SEO copywriting in 2026: what changed, what didn’t

AI summaries haven't killed SEO copy. They've changed what good SEO copy looks like. Here's the new playbook.

For about eighteen months, every other LinkedIn post has declared SEO copywriting dead. AI Overviews would eat the clicks. ChatGPT would replace search. Why write a 2,000-word guide when a language model can summarise it in three sentences?

And yet, here I are in 2026. Organic search still drives more qualified traffic than any other channel for most businesses I work with. Good SEO copy still ranks. The catch is that “good” now means something different from what it meant in 2022. The fundamentals haven’t been killed off, they’ve been re-weighted. This is the playbook I’m using right now.

What actually changed

1. The SERP is now a hybrid surface

AI Overviews (formerly SGE) sit at the top of a meaningful share of informational queries. They don’t appear on every search, transactional and navigational queries are mostly untouched, but for “how”, “what is”, and comparison queries, the Overview often answers the question before the user scrolls.

Click-through rates on position one have dropped for those query types. That isn’t the same as traffic dying. The clicks that do come through are more qualified, because the casual “what is X” reader got their answer up top and moved on. The people who click are the people who actually need more.

This means copy needs to do two jobs at once: be citable inside the AI Overview, and be worth clicking when the user wants depth.

2. Citations matter as much as rankings

Being cited inside an AI Overview, ChatGPT response or Perplexity answer is the new “rank one for the snippet”. The mechanics are different. AI systems pull from passages, discrete chunks of text that directly answer a question, not from whole pages. A 2,000-word article can contain a single 60-word passage that gets quoted across a dozen AI answers.

So the structural job has shifted. You’re not just writing a page that ranks; you’re writing pages full of standalone, quotable passages that answer specific questions in plain language.

3. E-E-A-T grew an extra E

Experience joined Expertise, Authoritativeness and Trustworthiness a couple of years ago, and in 2026 it’s the dominant signal of the four. AI models and Google’s quality systems are both increasingly good at detecting copy that’s been written from research versus copy that’s been written from doing.

“We tested this on twelve client sites” beats “studies show”. A screenshot of your dashboard beats a stock photo of a dashboard. A specific failure (“we tried this approach in March and it tanked CTR by 18%”) beats a generic warning.

4. Schema is now table stakes

JSON-LD structured data used to be a nice-to-have for rich results. In 2026, it’s how you tell AI systems what your page actually is. Article, FAQPage, HowTo, Product, LocalBusiness, Person, the right schema turns a wall of text into structured facts the model can reason about.

The schema that moves the needle most for content sites: Article with a real author object (Person schema with sameAs links to LinkedIn, ORCID, or industry profiles), FAQPage on any page that contains genuine Q&A, and HowTo on step-by-step guides.

What hasn’t changed

This is the part that gets lost in the AI panic. The fundamentals of SEO copywriting are still the fundamentals.

Search intent still wins. If someone searches “best CRM for small agencies”, they want a comparison, not a 3,000-word philosophical piece on relationship management. Page-type mismatch is still the single biggest reason well-optimised pages fail to rank, and AI Overviews have actually made this worse, they now hoover up the informational queries that comparison pages used to rank for as a side-effect.

Internal linking still works. Hub-and-spoke architecture, where a pillar page links out to cluster pages and back, is still the most reliable way to build topical authority. AI systems read your link graph the same way Google does, pages that are well-connected to other relevant pages get treated as more authoritative.

Title tags and H1s still carry weight. A clear, query-matching title is still the first thing both Google and an LLM look at. The advice to “write for humans, not search engines” is still a half-truth, write a title that a human would click and that contains the query they searched.

Page speed and Core Web Vitals still matter. INP replaced FID in 2024 and it’s the metric most sites are still struggling with. A page that ranks but takes four seconds to become interactive is a page that loses clicks to whoever’s in position two.

The 2026 SEO copywriting checklist

What I do on every piece of long-form content now, in order:

1. Match the SERP, not the keyword. Pull the top ten results. If they’re all listicles, write a listicle. If they’re all guides, write a guide. Trying to rank a comparison page for an informational query is wasted effort.
2. Write at least one passage per H2 that could stand alone. 40-80 words. Plain language. Directly answers a likely question. This is your AI-citation surface area.
3. Front-load the answer. The first sentence under each heading should state the conclusion. Supporting detail follows. AI systems are biased toward the first hundred words of any section.
4. Add specific, falsifiable claims. Numbers. Dates. Names. “Three of my four ecommerce clients saw INP drop below 200ms after switching to a lighter theme” beats “many sites see improvements”.
5. Write an author bio that means something. Real name, real credentials, real link to a LinkedIn profile that has activity on it. Person schema in the JSON-LD, with sameAs pointing at the same profile.
6. Ship the schema. Article + Person + FAQPage where relevant. Validate it in the Rich Results Test before publish, not after.
7. Internal-link aggressively but honestly. Every long-form post should link to at least three other pages on the site that are genuinely useful to the reader. Anchor text should be descriptive, not “click here”.
8. Update, don’t just publish. A page from 2024 that you’ve updated and re-dated in 2026 with new sections, new data, and new examples will outperform a fresh post nine times out of ten. Freshness is a signal; depth-of-history is a stronger one.

The thing nobody wants to say

Most SEO content in 2026 is worse than it was in 2022. The reason isn’t that Google or AI killed it, it’s that everyone started running drafts through ChatGPT without editing, and the resulting beige paste of “in today’s fast-paced digital landscape” content is being out-published at a rate humans can’t keep up with.

That’s actually the opportunity. The bar for “obviously written by someone who knows what they’re talking about” has never been lower. A page with a real opinion, real numbers from real work, and a real human voice now stands out from the noise more sharply than at any point in the last decade.

SEO copywriting isn’t dead. The lazy version of it is. What replaced it is something closer to journalism, show your work, name your sources, write things you’d be willing to defend in a room full of practitioners. That’s what ranks now. That’s what gets cited. And that’s what will still be working in 2028.