One scan, two lenses. The SEO lens grades your titles, descriptions, headings, and index hygiene. The GEO lens tests what ChatGPT, Claude, and Perplexity actually hit when they fetch you, from a datacenter IP, the way real retrieval works. Twenty-two checks, each with the fix attached.
Reads about a dozen public URLs once. Three scans per minute per visitor. Nothing stored.
Search traffic now splits between ranked links and generated answers. A site tuned for one and blind to the other leaves the growing half on the table, and the two failure modes are completely different.
The classic discipline: does the page state what it is, for whom, in the places crawlers read first?
This scan grades title quality, meta descriptions, heading structure, accidental noindex, viewport, image alt coverage, and favicon presence.
The new discipline: when an answer engine fetches your site from a datacenter, without JavaScript, on a budget, can it read anything at all?
This scan tests reachability as GPTBot and ClaudeBot, server-rendered content, strict JSON-LD, robots.txt access, llms.txt, sitemap, honest 404s, and markdown negotiation.
Every scan runs the same sequence from Cloudflare's network. Weights favor reachability, because a blocked site fails everything downstream.
About a dozen public URLs in three polite waves: the homepage as a browser, as GPTBot, and as ClaudeBot, plus robots.txt, llms.txt, sitemap.xml, a deliberately missing page, a markdown-negotiation probe, and the trust pages (/about, /contact, /privacy).
Twenty-two weighted checks. Rate-limited responses get one retry, and if a site throttles the scan anyway, the report says so instead of silently scoring the wreckage.
The report groups results the way you should fix them: reachability first, then content, then indexing hygiene, then structured data, then the machine layer, then on-page polish.
The full checklist, in fix-first order. Weights are shown as the share of the 100-point score.
| Area | Checks | Weight |
|---|---|---|
| Reachability | Homepage from a datacenter IP; reachable as GPTBot; reachable as ClaudeBot | ~22% |
| Content | Real server-rendered content (h1 + 500 characters without JavaScript); response under 3 seconds | ~15% |
| Index hygiene | No accidental noindex; missing pages return a real 404; trust pages exist | ~18% |
| Structured data | Strictly valid JSON-LD; complete Organization schema (contact, address, sameAs) | ~14% |
| Machine layer | robots.txt lets AI crawlers in; llms.txt exists and is linked; sitemap.xml; markdown negotiation | ~19% |
| On-page SEO | Title length; meta description; single h1; viewport; image alt coverage; favicon | ~12% |
This tool is the free first rung of Growthr's flat-fee SEO/GEO ladder. Each step credits into the next.
Thirty minutes with a live teardown of your site. Credits toward the audit.
Full scored review of technical SEO, on-page, schema, and AI visibility, in 48 hours. Credits toward Boost.
Growthr ships every fix in the plan, about a two-week sprint, for sites up to roughly 25 pages.
Ongoing monitoring and upkeep. Optional, cancel anytime.
GEO, also called AEO or AI search optimization, is the practice of making a site readable, quotable, and recommendable by AI answer engines such as ChatGPT, Perplexity, Claude, and Google AI Overviews. Where SEO optimizes for ranked links, GEO optimizes for being the answer: reachable from datacenter IPs, facts in raw HTML, strict structured data, and machine-readable files like llms.txt.
Ordinary checkers grade on-page SEO. This scan runs that lens plus a GEO lens most tools skip: whether GPTBot and ClaudeBot can reach the site at all from a datacenter IP, whether content is server-rendered, whether JSON-LD parses strictly, whether robots.txt lets AI crawlers in, and whether the machine layer exists.
No. The free scan is the automated pass: 22 checks, one page, 20 seconds. The paid audit is a human review of the whole site, its search data, and its AI-search standing, delivered as a scored report with a prioritized fix plan within 48 hours. The $499 credits toward implementation if Growthr ships the fixes.
In order: reachability, then server-rendered content, then indexing hygiene, then structured data, then the machine layer, then on-page polish. The report groups its sections the same way.