ChatGPT

OpenAI's ChatGPT answers current questions with browsing enabled, and its citations are drawn from the Bing index. Roughly 87% of the URLs ChatGPT cites match a Bing top-10 result, which makes Bing ranking the single strongest lever on whether you are surfaced here.

How each answer engine grounds its answersSeven answer engines on the left connect to their grounding sources on the right: ChatGPT to the Bing index; Google AI Overviews, AI Mode and Gemini to the Google index and Search; Perplexity, Grok and Claude to live web search.ChatGPTGoogle AI OverviewsGoogle AI ModePerplexityGeminiGrok (xAI)ClaudeBing indexGoogle index / SearchLive web search

How it grounds

When a prompt needs fresh information, ChatGPT issues search queries against the Bing index, pulls a handful of top results, and synthesises an answer that cites them as source cards. Pages that never reach Bing's first page are rarely quoted, so ChatGPT visibility tracks Bing organic rank and freshness far more closely than Google rank.

What moves the score

  • Bing top-10 rank — the dominant factor; if you are not on Bing's first page for the query, you will rarely be cited.
  • Freshness — recently updated pages are preferred for time-sensitive prompts.
  • Quotable content — adding quotations, statistics and cited sources lifts extraction and citation odds (about +30–41% in the Princeton GEO study).
  • Earned-media corroboration — being referenced on Reddit, Wikipedia or G2 raises the odds an engine trusts and names you.

In Lokrix

ChatGPT is one of the five engines that run live by default. Each prompt is sampled K times across temperature, persona and locale, producing a presence probability with a 95% confidence interval rather than a single answer. See live vs. predictive for when live runs are used, and the levers for how to improve here.