In 2026, the question isn't whether you're "doing SEO"; it's whether Google can tell what your site is about and who is behind it. The work moved past chasing exact-match keywords and toward being a recognizable source on the topics you cover.
If you're still obsessing over keyword density, you're playing a game that ended years ago.
Things, not strings
Google has described its Knowledge Graph with that phrase since 2012. The last size it published officially, in May 2020, was over 500 billion facts about five billion entities. Bigger numbers circulate, but they're third-party estimates Google never confirmed. The point doesn't depend on the number anyway: search connects people, places, things and concepts, instead of only matching strings of text.
For anyone managing content, my client work keeps pointing at the same priorities:
- A cluster of related posts on a topic, instead of one-off articles.
- Structured data (JSON-LD) that says who you are and matches what's visible on the page.
- Writing for the intent behind a query, and only then for its search volume.
None of these guarantees a ranking. They make it easier for Google, and for readers, to understand what you actually know.
AI Overviews: how common, and what they cost
How often AI Overviews show up depends on who is counting and which queries they count. The most careful measurement I've found comes from Pew Research Center: tracking the real searches of 900 U.S. adults, about 18% of their Google searches in March 2025 produced an AI summary. Tools that track commercial keywords report much higher shares, and the number keeps moving.
Pew also measured the cost. Users clicked a traditional result on 8% of visits when an AI summary appeared, against 15% when it didn't. Google's position is that the clicks that still happen are higher quality, meaning people spend more time on the site. Both can be true at once: fewer clicks, better ones.
On getting cited, Google's documentation is blunt. There are no additional requirements or special optimizations to appear in AI Overviews, no AI text files, no special schema, and regular SEO best practices still apply. What I do on my own pages and on client pages:
- Answer the question near the top, not three paragraphs down.
- Use clear H2 and H3 headings, so a section still makes sense when it's quoted alone.
- Publish something a model can't produce on its own: data, a case study, a position you can defend.
That's my practice. Google doesn't publish a checklist for this.
Core Web Vitals: INP replaced FID
INP (Interaction to Next Paint) replaced FID as a Core Web Vital in March 2024, and a good INP is 200 milliseconds or less. Google confirms that Core Web Vitals are used by its ranking systems, and in the same page it says it "always seeks to show the most relevant content, even if the page experience is sub-par." Good scores don't guarantee top rankings either.
So a slow site won't sink great content on its own. What it does is lose the visitors who arrive, and a WordPress site weighed down by unoptimized scripts loses them on the first tap. That's why I put WordPress performance and technical SEO upstream of content decisions: it decides whether the traffic you earned turns into anything.
A slow page wastes the ranking it already has.
E-E-A-T is about trust, not a score
With AI-generated content everywhere, trust is the hardest thing to fake. Google's quality rater guidelines, updated in January 2025, added generative AI to the definition of scaled content abuse, with examples like a page "created with generative AI with likely no original content and provides no value to users."
Two caveats matter here. Raters don't rank pages: Google says their ratings "do not directly impact ranking" and are used to benchmark the quality of results. And Google says E-E-A-T "itself isn't a specific ranking factor".
Your experience, published work and public track record still matter, because they're what readers look for when deciding whether to trust a page, and what Google's systems try to identify through other signals. A bio doesn't rank anything by itself. Proof of expertise inside the content is what makes the content worth ranking.
What to do with all this
SEO in 2026 rewards being the source other pages summarize. Use AI to speed up research, and keep the part a model can't produce for yourself: what you saw, what you measured, what you got wrong. Then check every claim you publish against its source, including the claims in posts like this one.
Which of your pages would still be worth citing if the AI summary above it were accurate?
If your content is good and your Core Web Vitals are quietly capping it, WordPress performance work is where I'd start. Send me the URL.