AI Prompt Generator

How to actually test your AI visibility manually

Start with unbranded prompts. Ask what the best options are in your category, how to choose, and what the differences are among leaders. Note who appears, in what order, and whether the advice is accurate. Then run brand-specific prompts: is your company known, how is it described, and what alternatives are suggested? Gaps between those two modes tell you whether you have a category-association problem, a reputation problem, or both.

Reduce personalization bias when you can: open a fresh chat, use guest or temporary modes if available, and avoid leading the model with your company name before unbranded tests. History and memory features can make answers look better (or stranger) than a cold start. Still, no manual session is a laboratory — treat results as directional samples.

Repeat the same prompts across ChatGPT, Perplexity, Gemini, and Claude when it matters. Retrieval-heavy products often surface different sources than chat-first products. Log what you see (the checklist on this page is a temporary notepad — copy it out). When patterns repeat — never mentioned, mentioned with wrong category, buried under competitors — prioritize entity consistency, third-party proof, and crawl access. Manual testing builds intuition; structured free tools and paid audits turn that intuition into a plan.