What could bring
you more customers?

It fetches one public HTML page and its robots.txt. It checks the HTTP response, declared Googlebot and OAI-SearchBot access, indexing directives, title, description, H1 headings, canonical URL, image alt attributes, and JSON-LD presence. It reports training controls separately and suggests up to three things to review.

The technical check does not measure live AI recommendations. The question planner uses your page title as an editable starting point, plus your offer, audience and market, to suggest 12 template-based content opportunities. You can copy a question into an AI service yourself. Arrow does not query AI services or collect their answers; one observed answer is not an overall visibility score.

Yes, manually. Run a question in your chosen AI service and paste the answer, source URLs, date and platform into the observation tracker. It detects brand and competitor text matches and checks whether your domain appears in the source URLs you supply. It does not verify recommendations, fetch AI answers, or calculate a general visibility score. Observations remain in memory in this tab until reload or close; export them before leaving.

After a successful website check, the planner shows the inspected URL, page title, main heading and description. It suggests an editable topic from the page title. The site check prepares five starter questions from your customer goal and a suggested page-title topic. Open the refinement controls to rewrite the topic or add your audience and market. Seven more questions are available on demand. Each has an intent and an optional content angle. These are local templates, not live keyword research or measured demand.

Yes, the controls are separate. OpenAI uses GPTBot for training and OAI-SearchBot for ChatGPT Search. Google-Extended does not control Google Search inclusion or rankings. Review the official documentation before editing your robots.txt and choose the rules that match your preferences.

Google does not require special AI markup or AI text files for its AI search features. Use relevant structured data that accurately represents visible content. This checker does not penalize a missing llms.txt or assume every page needs FAQPage markup.

The check is free and does not require an account or email. It is limited to 12 checks per connection per hour, subject to a shared service limit. Your report and conversation remain in the current tab; the checker does not create a public report page. Hosting access logs may still exist. Query strings and fragments are removed from submitted URLs. An optional Arrow review requires your email and agreement to send your URL, goal and summary to Arrow. Usage events are measured only with analytics consent; pasted AI answers are not sent by the observation tracker.

The report guide uses a fixed explanation library and the observations in your current check. It can explain priorities, crawl rules, training preferences, titles, structured data, and image alternatives. It does not call an AI model or independently verify claims outside the inspected page.

A firewall, login, timeout, redirect chain, unsupported response, or page larger than 1 MB can prevent inspection. This checker supports public IPv4 destinations over standard HTTP or HTTPS ports and does not execute JavaScript. An unsuccessful check does not prove that Google or an AI crawler is blocked.

Start with unintended indexing restrictions and missing page essentials. Then make the page useful: answer the real question clearly, keep business facts accurate, use descriptive headings and informative image alternatives, and link related resources. Technical readiness supports discovery; it cannot guarantee impressions, rankings, or citations.