CiteGraph

CiteGraph (citegraph.app) shows whether AI assistants recommend your product, why they don't, and then does the work to fix it. It asks ChatGPT, Claude, Gemini and Perplexity the questions your buyers really ask, several times each, keeps every answer word for word, and writes the pages for the questions you lose. THE PROBLEM Buyers now ask an assistant "what's the best tool for X" and get three or four names. If you are not one of them, you were never considered, and your analytics will not tell you. SEO tools cannot see it, and checking by hand misleads: the same question gives a different answer every time, so one check is an anecdote. WHAT A SCAN MEASURES Each project (one domain) has 10 real buyer questions, mined from how people phrase the problem in public threads and on your own site. A scan asks each one 3 to 5 times on all four engines through their official APIs with live web search: 120 to 200 answers. You get: Named rate: how often the assistant recommends you Cited rate: how often it links to your own pages Share of voice: which rivals take the answers you don't Sources: every third-party page the engines relied on Every rate carries a margin of error, so you can tell real movement from noise. BLIND SPOTS For every lost question you read the full answer, the rival that won, the reason the engine gave and the pages it cited. CiteGraph drafts the fix: the exact claim your site is missing and where it needs to exist. A side by side view sets your best page against the page the engine cited and lists what theirs has that yours lacks. IT WRITES THE PAGES CiteGraph is not only a tracker. It writes full pages for lost questions, grounded in a ledger of facts read from your own site, so nothing is invented. Starter includes 30 pages a month. AUTOPILOT Connect the GitHub repository behind your site and CiteGraph publishes posts with images on a schedule: 2 a week on Starter, up to 3 a day on Scale. Delivery is a pull request to review, or published for you. Each post is aimed at the engine it suits and scored the way all four engines would read it. Autopilot then checks the post is live, linked from your index and in your sitemap. THE CITABILITY MODEL CiteGraph trains its own model on 1,920 pages the engines cited, from 1,380 websites. Each engine has a type: ChatGPT cites a product's own pages, Claude prefers direct how-tos, Gemini favours this year's ranked lists, Perplexity reads comparisons. The model describes what cited pages look like. It is a taste, never a promise. ALSO INCLUDED Weekly rescans, per-engine history and a digest email Competitor discovery and the phrases AI uses to describe each brand Plays: ranked moves, each with a ready-to-edit draft Site readiness: 15 checks, each with its fix AI crawler tracking: which bots read which pages, verified by IP An MCP server, so your own AI agent can read scans and draft fixes PRICING Starter $29/month: 2 domains, 3 runs per question, 6 rivals, 1 seat Growth $69/month: 3 domains, 5 runs, 10 rivals, CSV export, fact sheets, 5 seats Scale $149/month: 6 domains, white-label reports, priority scans, unlimited seats Every plan includes all four engines and starts with a 7-day trial. Starting it charges $1 to confirm the card, and the plan is not charged until the trial ends. Your first scan runs the moment you arrive. No subscription? Your first purchase is 2 scans for $12, then $12 a scan. 120-DAY GUARANTEE No tool controls what an AI says, so the promise is a refund, not a ranking. If no scan shows ChatGPT, Claude, Gemini or Perplexity naming your product within 120 days of your first payment, every subscription payment from that period comes back. Your side: keep weekly scans on and publish at least three of the pages CiteGraph writes. FREE TOOLS, NO SIGNUP AI visibility checker: one buyer question asked of ChatGPT, with the verbatim answer Live site check: every page on your site judged against what each engine cites AI readiness checker Public category lea