GEO vs SEO in 2026 — What Google's May Guidance Changed

Originally published on rikuq.com. Republished here for Dev.to's readers.

Google's May 2026 AI search guidance settled one debate: for Google's surfaces, AEO and GEO are "still SEO." Same crawl, same ranking, same E-E-A-T factors. Optimizing for AI Overviews is optimizing for Google search, period.

That sentence is accurate, and it ended a year of speculation about whether Google would build a separate AI crawl or weight different signals on its AI surfaces. They didn't. They reaffirmed that strong fundamentals win everywhere they own.

But the guidance ended one debate while leaving another wide open: what about the engines Google doesn't own?

ChatGPT runs on Bing. Claude runs on Brave. Perplexity runs on its own index. Each of those engines has its own crawl behavior, ranking signals, and citation patterns. None of them issued matching guidance saying "we're also just classical SEO." Optimizing for them genuinely is a separate discipline — what I cover in the four-index reality post.

This post is the honest read on what changed in May 2026, what didn't, and what you should actually do as a content creator or solo founder in the second half of the year.

I'm Ravi. I run Citare, a platform built specifically for tracking AI search visibility across these surfaces, so I've watched this debate from inside the operational reality.

The actual content of Google's May 2026 guidance

Three substantive points worth understanding precisely:

1. AI Overviews + AI Mode + Gemini use Google's main index.

No separate AI crawl. No special "AI-friendly" schema. No llms.txt requirement specific to Google. The same E-E-A-T signals, the same backlinks, the same internal linking architecture, the same site speed considerations — everything that worked for Google search in 2024 works for Google's AI surfaces in 2026.

This is a correction of widespread bad advice from 2024-2025. Many SEO content writers spent the last 18 months selling "GEO-specific" optimizations (custom JSON-LD types, AI-optimized schema, content rewrites for "LLM consumption"). Google's guidance was effectively: stop. None of that helps. Foundational SEO discipline does.

2. E-E-A-T weight increases on AI surfaces specifically.

The model has to decide what to cite. Weak-signal sources — anonymous content, content without verifiable authorship, content from sites without authority signals — get filtered out of AI Overview citations even when they rank on regular Google search. The honest implication: investing in real author identity, real bylines, and real entity verification is no longer optional for content that wants AI Overview presence.

3. Originality and first-party data are now table stakes.

The March 2026 core update already started rewarding originality. The May guidance made it explicit: rephrased takes on existing content collapse on AI surfaces faster than they do on classical search. The model can recognize when an article is summarizing what 50 other articles already said. AI surfaces don't cite restatements; they cite the source.

For content creators this means: every post needs at least one piece of first-party data (a number you measured, a screenshot you took, a verdict you reached) that no competitor can copy. If your post is rewritable from publicly available sources, AI engines will cite the publicly available sources, not you.

What Google specifically did NOT cover

The guidance was Google-shaped. They covered Google's surfaces and what works there. They didn't address:

ChatGPT/Bing behavior. Microsoft hasn't issued matching guidance.
Claude/Brave behavior. Anthropic hasn't published optimization guidance.
Perplexity behavior. Perplexity has issued some guidance but it's narrower than Google's.
Citation behavior across surfaces — Google's guidance was about ranking, not about how AI engines surface citations to users. Citation UX varies dramatically (Perplexity shows inline citations, ChatGPT shows linked sources at the bottom, Claude shows sources in a side panel, AI Overview shows a few cards).

The half of the AI search world Google doesn't own is still a genuinely different discipline. Anyone telling you "GEO is dead, just do classical SEO" is half-right and half-wrong — right about Google, wrong about the other ~40 percent of AI-mediated discovery.

The "GEO is still SEO" framing — when it's true vs not

True when:

You're optimizing for Google search (any surface — regular results, AI Overview, Gemini, knowledge panels)
Your audience primarily discovers content via Google
You're investing in fundamentals (E-E-A-T, backlinks, structured data, originality)

False when:

You need visibility in ChatGPT, Claude, or Perplexity specifically
Your brand is cited differently across AI engines (which it is, almost always)
You're trying to win citation slots in AI engine responses, not just ranking on classical search results
You're tracking AI search visibility, not just Google rankings

The honest one-line summary: GEO is still SEO inside Google's universe. Outside it, real multi-engine GEO is a distinct discipline.

What's genuinely changed in the optimization playbook

Five concrete shifts since 2024:

1. Padding for word count is dead.

AI extraction punishes filler. Articles bloated to "rank for length" now lose to tighter articles that say more in fewer words. Editorial discipline > length.

2. First-party data went from nice-to-have to table stakes.

Every serious post needs at least one number, screenshot, dataset, or experiment that no competitor has. AI engines cite the source of facts; if you're not the source, you're not the citation.

3. Real author identity matters more.

Bylines with photos, bios linking to verified social profiles, Person schema with sameAs entries to other owned properties (the entity graph). AI engines weight this heavily for citation trust.

4. FAQ schema is now extraction infrastructure.

Not just a Google SERP feature — AI engines extract FAQ Q&A pairs directly into responses. Articles with well-structured FAQs get cited more across AI surfaces, regardless of underlying engine.

5. "Last updated" dates became a citation factor.

Recency signals weight heavier in AI surfaces than they ever did in classical SEO. Articles with visible "Last updated [date]" and actual updates get cited; stale articles get filtered out even when they rank.

What did NOT change (and won't)

Classical SEO fundamentals still drive most of the work:

Backlinks still matter — especially on Bing (ChatGPT's index)
Schema markup still matters — especially Article, FAQPage, BreadcrumbList, Person
Site speed still matters — Core Web Vitals are a Google signal still
Internal linking still matters — for both classical SEO and AI engine context understanding
E-E-A-T still matters — now more than ever
Content quality still matters — and is more measurable now via AI extraction patterns

If you've been doing SEO right, 80 percent of your existing work still applies. The 20 percent shift is what this post is about.

The practical playbook — what to do as of May 2026

Universal (do these regardless of engine):

Real author identity on every post. Photo, bio, Person schema with sameAs to your other properties.
At least one first-party data point per post. Number, screenshot, dataset, verdict.
FAQ schema with FAQPage JSON-LD on every post where Q&A pairs are natural.
Visible "Last updated [date]" prominent at the top of every post. Update quarterly.
Tight, decisive writing. No hedging. No filler for word count.
Clean technical foundation. Site speed, structured data, internal linking, mobile-friendly. The basics.

Google-specific (works for AI Overview + Gemini + classical):

Heavy investment in E-E-A-T — author credentials, demonstrated expertise.
Backlinks from authoritative domains in your topical neighborhood.
Topical cluster building — comprehensive coverage of related questions, internal linking between them.

Multi-engine GEO (works for ChatGPT/Claude/Perplexity beyond Google):

Bing-specific: invest in backlinks (Bing weights them heavier than Google does today).
Brave-specific: invest in freshness and independent sources (less domain-authority bias than Google).
Perplexity-specific: invest in citable inline facts — specific numbers, named entities, extractable claims.
Track per-engine visibility — single-engine rank tracking misses 40-60 percent of where your brand appears. Use a tool that watches all four indexes.

Skip these (they don't help):

Custom "AI-optimized" schema beyond the standard types — no evidence any engine rewards it.
Content rewrites specifically "for LLM consumption" — wasted effort; quality content works everywhere.
Heavy llms.txt investment beyond shipping a basic one — uncertain ROI.
"AEO-specific" agencies promising 10x AI search visibility — nothing structurally separates this from good SEO + multi-engine awareness.

The honest summary

Google was right about Google. Their AI surfaces use the same ranking as classical search, and the optimization is the same discipline. The 2024-2025 "GEO is a totally separate field" advice was largely wrong, and Google's May guidance correctly buried it.

Google was incomplete about the non-Google engines. ChatGPT (Bing), Claude (Brave), and Perplexity all behave differently, have different optimization profiles, and require separate attention if you want presence on those surfaces.

The practical work in 2026 is: nail classical SEO fundamentals (they're now table stakes), invest in first-party data and real author identity (the post-March-2026 differentiation), and track per-engine visibility across all four indexes if you care about the half of AI search Google doesn't own.

Anyone selling you a more complicated story than that is selling you something. The discipline is straightforward; the execution requires consistency.