GEO Audit (AI Search)
Get cited by ChatGPT, Perplexity, Claude
Generative Engine Optimization is the new SEO. Pantra checks whether AI crawlers can reach your site, whether your schema is structured for citation, and whether your content is in the Q&A format that GPT, Claude and Perplexity pull into answers.
Zero-click search is eating Google
More than 30% of Google searches now end without a click — the AI answer box is enough. Meanwhile ChatGPT gets 3.5 billion queries per month, Perplexity is growing 50% quarter-over-quarter, and Claude is the default assistant inside half of Fortune-500 support tools. If your site is not cite-able by these engines, you are invisible to an audience that is doubling every year.
Every GEO check we run
| Check | Severity | What we check | Why it matters |
|---|---|---|---|
| GPTBot access | Critical | robots.txt allows GPTBot (or does not disallow it) | Without GPTBot access, ChatGPT cannot crawl your content for RAG. |
| ClaudeBot access | Critical | robots.txt allows ClaudeBot and Claude-Web | Claude citation is growing fast in enterprise AI use cases. |
| PerplexityBot access | Critical | robots.txt explicitly allows PerplexityBot | Perplexity is the fastest-growing AI search engine of 2025. |
| GoogleOther / Google-Extended | High | Not disallowed — needed for Google AI Overviews and Gemini answers | Google-Extended is the opt-in token for Google AI training and AI Overviews. |
| Server-side rendering | High | Core content present in initial HTML (not only after JS hydration) | Most AI crawlers do not execute JavaScript. Client-only apps are invisible. |
| Structured JSON-LD | High | Article, FAQPage, HowTo or Product schema present where appropriate | Structured data increases citation likelihood by 3-4x. |
| llms.txt | Medium | Present at /llms.txt, summarizing the site in Markdown for LLMs | The emerging standard (Anthropic, Mintlify) for LLM-friendly site maps. |
| Q&A content structure | Medium | Headings phrased as questions, answers in the first paragraph below | LLMs pull direct question→answer pairs for citations. |
| Bing Webmaster verification | Medium | Bing site verification file or meta tag present | Perplexity uses Bing as its primary index; Bing verification accelerates coverage. |
| Openly licensed content signal | Low | Creative-Commons or equivalent licence visible | LLMs preferentially cite content with clear reuse licences. |
| Authoritative external links | Low | Page links to established sources (Wikipedia, RFCs, academic) | Citation-ready content is co-referenced with authoritative sources. |
How the GEO scan works
1. Fetch robots.txt and llms.txt
We read /robots.txt and /llms.txt and match each AI bot against allow/disallow rules, honoring the specificity order in the standard.
2. Render-parity probe
We compare plain HTML (no JS) with headless-browser-rendered output. If the two differ dramatically, your content is invisible to AI crawlers that do not execute JS.
3. Schema extraction and validation
Every JSON-LD block is parsed, validated against schema.org, and scored for completeness (all required fields, image presence, author attribution).
4. Q&A structure scoring
We detect question-shaped H2/H3 headings and measure the density and length of the first paragraph below each — the shape LLMs actually cite.
5. GEO score and remediation
The GEO score is a weighted 0–100 combining crawler access, render parity, schema depth, Q&A structure and verification signals. Every failing check produces a stack-specific fix-prompt.
Sample fix-prompt
Make my Lovable app discoverable and citable for AI search:
1. Create /public/robots.txt with explicit allow for AI crawlers:
User-agent: GPTBot
Allow: /
User-agent: ClaudeBot
Allow: /
User-agent: PerplexityBot
Allow: /
User-agent: Google-Extended
Allow: /
Sitemap: https://mysite.com/sitemap.xml
2. Create /public/llms.txt summarizing the site in Markdown:
# My Product
> One-line description.
## Docs
- [Getting Started](/docs/start): Install and run in 2 minutes.
## Pricing
- [Pricing](/pricing): Plans from $X/month.
3. Add FAQPage JSON-LD to /faq listing every Q&A on that page.
4. Move the hero headline, subheadline and primary CTA content out of
client-rendered React state and into the initial HTML response.
5. Reshape headings on key pages to be questions ("What is X?", "How
does Y work?") with the answer in the first paragraph below.Stacks we auto-detect
GEO vs. classic SEO vs. no strategy
Classic SEO tools barely touch AI crawlers. "No strategy" means betting that Google keeps 100% of search traffic forever.
| Aspect | Manual | Other tools | Pantra |
|---|---|---|---|
| AI crawler checks | Manual robots.txt review | Rare / none | GPTBot, ClaudeBot, PerplexityBot, Google-Extended |
| Render-parity detection | View-source vs. browser | No | Automated on every scan |
| llms.txt check | — | No | Yes |
| Schema citation readiness | Rich results test | Partial | Full FAQ/HowTo/Article depth |
| Monthly cost | Your time | $99–$449 | From $19 |
Works great with
Which plan includes this
Included on every plan. Pro adds the GEO Dashboard for cross-engine citation tracking.
Common questions
Is GEO really different from SEO?add
Yes. Google ranks pages; AI engines extract passages. SEO optimizes for click-through, GEO optimizes for being cited as the answer. Different formats win.
Do I have to give up Google SEO to do GEO?add
No. The two overlap 70%. Good H-structure, schema, sitemap and fast pages help both. GEO just adds AI-crawler access, render parity and llms.txt on top.
Should I block AI crawlers to protect my content?add
If your business depends on paid access, yes. If your business depends on discovery (most SaaS, blogs, agencies, docs), blocking is self-sabotage.
How do I know if ChatGPT actually cites my site?add
The Pro plan includes the GEO Dashboard that probes ChatGPT, Perplexity and Claude for citations of your domain on tracked queries.
Do all AI engines respect robots.txt?add
Officially yes. OpenAI, Anthropic, Perplexity and Google all document and respect robots.txt directives. Fringe scrapers do not — but they would not bring citation traffic anyway.
What is llms.txt and is it actually used?add
A Markdown file at /llms.txt summarizing the site for LLMs. Proposed by Anthropic and Mintlify, now adopted by Stripe, Vercel, LangChain and hundreds of docs sites. Emerging but worth having.
My site is a pure SPA — is GEO impossible?add
Not impossible, but hard. The fix-prompt will tell you to adopt SSR or static pre-rendering for public routes. Lovable supports this, Next.js is built for it, Bolt and v0 need a migration.
How long until GEO fixes show up as AI citations?add
GPTBot recrawl is typically 2-7 days; Perplexity via Bing 1-3 days once Bing has indexed the update. Plan for a 2-4 week trend line, not overnight.