How to show up when ChatGPT recommends products

When ChatGPT, Claude, Perplexity, or Gemini answer a buying question with web search on, they retrieve real pages: review platforms, comparison articles, directories, Reddit threads, and brand websites, then synthesize an answer from what those pages say. If your brand isn't described clearly on the pages they read, you don't get named. It's that direct.

AI crawlers identify themselves: GPTBot and OAI-SearchBot (ChatGPT), ClaudeBot (Claude), PerplexityBot, and Google-Extended (Gemini). If your robots.txt blocks them, none of your content exists as far as the assistants are concerned, no matter how good it is. Check yoursite.com/robots.txt for those names. While you're there, consider adding an llms.txt file: a plain-text summary of what you do, who you serve, and where, in the format engines parse best.

Assistants match brands to questions through condition- and use-case-specific language. A clinic whose site says “multidisciplinary care” loses to one whose pages say “chiropractor for knee injuries and shoulder pain”. A shoe brand that says “engineered comfort platform” loses to one that says “stability shoes for flat feet”. Write the actual problem you solve, in plain words, on the page.

Each engine leans on different surfaces: Gemini reads the Google index and Business Profiles, Perplexity favors Reddit and fresh review pages, ChatGPT and Claude weigh high-authority editorial and review platforms. For a local business that means a complete Google Business Profile and steady reviews; for a product brand, editorial reviews and honest community presence; for software, the major review platforms. The fastest way to know which sources matter for you: look at what the engines actually cite when they answer your customers' questions without naming you.

Schema markup (LocalBusiness, Product, Organization, FAQPage) is how engines confirm what your business is, does, and where it operates. It's a copy-paste task for whoever manages your site, and it disproportionately affects how confidently engines classify you.

AI answers are probabilistic: ask twice, get slightly different names. A single check tells you almost nothing. The honest way to track progress is repeated daily measurement with a confidence interval, so when your number moves, you know it moved by more than noise. That measurement loop, plus recommendations built from the exact sources the engines cited when they skipped you, is what Atvelo does.