ChatGPT named RankZero in 0 of 559 answers. Perplexity, asked the same questions over the same 30 days, named us 50 times in 900.
That is our own brand. Across all 10 brands we run, the pattern holds: Perplexity named the brand in 29.6% of its answers, Gemini in 22.6%, Google AI Overviews in 16.4% and ChatGPT in 10.1%. Same buyer questions, same brands, same 30 days to 9 October 2026, 31,760 answers in total.
So "how visible are we in AI search?" has no single answer. It has four, and the one most teams check first is usually the lowest. This is edition one of a quarterly report on how often AI engines mention brands: the numbers first, then what to do with them, then exactly how we counted.
Key Takeaways
- Across 10 brands and 31,760 answers (30 days to 9 October 2026), Perplexity named the brand almost three times as often as ChatGPT: 29.6% against 10.1%. It named the brand more often than ChatGPT for all 10 brands.
- ChatGPT was the lowest-naming engine for 7 of the 10 brands (one of the seven a tie). Two brands, ours included, were named 0 times by ChatGPT.
- A single blended visibility score hides this. Track and report each engine on its own.
- Another published study puts ChatGPT first. The difference is what gets counted, so only compare figures measured the same way.
- This edition is a baseline. ChatGPT's tracking model changed in September, so nothing here is a trend yet.
The Engine Gap in One Table
Pooled across the 10 brands, in the 30 days to 9 October 2026:
| Engine | Answers | Answers naming the brand | Rate |
|---|---|---|---|
| Perplexity | 9,798 | 2,903 | 29.6% |
| Gemini | 9,518 | 2,154 | 22.6% |
| Google AI Overviews | 6,336 | 1,040 | 16.4% |
| ChatGPT | 6,108 | 615 | 10.1% |
| All four | 31,760 | 6,712 | 21.1% |
Every brand was asked its own set of buyer questions, and every question went to all four engines. A brand counts as named when the answer mentions it by name. Links alone do not count.
The figures are pooled: we add up answers across all 10 brands rather than averaging brand by brand. That is the only way we publish client figures, and no brand or sector is shown on its own.
The spread is the headline. The brand a buyer hears about depends on which engine they happen to open.
What to Do With the Engine Gap
If the engines disagree this much on the same questions, a brand needs to read, report and work on them separately. Five moves follow from the table, in the order we would make them.
1. Report each engine on its own
A single AI visibility score averages a 29.6% engine with a 10.1% engine and reports something that describes neither. Split every report by engine. When the blended number moves, you need to know which engine moved it.
Take a Head of Growth, call her Priya, who sees her brand's visibility rise from 14% to 18% in a month. Split by engine, all of that rise is Perplexity, and ChatGPT is flat at 2%. The board deck said "AI visibility up". The honest line is "Perplexity up, ChatGPT untouched", and it leads to a different next quarter. (Share of voice in AI search works the same way: split it before you read it.)
2. Check ChatGPT first, because it is usually the gap
ChatGPT was the lowest engine for 7 of our 10 brands. For most brands that makes it the biggest gap and the one to look at first. Run your own buyer questions in ChatGPT, not your brand name, and note which competitors it names instead. Our ChatGPT tracker runs those questions on a schedule so one bad afternoon does not set your strategy.
If ChatGPT names your competitors and not you, the causes are usually specific and fixable. We walked through them in why a SaaS brand does not show up in ChatGPT.
3. Use Perplexity to see which pages carry your category
Perplexity names brands most often, and it shows its sources with the answer. That makes it the cheapest place to see which pages an engine leans on for your questions: review sites, comparison lists, forums, your competitors' own pages. Those pages are your target list.
What one engine cites is not proof of what another one reads. We have not measured that overlap, so treat Perplexity's sources as a starting list, not as the answer for ChatGPT. The engine-by-engine differences are laid out in our ChatGPT vs Perplexity vs Claude guide.
4. Only compare like with like
A rate means something only next to a rate measured the same way: same questions, same engines, same model, same window. Change any one of those and the comparison breaks.
We learned this on our own numbers. In August our visibility read 6.71%. In October it read 3.2%. Nothing in the market had halved: the ChatGPT model we track with changed in September, and the competitor lists widened. The two numbers are not comparable, and we do not compare them. Build the question set once and keep it stable, so a change in the number is a change in the market.
5. Count failed runs out, not as zeros
An engine that fails to answer is having an outage. Counted as an answer, it reads as a drop in visibility. Our Google AI Overviews capture returned almost nothing from 5 October onward, and those runs are left out of every figure here rather than counted as answers that named nobody.
Want this split for your own brand? Run the free AI visibility checker, or book an audit call and we will walk you through all four engines on your own buyer questions.
Our Own Numbers: RankZero by Engine
We are one of the 10 brands, and we are near the bottom. In the same 30 days, on 31 tracked questions that do not mention RankZero:
| Engine | Answers | Named RankZero | Rate |
|---|---|---|---|
| Perplexity | 900 | 50 | 5.6% |
| Google AI Overviews | 600 | 9 | 1.5% |
| Gemini | 870 | 11 | 1.3% |
| ChatGPT | 559 | 0 | 0.0% |
Zero is not a rounding problem. 559 times, someone asking the sort of question our buyers ask got an answer from ChatGPT that did not include us.
We are publishing it because it is the honest starting line. An agency that sells AI visibility work should show its own baseline before it shows anyone else's. The next editions will show whether the work we are doing on our own site moves it, engine by engine.
The shape matches the pooled table. Perplexity is our best engine and ChatGPT our worst, the same order as the 10-brand totals, only lower.
Why the Engines Differ
We have measured how often each engine names a brand. We have not measured why. What follows is our reading, not a finding.
- Perplexity searches the web for nearly every answer and cites what it finds. Brand names sit on the pages it retrieves (lists, reviews, comparisons), so they land in the answer.
- Google AI Overviews is built on Google Search results, so it inherits much of what already ranks for a question.
- ChatGPT often answers in general terms. On a broad buyer question it will explain how to choose before it names anyone, and some answers never get to a name at all.
The practical point does not depend on which reading is right. A brand's presence in AI search is four different numbers, and the work that lifts one does not automatically lift the others. That is why our AI search visibility tool reports by engine first, and how a citation becomes a mention is covered in our AI citation entry.
Why Another Study Says ChatGPT Names Brands Most
At least one published study ranks ChatGPT first. Sourceable's analysis of 240 brands across 1,800 category prompts (March to May 2026) put ChatGPT at 41%, Gemini 36%, Perplexity 33% and Claude 29%.
That does not make either study wrong. It shows how much the result depends on the method:
- What counts as "relevant". A rate over answers judged relevant to a brand is a different number from a rate over every answer to a fixed question set.
- Whose questions. A broad category panel measures the market. Each brand's own buyer questions measure that brand's position.
- Which model and when. Engines change the models behind them, and a model change moves the rate on its own. Ours did in September.
Read any engine comparison, ours included, as true for its method and its window. Then measure your own brand the same way twice before you believe a change.
How We Measured
Everything above comes from RankZero's own tracking of the brands we run, the same first-party approach as our AI crawler study.
- Window: the 30 days to 9 October 2026.
- Brands: 10 brands we run, including our own. Client figures are pooled across all 10, with no brand or sector split.
- Answers: 31,760 answers to tracked buyer questions. Each brand has its own question set; every question goes to all four engines.
- Engines and models: ChatGPT through OpenAI (gpt-5.6-luna and gpt-5-mini logged), Perplexity (sonar), Gemini (gemini-2.5-flash), Google AI Overviews captured from Google's results page.
- Schedule: Perplexity, Gemini and AI Overviews daily. ChatGPT runs each question every other day since 14 September (half the set one day, the other half the next), which is why it has fewer answers.
- Named: the answer mentions the brand by name. A citation link without the name does not count.
- Failed runs: an engine run with no answer text is a failure, not an answer, and is left out.
- Known gap: the AI Overviews capture returned almost no answers from 5 October, so its figure rests mostly on 9 September to 4 October.
- Not comparable with earlier figures: the ChatGPT tracking model changed around 7 September and competitor lists widened in September. This edition is the baseline.
- Limits: 10 brands is a small group. These are the brands we run, not a sample of the market, and the figures say how these engines treated these brands in these 30 days.
The Next Edition
This report updates every quarter, with the same method so each edition can be read against the last. Each one keeps the four-engine table on this page, dated, and adds a new window beside it.
We also plan to add which kinds of sources each engine cites for these questions, and what moved after a dated piece of work on a brand. When a number changes because we changed how we measure, we will say so next to the number.
Frequently Asked Questions
How often does ChatGPT mention brands in its answers?
Across 10 brands and 6,108 ChatGPT answers in the 30 days to 9 October 2026, ChatGPT named the brand in 10.1% of answers. It was the lowest of the four engines we track for 7 of the 10 brands.
Which AI engine names brands most often?
In our data, Perplexity: 29.6% of 9,798 answers named the brand, against 22.6% for Gemini, 16.4% for Google AI Overviews and 10.1% for ChatGPT. Sourceable's study, measured differently, ranks the engines in a different order.
Why does my brand show up in Perplexity but not in ChatGPT?
It is the pattern for every brand in our data: Perplexity named all 10 brands more often than ChatGPT did. Perplexity searches and cites the web for almost every answer, while ChatGPT often answers in general terms. Treat ChatGPT as its own piece of work and track it separately.
Can I compare my AI visibility score from one month to the next?
Only if the questions, engines, models and window are measured the same way both times. A model change alone can move the number. Ours read 6.71% in August and 3.2% in October, after the tracking model and the competitor lists changed, so we do not compare the two.
Your brand is four numbers, not one. If you want to know which engine is the gap for you and what would close it, book an AI search audit call. We run the work as a done-for-you AI SEO agency, and we will show you your own four-engine split on the call.