Technical GEO (generative engine optimization)

ChatGPT can't read your website

80–88% of informational queries now trigger an AI summary, and most queries no longer produce a click. Yet a great many sites are blank to AI crawlers — blocked by JavaScript, by a firewall, or by their own robots.txt.

Scan first. Decide about us afterwards.

Three ways to be invisible — owners usually know about none of them

Blocked by JavaScript

GPTBot, ClaudeBot and PerplexityBot do not execute JavaScript. On a client-side rendered site they see a blank page. Only Googlebot renders JS — so ranking fine on Google tells you nothing about whether AI can see you.

Vercel / MERJ, analysis of 500M GPTBot fetches

Blocked by the firewall

Bot protection like Cloudflare can return 403 to AI crawlers on default settings. Nothing in robots.txt reveals it and you are never notified. The site is fine for humans and a 403 for ChatGPT.

Verifiable with the scanner on this site

Blocked by yourself

Many sites added crawler blocks to robots.txt during the 2023–2024 'stop the scrapers' wave and never revisited them. Today that line means opting out of AI search.

One line to fix

Where we differ from other GEO vendors

The usual GEO service sells an 'AI visibility score' and a dashboard. We think that number cannot currently be measured — and the vendors selling it can't prove it either.

Why visibility scores are unreliable

Fewer than 1 in 100 runs of the same prompt returns the same set of brands
40–60% of cited domains change within a month
A move in citation share from 8% to 11% is statistically indistinguishable from noise
A review of 45 studies found no technique with a stable, long-term, cross-platform effect

Rand Fishkin's 2026 experiment with 600 volunteers; 2026 review of 45 studies

So we only sell what can be verified

Whether the crawler can fetch you — one curl proves it, and it's binary
Whether the firewall blocks it — refetch under a different identity and you know
Whether structured data is emitted — it's in the source
How fast you respond — that's a number

Fix these and visibility becomes possible; leave them and it is impossible. We make 'possible' true. We do not promise rankings we don't control.

We only do what the evidence supports →

Services

Three tiers. The difference is whether we tell you, or we do it.

Technical audit

NT$ 25,000one-time

We inspect, prioritise and explain. You or your engineers do the fixing.

Fix engagement

NT$ 90,000one-time

The audit, then we implement the fixes and verify them in your environment.

Retainer

NT$ 45,000per month

For sites that keep shipping. This debt comes back, especially after a front-end rewrite.

Evidence-backed

Server-side rendering	Binary prerequisite — client-side rendered sites have zero AI visibility	Vercel / MERJ，5 億次 GPTBot 抓取
Statistics and citation density	+37% to +41%	KDD 2024 同儕審查（Princeton / Georgia Tech / IIT Delhi / Allen AI）
Third-party brand mentions	0.664 correlation (backlinks only 0.218)	Ahrefs，75,000 個品牌研究
Answer-first structure	Direct statements ~2x more likely cited; start/end outperform the middle by 30%+	Stanford TACL 2024；Kevin Indig，120 萬則 ChatGPT 引用
Schema markup	+22%; FAQPage nearly doubles ChatGPT citations	Semrush，304,805 個 URL；SE Ranking
Topical clusters / fan-out coverage	+161% citation likelihood	Surfer SEO，10,000 關鍵字研究

See the full evidence inventory →

Questions

What is GEO, and how does it differ from SEO?

GEO is generative engine optimization: getting cited by ChatGPT, Gemini, Perplexity and Google's AI summaries rather than ranking in a list of blue links. The overlap with SEO is large — roughly 80% of AI visibility is solid fundamental SEO — but a few differences matter: AI crawlers don't execute JavaScript, 62% of AI citations come from outside the top 10 organic results, and ChatGPT and Perplexity share only 11% of their source domains.

We rank well on Google. Do we still need this?

Yes — and ranking well makes the problem easier to miss. Googlebot executes JavaScript; GPTBot, ClaudeBot and PerplexityBot do not. A React or Vue site can rank first on Google and be completely invisible to ChatGPT at the same time. The two are not connected.

What does the free scan do to my site?

It reads three public resources only: the page you give it, robots.txt and llms.txt, each fetched once under the GPTBot identity. It does not log in, submit forms, crawl the whole site, or retain your content. If the page returns a non-200 to GPTBot, we refetch once as an ordinary browser for comparison — that is the only way to tell whether a firewall is targeting the crawler.

Want us to fix it?

This report is already a complete problem list. Hand it to your own engineers, or hand it to us.

Scan my site free Contact