Google Cut My Reach 56%, AI Citations Jumped 21x. Here's Why

Something split in two during May. Google's search rankings turned against my site hard, while my citations inside AI answer engines climbed just as hard, over the very same weeks. Reading both dashboards side by side has been very clarifying as of why, because the reason behind both moves turns out to be one and the same.

Let me show you both halves, then the single idea that ties them together.

Half one: Google pulled back, sharply

In late May, Google rolled out a broad core update. These land every few months and reshuffle which pages the algorithm treats as most worth showing. My site landed on the losing side of this one, and the numbers are not subtle.

Comparing the four weeks after the update against the four weeks before:

• Impressions fell by roughly 56%.

• Average ranking position roughly halved in quality, sliding from around 7 to around 15. My pages didn't just appear less, they ranked far worse, dropping from the bottom of page one to the middle of page two.

• Clicks held flat and CTR actually rose, from about 0.3% to about 0.8%. Fewer impressions, but the ones that survived were better matched to what people wanted. That detail matters, and I'll come back to it.

A core update is nothing more than a reassessment: the algorithm took a fresh look at my pages against everything else on the web and decided a good number of them deserved a lower spot than they held a month earlier. In some ways that stings more than a penalty, because there's nothing to "fix." There's only content to make genuinely better.

So I went and read what Google itself now says about visibility, and that's where the two halves of this story connect.

What Google says it actually rewards

Google publishes a guide on optimizing for its generative AI features, and the striking thing is how little of it is about mechanics. The single principle it places above everything else is that unique, genuinely useful content shapes your visibility more than any other factor in the guide.

The contrast it draws stopped me. A first-hand account based on real experience stands apart; a summary of things already published elsewhere simply restates what's already out there. It names the failure mode without flinching: commodity content. This is content built on common knowledge that could have come from anyone and adds little real insight, and it's what gets pushed down.

Their own example of commodity content is a generic listicle, something like "7 tips for first-time homebuyers." Their example of the good kind is a specific, lived, experience-based piece that only one person could have written.

Hold that next to a site with a hundred-plus templated topic pages, and the core update explains itself. The pages that lost rank were the ones closest to commodity: the same structure with the audience label swapped, useful enough but carrying nothing you couldn't find elsewhere or generate in seconds.

The guide also pushes back on a lot of popular advice, including parts of what I'd absorbed. Google says plainly that for its own search you don't need special machine-readable files, you don't need to chop content into little chunks, and you don't need to rewrite everything for the machines. It treats most of the so-called optimization hacks as noise. What it refuses to call noise, what it places first, is non-commodity, first-hand, people-first content. So they care far more about substance and far less about the mechanics than I'd initially assumed.

Half two: AI citations went the other way, hard

Here's what makes the timing remarkable. Over roughly the same stretch that Google was demoting my pages, my citations inside AI answer engines did the opposite. They climbed from around 18 to around 381. That's in the order of a 21x increase, and close to half of that growth landed in just the last couple of weeks, the very period Google was pulling back.

The rise also broadened instead of narrowing, which matters more than the headline figure. A few weeks ago a single page drove more than half my citations, and a single question drove a quarter of them. Now they're spread across seventeen pages, and the top question accounts for only about 16% of the total, down from 26%. Pages that barely registered before are now pulling dozens of citations each.

For a few days this felt like a contradiction. How can the same site be judged less worthy by one system and more worthy by another, in the same weeks? The answer is that the two systems were judging different pages by the same rule.

The one idea that explains both halves

Both metrics reward the same thing: content that carries something only you could have made. They simply applied that test to different parts of my site at the same moment.

The pages Google demoted were the commodity-leaning ones: broad, templated, thin on first-hand substance. The pages AI engines started citing were the ones with a genuine, specific, hard-to-copy angle.

Same period, same site, opposite outcomes, because each system was weighing a different slice, and both applied an identical standard: is there real, non-commodity value here, or is this something a model could produce on its own?

That turns the core update from a loss into a piece of instruction that I can further use on my website. It told me, bluntly, which of my pages carry real value and which are filler. The AI citation data told me the same thing from the opposite direction, by rewarding exactly the pages with a distinct point of view.

I'd love to claim I read Google's guide and engineered the citation jump from it. The honest version is better. I reached the same conclusion through my own experiments before I read the guide, then found Google had written down the identical principle. This only confirmed my initial supposition.

What I'm doing with this

I will stop spreading effort across commodity pages that were flagged as low value both by the search as well as the AI citations. I will then concentrate it on fewer pages that carry real first-hand judgment. The goal is to make each page genuinely worth citing and genuinely worth ranking, which, as it turns out, are now the same job.

The CTR is also another confirmation I got from this update. When Google stripped away the weak impressions, my click-through rate roughly doubled.

What I'm thinking about this week

Which of my pages are genuinely non-commodity, and being honest about the answer.

I will be sorting the topic pages into two piles: ones that carry some value confirmed by some positive increase in citations or search results vs ones that are a template with a label changed. I'll report back on whether concentrating the effort moves either dashboard.

Reply to this email with one of:

• Whether the May core update helped you, hurt you, or did nothing, and what you think it was reading

• Your honest estimate of what share of your own content is truly non-commodity

• A page of yours that earns attention on both Google and AI engines, and what you think makes it work

P.S. The playbook on earning AI citations is here: How to Get Cited by AI. The rest of the toolkit, in the order I'd use it: Visibility Audit → SEO Strategy Playbook → Title & Meta Audit → Sovereign Idea Workflow → Business Model Canvas. All free.

P.P.S. Know someone who got hit by the May core update and is panicking? Send them this. It reframes the whole thing. They can subscribe at thesovereigntechnologist.com.

That’s all for this week.

See you next Thursday.