44.2% of LLM citations come from the first 30% of a page. AI Overviews cut position-1 organic click-through rate by 54%. Search Engine Land published the numbers this week and they collapse a year of GEO hand-waving into one rule: the first 300 words of your homepage are no longer your hook — they are the cell the model copy-pastes when a buyer asks ChatGPT, Claude or Google AI Overviews about your category. Hub365 calls that band the citation injection zone, and we treat it like inventory.
What just happened
Search Engine Land published a fresh GEO benchmark on May 16, 2026 and the four numbers underneath the headline are the ones a business owner needs.
- 44.2% of all LLM citations originate inside the first 30% of a page's text. Top-of-page is no longer a stylistic preference — it is a citation injection point.
- ChatGPT cites a source 87% of the time but only mentions the brand inside the answer 20.7% of the time. Gemini does the inverse: it mentions the brand 83.7% of the time and links 21.4%.
- Google AI Overviews now hit 61% brand mention and 84.9% citation rate — the gap with the chat engines is closing fast.
- Position-1 organic CTR fell from 1.41% to 0.64% when an AI Overview appears above it. A 54% click loss on the slot every SEO agency has been promising for fifteen years.
The benchmark was compiled mid-May 2026 and Search Engine Land is calling it the inflection of "answer-engine majority" — the point at which most commercial queries are answered above the blue links, not by them.
Why it matters for a 5 to 50 employee business
The number that hurts is the 54%. If you are an HVAC company in Dallas, a dental clinic in Mexico City or a bilingual agency in Miami, you have been paying somebody to chase position 1 for the last decade. That slot just lost half of its clicks to a paragraph above it that you did not write. The replacement game is being the brand the paragraph is quoting.
This is the asymmetry Hub365 keeps writing about. Enterprises get the consulting-mediated rollout — OpenAI just formalized the Deployment Company with TPG, Bain, Capgemini and McKinsey and acquired Tomoro's 150 Forward Deployed Engineers; Anthropic expanded its PwC alliance the same week. SMBs do not get a deployment partner from the model labs. They get a free workshop and a "good luck." The GEO collapse is the same shape: enterprise marketers will retainer a brand-mention agency in Q3. The 5 to 50 employee business needs a packaged version of the same work, billable in week one.
The Hub365 read
We rewrote our own first 300 words in March. The discipline is not literary — it is logistical. Three rules survived the rewrite.
- One quotable claim per page, above the fold. The model is looking for a sentence it can lift verbatim. If your homepage opens with "We help businesses grow," there is nothing to lift. Ours opens with what Hub365 ships, in numbers, in the first paragraph. The same rule applies to your product pages, your About, and your FAQ.
- A real FAQ with five to six questions per page, schema-marked. AI Overviews and ChatGPT both lean on FAQ schema when they need to attribute. Hub365 emits FAQPage JSON-LD on every page now. It is not optional anymore.
- An /llms.txt at the root. This was 3% of Fortune 500 .com domains in March. It is 14% as of last week — Stripe, Cloudflare, Atlassian, Shopify — with zero press releases. The answer-engine equivalent of robots.txt is becoming table stakes silently. Hub365 ships one with every GEO retainer.
The piece nobody else is selling yet: tracking the citations. We measure mention rate and citation rate on a monthly cadence across ChatGPT, Claude, Perplexity, Perplexity Comet and Google AI Overviews for the queries that route to a quote in your industry. The number on the dashboard is not "are we ranking" — it is "are we being cited." That is the metric that survives the next twelve months.
What to do this week
You do not need to wait for a retainer to start. Three moves, in order, billable to your own time.
- Audit your first 300 words. Open your homepage. Count the words above the first H2. Read the first sentence and ask: would a model lift this verbatim? If the answer is no, you have a citation injection zone but nothing in it.
- Replace one paragraph with one number. Pick the most concrete fact about your business — average response time, number of clients, the result you are best at — and put it inside the first 100 words. Do not soften it with "we believe" or "our mission is." The model is looking for a noun, a verb and a number.
- Ship a five-question FAQ at the bottom of your top three pages. Same questions a buyer actually types into ChatGPT. Mark them with FAQPage schema. If your CMS does not emit FAQ schema, that is the project — not another blog post.
If you want to skip the audit step, Hub365's GEO audit reads your homepage, top 5 product pages and FAQ in 48 hours and ships back the rewrites, the FAQ schema and the /llms.txt as one deliverable. It is the same playbook we run on hub365.ai every week — same Blog Bible, same citation injection rule, same brand voice gate. The same playbook lives one click deeper inside the Hub365 AI visibility playbook if you want to read before you book.
What is the citation injection zone? The first 30% of a page — roughly the first 300 words. Search Engine Land's May 2026 benchmark shows 44.2% of all LLM citations originate inside that band. If your best line is below it, the model is statistically unlikely to lift it.
Did SEO die? No. Position-1 organic CTR dropped from 1.41% to 0.64% when an AI Overview appears — a 54% click loss for the top result. SEO still routes traffic; it now routes a shrinking slice. The new goal is being cited inside the answer, not just ranked above the fold.
What is GEO and how is it different from SEO? GEO — Generative Engine Optimization — is the practice of being cited by LLMs (ChatGPT, Claude, Gemini, Perplexity, AI Overviews) inside the answer they generate. SEO optimizes for clicks; GEO optimizes for being inside the answer the user reads instead of clicking.
Does Hub365 rewrite my whole site? No. The GEO audit reads your existing homepage, top 5 product pages and FAQ, and rewrites only the first 30% of each plus adds an FAQ schema and an /llms.txt at the root. You keep your brand, your copy, your CMS.
How do I know it worked? Hub365 tracks mention rate and citation rate per query across ChatGPT, Claude, Perplexity, Perplexity Comet and Google AI Overviews on a monthly cadence. The number that matters is "cited inside the answer for the question your buyer types," not "ranked #1 in Google."
Where do I start? Send CITED on WhatsApp for a 15-minute scan of your homepage's first 300 words. If the citation injection zone is empty, we will tell you on the call. If it is already loaded, we will tell you that too.
Sidebar — the four GEO numbers a 50-person business has to know
- 44.2% — share of LLM citations that come from the first 30% of a page.
- 54% — drop in position-1 organic CTR when an AI Overview appears above it.
- 84.9% — Google AI Overviews citation rate, up sharply this quarter.
- 300 words — the working size of the citation injection zone on most homepages.
The model labs gave the Fortune 500 a deployment partner this week. They gave the 5 to 50 employee business a 54% click loss and a free workshop. The citation injection zone is the cheapest fix on the board — and the first place Hub365 ships when a new client lands. The page you already own is the surface. The first 300 words are the inventory. The model is reading them right now.
