Guide for buyers of tracking
A ChatGPT brand tracker asks a fixed set of buyer questions on a fixed schedule and records, for every answer, whether your brand was named, in what position, in what tone, and which sources the answer cited. Anything that produces a single score without letting you open the answers behind it is not a tracker; it is a number. This page explains what to demand from any tracker — including ours — and when you do not need one at all.
| Measurement | Why it is non-negotiable |
|---|---|
| Named-rate, per question, per engine | Repeated identical prompts return different brands 40–60% of the time month to month. A one-off check is a coin flip; only a rate across several runs means anything. |
| Position and recommendation class | Being mentioned fifth with a caveat is not the same as being recommended first. The sentence that decided the classification should be shown, not summarised. |
| Citations, at URL level | In our own audit of 232 logged answers across three engines, 92% of cited sources were third-party pages, and 157 of the 196 answers naming the brand did not cite the brand's own site. The work is mostly off-site; a tracker that only shows a score cannot tell you where. |
| The verbatim answers | Every number should open to the answer that produced it. If you cannot audit the evidence, you cannot defend the metric to your board — or notice when the tool is wrong. |
| The market you actually sell in | Location decides whether Google's AI even answers: in our runs, "best hair transplant clinic" asked from Lisbon returned an AI Overview naming local clinics; the identical question from London returned none. A tracker that measures from the wrong country describes somebody else's market. |
The fastest way to evaluate any tracker is to look at a real output. See a real BrandsNode report — actual questions, actual answers, actual citations, with the method visible.
If you have one brand, one market, and an afternoon a month, you can run this yourself: our guide How to track your brand in ChatGPT is the full manual method — the same method any honest tool automates. A spreadsheet stops being enough when any of these is true: you need several runs per question per engine (variance makes single checks worthless); you sell in more than one country or language; you need the citation log, not just the named-rate; or someone above you will ask "and what do we do about it?"
| Option | What you get | Where it breaks |
|---|---|---|
| Spreadsheet + manual runs | Full control, zero cost, real understanding of the method | Variance requires many runs; multi-market and citation logging become a part-time job |
| Generic monitoring tool | Automated named-rate and share-of-voice across many engines, from roughly the price of a lunch per month | Invented prompt sets, engine-count marketing, scores that do not open to evidence, no view of your specific market's buyers |
| Market-level measurement (what we sell) | Questions derived from demand evidence in your market and language, several runs per question, every answer and citation logged, and a monthly readout of where buyers lose you and what to change | Costs more than a generic monitor; only worth it if you will act on it |
We are not neutral about the third row — it is our product. The first two rows are still genuinely the right answer for many teams, which is why we published the manual method in full.
A trustworthy report shows, at minimum: the question set with its provenance; the named-rate and position per engine with the number of runs behind each figure; the tone and the sentence that decided it; the cited sources with who owns them; and a prioritised list of what to fix, tied to the specific questions where you lost. Ours is public: sample report. Compare anything you evaluate against it — including against us.
We will run one brand through the full method — questions derived from your market's demand, asked across ChatGPT, Perplexity, Gemini and Google AI Overviews, several runs each, every answer and cited source logged — and send you the report.
Apply for a free auditOne brand, no cost, no card, no obligation. We reply with the audit or with an honest reason it would not tell you anything useful.