ChatGPT visibility, measured daily

Is ChatGPT recommending you, or your competitors?

Somewhere in the past hour, a buyer in your category asked ChatGPT what to use, and ChatGPT answered with a handful of names. Geonimo asks ChatGPT the questions your buyers ask, every day, and measures which brands come back, yours or your rivals.

One check is an anecdote: ChatGPT is non-deterministic, so the same question returns different brands on different days. Real tracking is repeated sampling: the same prompts, daily, aggregated into rates and trends with confidence intervals.

Free check · No signup · Verifies GPTBot & OAI-SearchBot access + llms.txt in seconds

What Geonimo measures on ChatGPT

The answer a real buyer sees

Geonimo samples ChatGPT at the consumer surface: live grounding, citations, shopping carousels and ads included, not a stripped-down API view that behaves differently.

Mentions, position, sentiment

Every tracked brand is extracted from every answer with its position and the sentiment of the surrounding text: being named first and named last are different outcomes.

Grounded vs from-memory

ChatGPT answers many questions from model memory without searching the web. Geonimo reports the grounded share, so you know which content changes can move numbers now and which face model memory.

Fan-out and citations

When ChatGPT searches before answering, Geonimo captures the queries it ran and the sources it cited or silently read, usually where the "why am I not mentioned?" answer lives.

Why a single ChatGPT check misleads

Ask "best CRM for consultants" once and you have one sample of a distribution, not a measurement. Geonimo runs your prompt set daily and publishes every share with a 95% confidence interval; day-to-day movement inside the noise floor is labeled "ns", never sold as a win. Visibility is measured on unbranded buyer questions only, because a prompt that names your brand can only produce an echo.

What ChatGPT reads before it answers

This is not speculation: OpenAI’s own help documentation says ChatGPT decides on its own to search when a question can benefit from current information, and typically rewrites your prompt into one or more targeted queries before searching. Geonimo records those queries and the cited sources, then fetches every cited page to check whether it actually names you. A trusted, repeatedly-cited page that names competitors but not you is your highest-leverage outreach target.

How to track your brand in ChatGPT, step by step

  1. 1Write 20 to 40 unbranded buyer questions in your customers’ words: recommendation prompts ("best X for Y"), comparison prompts ("X vs Y for Z") and problem prompts ("how do I..."). Never include your own brand name; a branded prompt can only produce an echo.
  2. 2Ask each prompt in a fresh session with memory and custom instructions off, so earlier conversation cannot contaminate the answer.
  3. 3Record five things per answer: mentioned or not, position among the recommendations, how you were framed, which competitors were named, and the sources cited when ChatGPT searched the web.
  4. 4Repeat on a fixed schedule and track rates over time, never single answers. ChatGPT is non-deterministic, so only repeated sampling separates real movement from churn.
  5. 5Aggregate each rate with a confidence interval before reporting it, so a two-point wiggle inside the noise floor never gets sold as a win.

Done by hand, this costs about an hour per week for 30 prompts and stops working the week someone gets busy. Geonimo runs the same protocol daily across ChatGPT, Google AI Mode and Perplexity, with the extraction, competitor detection and statistics handled for you, and new prompts analyzed within minutes rather than on tomorrow’s schedule.

The ChatGPT visibility metrics that matter

  1. 1Mention rate: the share of your tracked buyer prompts where ChatGPT names you, published with a 95% confidence interval. Your single baseline number.
  2. 2Share of voice: your mentions divided by all tracked brands’ mentions on identical prompts. The honest competitive KPI, because your score can rise while your share falls.
  3. 3Position: first recommendation and last mention are different outcomes. Recommendation lists are ordered, so ordinal position is tracked per answer.
  4. 4Sentiment and framing: "the industry standard" and "a budget option with mixed reviews" are both mentions. Only one wins deals.
  5. 5Grounded share: how often ChatGPT searched the web before answering versus answering from model memory. This split tells you which content changes can move numbers now.
  6. 6Verified citations: the pages ChatGPT reads and cites, credited to your brand only after the page is fetched and actually names you.

Why the effort is worth it: Semrush’s traffic study measured the average AI search visitor as 4.4x as valuable as the average organic search visit, based on conversion rate. Fewer sessions, far more intent.

Manual checks vs DIY scripts vs a tracker

Manual sampling is free and works as a first baseline, but weekly sampling of a noisy distribution needs a month or two to distinguish movement from variance. DIY API scripts add consistency, with two traps: the API without search tools behaves differently from the ChatGPT surface real buyers use, and mention extraction is harder than string matching (brands appear with typos, abbreviations, or as domains). Someone also has to maintain the script through every model change.

A purpose-built tracker removes both problems: Geonimo samples the consumer surface daily, extracts mentions against your tracked brand and competitor list with position and sentiment, verifies citations against the cited pages, and publishes every share with its confidence interval.

What a good ChatGPT visibility score looks like

There is no universal benchmark: a category with three credible brands concentrates mentions, a category with thirty fragments them. The honest yardsticks are your trend against your own baseline and your share of voice against the top competitor on the prompts that drive revenue. Category leaders typically take a large share of the recommendation slots on their money prompts; if a rival holds the first-recommendation position on prompts you should own, that specific gap, on those specific prompts, is the number to move.

ChatGPT tracking, answered

Does ChatGPT give everyone the same answer?

No. The same prompt asked twice can return different brands, in different orders, with different framing, and behavior shifts with web search on or off and after model updates. That is why Geonimo samples daily and reports trends with confidence intervals instead of one-off checks.

How can I track brand mentions in ChatGPT for free?

Manual sampling works as a baseline: 20–40 real buyer prompts in fresh sessions, weekly, recorded consistently. Our guide walks through it, and the free Geonimo audit automates the first pass with results within 24 hours.

Do you track the ChatGPT API or the real product?

The real product surface. The API without search tools behaves differently from what ChatGPT users actually see, so measuring it would measure the wrong thing.

Is this a ChatGPT rank tracker?

Yes, in the sense that matters: Geonimo tracks the position of your brand within ChatGPT answers (first recommendation vs also-mentioned), your mention rate, and your rank against competitors on the same prompts. There are no classic keyword rankings in AI answers, so position-in-answer is the honest equivalent.

How many prompts do I need to track ChatGPT properly?

Most brands start with 20 to 40 real buyer prompts covering recommendation, comparison and problem questions, then expand by product line or market. Geonimo plans include generous prompt allowances and every prompt runs daily across all three engines.

Does ChatGPT search the web for every answer?

No. A large share of answers come from model memory without a live search. Geonimo reports the grounded share for your prompt set, which matters because content fixes can move grounded answers quickly while memory-based answers respond slowly, through the model’s training and brand priors.

How much does ChatGPT visibility tracking cost?

Geonimo starts at $199/month ($159/month billed yearly) with all tracked engines included on every plan and no per-engine add-on fees. The free audit comes first, so you see your real ChatGPT baseline before paying anything.

Which metrics matter for ChatGPT visibility?

Mention rate (with its confidence interval), share of voice against competitors on identical prompts, position within the answer, sentiment, and the citations behind the answers. Geonimo reports all five daily.

See your ChatGPT baseline this week

Book a free audit: first insights on your ChatGPT visibility within 24 hours. No credit card required.

Book your free audit