Around 2011, a young SEO figured out that buying links worked. Her CEO promised her a drum set if she got the company to number one. Then one morning in August 2012, the rankings were simply gone.

Connect on LinkedIn · Visit my site

Lily Ray told this story to Search Engine Land this week, and the timing is what makes it worth repeating. She's watching people run automated content workflows and GEO experiments with the same confidence she once had about buying links.

Penguin was Google's 2012 crackdown on manipulative link building. Overnight, sites that had been ranking second or third for their most valuable keywords were gone from the results entirely.

Ray says that experience shaped how she has approached SEO ever since. It taught her something specific: the fact that a tactic is working right now tells you nothing about whether it will keep working.

Ray moved toward fundamentals after Penguin. Title tags, headings, image optimization, internal linking. Boring, durable work.

Then Google's 2018 Medic update hit one of her agency's clients hard, and she went deep into Google's Search Quality Rater Guidelines and E-E-A-T. That stands for experience, expertise, authoritativeness, and trustworthiness, and it's Google's shorthand for how much a source can be relied on.

Her key point about E-E-A-T is one a lot of people still miss. It isn't a setting you turn on in your own site. It's how other companies, reviewers, and customers evaluate your brand, which means you can't fully control it from your own CMS.

Ray spent hundreds of hours studying sites hit by the helpful content update. Plenty of them deserved it: generic writing, stock photography, no sign of a real person behind the work.

But she estimates roughly 40% of the affected sites were genuinely trying to do things right. Recipe bloggers and travel bloggers who had followed Google's published guidance got hit alongside everyone else.

She says she spoke with dozens of business owners who lost their entire livelihoods. That's why she's cautious about GEO tactics now, not because she thinks the tactics are silly, but because she's watched what a broad correction does to people standing near the blast radius.

Ray points out that AI platforms are dealing with low-quality content right now. Generative AI made producing SEO spam cheaper and faster than it has ever been, and the AI tools are citing some of it.

She's also seeing early attempts to clean it up. ChatGPT, for example, uses fan-out queries that restrict some searches to specific domains rather than pulling from anywhere.

Her read on what comes next is based on pattern recognition, not prediction. Every time spam has worked well at scale, the platforms have eventually come up with a big, blunt way to shut it down.

There's a growing argument that you can drop traditional SEO and optimize only for AI search. Ray rejects it in strong terms, calling the idea completely and blatantly false.

Her position is that if you actually understand how AI search works, you'd conclude SEO matters more now, not less. The best and most sustainable thing you can do for AI visibility, she says, is good SEO.

"You can't shortcut your way into being recommended as the best. That's the whole point of the recommendation."

What counts as winning also changes. In AI search, success means being recommended as the best or most relevant brand for what you do, and Ray means earning that naturally rather than publishing listicles that name yourself.

If your output has climbed sharply this year, check whether traffic and conversions climbed with it. If they didn't, you're adding risk without adding value.

Ray's approach is a smaller set of genuinely valuable assets, kept current over time. Pick your ten strongest pages and improve them before you write anything new.

E-E-A-T lives largely in what other people say about you. Reviews, mentions, and third-party coverage all feed the same signal your own homepage can't produce on its own.

Did You Know? Ray found her first SEO job on Craigslist while studying politics and Spanish at New York University. She didn't think she had anything worth saying on a conference stage until around 2017.

The uncomfortable part of Ray's story isn't that buying links was wrong. It's that buying links worked, and worked well, for a long stretch before it didn't.

That's the trap. A tactic producing results is not evidence that the tactic is safe, because the results and the correction arrive on completely different schedules.

I don't know whether the GEO reckoning she's describing actually comes. But the question she ends on is the one I keep returning to: if a system were genuinely trying to identify the best service in your category, would it pick you?

Till next time,

Source