Short answer: ChatGPT and Perplexity, when they search the web, name the vendors whose own pages answer the developer's exact question. One page per task, plain pricing and limits, honest comparison pages, an llms.txt file and readable docs. On our own data API that moved ChatGPT answers citing us from 1.2% to 8.8% of tracked questions in about two weeks.
A developer asks "which API has historical prices for X, and what does it cost". The assistant runs a few searches, reads the top pages and quotes the ones that answer directly. Vendors get named when a page:
We run a trading-card price data API launched in September 2026. Each week we asked ChatGPT (gpt-4.1-mini with web search) and Perplexity the same 160 to 170 buyer and developer questions and logged every citation.
| Sep 28 | Oct 13 | |
|---|---|---|
| ChatGPT answers citing us | 1.2% (n=160) | 8.8% (n=170) |
| Developer and business questions citing us | 0 of 20 (Oct 6) | 4 of 20 |
All three of its first outside customers found it through AI answers. It is a small, early case, and we share the raw questions and answers on request.
We run this as a 30-day pilot for data and API companies: the 20-question check, 5 answer pages shipped with your approval, and a day-30 re-run. See the pilot, or email your API's URL and the 3 questions you most want to win to ops@maehwalabs.com for a free 10-question check.