The short answer
Share of answer is the percentage of AI answers, for a defined set of buyer questions, that name your company. Because AI engines rarely give the same answer twice, measure it across many questions and repeated runs, per engine, every month. Frequency of appearance is the signal. Position in a single answer is mostly noise.
Key takeaways
- Traditional rank tracking is ineffective for AI search because engine responses vary significantly across multiple runs.
- Share of answer replaces ranking with frequency, measuring how often your brand is mentioned across a consistent set of buyer prompts.
- Tracking brand mentions, cited sources, and descriptive accuracy across multiple AI engines provides actionable insights into your visibility.
Sooner or later, someone asks: "What rank are we in ChatGPT?" It's a fair question with a frustrating answer. There isn't one.
Ask the same question twice and you may get a different list. That's not a bug in your tracking. It's how these engines work.
Why doesn't rank tracking work in AI search?
Because the lists keep changing. In a January 2026 study by SparkToro and Gumshoe.ai, about 600 volunteers ran 12 identical prompts through ChatGPT, Claude, and Google's AI nearly 3,000 times.1
Less than 1 in 100: the odds of an AI engine giving the same list of brands twice for the same prompt. Source: SparkToro and Gumshoe.ai, via Search Engine Land, January 2026 1
The order was even less stable: an identical list in the same order came up closer to 1 in 1,000 times. But some brands appeared in 60% to 90% of answers for a given intent, even as their position shifted.1 That's the signal worth tracking.
What is share of answer?
It's a frequency metric, borrowed from the idea of share of voice.
\text{Share of answer} = \frac{\text{AI answers that name you}}{\text{total AI answers checked}} \times 100If you check 200 answers across your priority questions this month, and 46 name you, your share of answer is 23%. Track it per engine, per question group, and alongside your top competitors.
How do you measure it?
Choose 20 to 50 buyer questions.
Use the questions prospects ask on sales calls, grouped by intent: category, comparison, problem, and use case.
Pick your engines.
ChatGPT, Perplexity, Gemini, Claude, and Google AI Mode cover most B2B buyers.
Run each question several times.
The SparkToro study's main lesson is that one run tells you little.1
Log every answer.
Use the table below.
Calculate share of answer
for you and your top three competitors.
Repeat monthly,
with the same questions, so the trend means something.
| What to log | Why |
|---|---|
| Question and engine | Lets you compare engines and intents |
| Date | Answers shift as engines update |
| Brands named | The core of the metric |
| Sources cited | Shows which sites shape the answer |
| How you're described | Catches outdated or wrong descriptions |
Small samples swing wildly. A single run where you appear doesn't mean you've won the query, and a single run where you don't isn't a crisis. Judge trends across months, not individual answers.
The "sources cited" column is the most useful one. It tells you exactly which review sites, publications, and pages to influence next.
You can't rank in an answer. You can show up in more of them. Share of answer is how you prove it.
Read next: How to track AI search traffic in GA4 and Search Console
The bottom line
You can't rank in an answer. You can show up in more of them. Share of answer is how you prove it.
Sources
- Search Engine Land, AI recommendation lists repeat less than 1% of the time: Study, Danny Goodwin, January 28, 2026. Reports research by Rand Fishkin (SparkToro) and Patrick O'Donnell (Gumshoe.ai). searchengineland.com/ai-recommendation-lists-rarely-repeat-study-468076
About the author
Aimee van der Haar
Aimee founded Bearly Creative to give great startups and tech companies the marketing they deserve. She works across strategy, design, content, and AI search, and she still does the work herself.
