single year, measured on a panel of sites run by professional marketers actively trying to grow traffic. The clicks went. The searches didn't.

So here's the correction, and it's the entire basis of how I work now: the practice of SEO is in fine health. The SEO business model, billing monthly for rankings and clicks, is what's finishing.

I collapsed those two things into one claim in 2020, and it made me sound like every other person shouting that SEO was dead. Getting the direction right and the framing wrong is normal in this work. I'd rather show you the correction than curate the hits.

2 December 2020, answer engines as strategy

2 December 2020

“The best SEO tactic for 2021 and beyond. Are you using Answer Engines as a strategy for SEO?”

LINK · SCREENSHOT

Right: treating answer engines as a strategic layer rather than a tactical add-on. That's the mainstream position now. Google's May 2026 guidance broadly agrees, it argues the work belongs inside SEO rather than beside it, which is a disagreement about labels, not substance.

14 December 2020, the video call

14 December 2020

“The future of SEO is NOT a traditional text-based search, do you agree? Video is wiping the floor with SEO...”

Half right, and worth showing you for the half that wasn't.

LINK, PERMALINK OUTSTANDING · I made the same argument at length two months later in “YouTube and the end of the SEO industry, is the shift finally coming now?”, 9 February 2021

The core intuition held: text-based, list-of-links search wasn't the end state. But I backed video as the successor, and it wasn't. Video grew enormously; search didn't become primarily visual. What actually displaced the list of links was the generated answer, which nobody, me included, saw arriving in the form it did.

What it taught me: being right that something will change is far easier than being right about what replaces it. It's why I now sell measurement rather than predictions, and why every number on this site carries a margin of error.

22 August 2024, for completeness

22 August 2024

“What are your thoughts on Google...”

LINK · SCREENSHOT

Post-ChatGPT, so this is commentary rather than prediction. Included because I'd rather show the whole record than the flattering slice of it.

The scoreboard

Six pre-ChatGPT posts. The honest tally:

CallVerdict
The interface would shift from lists to answersRight, and early
"AEO" and "answer engines" would be the frameRight, ~4 years before the literature
Questions, not keywords, would be the unitRight
Agencies weren't preparing clientsRight, and largely still true, only 14% of marketers track AI search performance at all, per Conductor's 2026 survey of 250+ digital leaders
SEO "dead and forgotten" by 2025Wrong. The business model is in trouble; the practice isn't
Video would succeed text searchWrong. Generated answers did

Four right, two wrong. That's roughly the hit rate you should expect from anyone making public calls about technology, and you should be suspicious of anyone claiming better.

Why the dates are precise

You'll have noticed exact dates rather than "a few years back."

LinkedIn post identifiers are Snowflake-style, the creation timestamp is encoded in the identifier itself. Every date here is derived from the identifier in the post's own URL and cross-checked against what LinkedIn displays. If you want to verify any of them independently, the URLs are above and the method is about one line of code.

Small thing. Same principle as everything else here: if proof is the product, the proof has to survive being checked.

The timeline behind the posts

Early 1990s to 1998, before Google

I was running my own computer consumables business, and I optimised it myself, out of commercial necessity rather than as a discipline anyone had named yet. Ranking on Lycos, AltaVista and Ask Jeeves, five years before Google launched. At one point, for a laugh, I got the business ranking number one on Lycos for the single term bank. In hindsight I should probably have kept hold of that one.

1998 onwards, the PageRank era

The PageRank era began in 1998, when Larry Page and Sergey Brin introduced the algorithm that gave Google its advantage. By the time it arrived I already had seven or eight years of ranking sites in search behind me, so the shift was a change of mechanism rather than a change of trade: work out what the machine is rewarding, then earn it rather than fake it.

2016 to 2019, answer engines, first time round

At least eight years before ChatGPT I was ranking answers in voice search, on Alexa and the assistants around it. Technically it was good fun. Commercially it went nowhere, because very few people were actually searching for products and services by voice, and low adoption meant low client interest.

What it taught me: when a machine returns one answer instead of ten links, second place is worth nothing. I saw that on a device in people's kitchens years before anyone was discussing AI Overviews. Same problem now, vastly larger scale.

The algorithm decade

A decade in which digital media and social platforms all ran on the same core assumption: that the algorithm would actively curate and deliver an audience straight to your content. I spent a large part of it on video, ranking on YouTube and building video specifically to rank, because video was being pulled into the main Google index faster than anything else you could publish.

2019–2022, saying it out loud

The posts above, written while the industry consensus was still that search would keep sending links and keep sending traffic. CONTEXT: REACTION, WHO DISAGREED, WHAT YOU WERE DOING COMMERCIALLY AT THE TIME

2023–now

ChatGPT arrives, the category forms, and the thing I'd been describing since 2019 becomes the thing everyone needs. Question.Marketing is the practice built for it: question research feeding both outreach and answer-first pages, with results published in a public ledger, including the interventions that produced nothing.

linkedin.com/in/aeo

I hold the LinkedIn vanity URL linkedin.com/in/aeo. Registered YEAR, CONFIRM. The archive shows it was in place by August 2024 at the latest.

There's a story in it, and a technical point. When an AI system works out who is an authority on a topic, a high-authority crawlable profile whose URL is the category term is a useful entity signal. Ahrefs, across 75,000 brands, found branded and entity signals correlate 0.664 with AI Overview visibility against 0.218 for backlinks, roughly three times as predictive.

I didn't acquire it for that reason. In 2019 there was no reason to. But it's the kind of asset nobody can compete for, because there's only one.

What thirty years teaches you that eighteen months can't

Which fundamentals never move. Crawlability, structured information, genuine authority, answering the actual question. That's been the job since the nineties and it's the job now. Google's May 2026 guidance says essentially this, and it's right. It reads as revelation only to people who arrived recently.

Which panics are real. Mobile, social, voice, all overstated, and I said so at the time. This one is different in kind: for the first time the interface returns a synthesised answer rather than a set of options, on surfaces that don't use Google's index.

That being early is not the same as being right. I was talking about AEO in 2019 and it made me nothing at all for four years. Timing is most of it.

How to hold a position you can't fully prove. Ahrefs found the overlap between Google's top ten and AI Overview citations collapsing from 76% to 38%. BrightEdge found it rising. Both are competent teams. When that happens you measure your own situation and stop arguing about the average, which is why everything I publish carries a confidence interval.

How I work

I turn up. Doncaster-based, on-site across South Yorkshire, remote beyond it. I build with your team rather than around them, and hand over something they can run without me.

And I publish the numbers, baselines, interventions, results, on a fixed cadence, including the work that produced nothing. Six years of public posts with two wrong calls in them is the same instinct: nobody can verify anyone's claims in this market, so the only useful thing I can do is show you the whole record.

The client ledger → · What didn't work → · The AI Visibility Index →

Book a conversation