AI visibility: how often AI assistants name your brand when buyers ask without naming any brand.
How we measure AI visibility
We ask each AI assistant the same buyer questions every month and count the answers. Every answer is recorded and checked. Results are shown as counts first, with the percentage beside the count, and a refusal is shown and never counted as an answer.
Sample monthly report: Brand A
| Assistant | Brand A named in |
|---|---|
| ChatGPT | 4 of 10 answers (40%) |
| Google Gemini | 2 of 10 answers (20%) |
| Google AI Overviews | 3 of 10 answers (30%) |
| Claude | 0 of 10 answers (0%) |
| Perplexity | 1 of 10 answers (10%) |
What we count
AI recommendation: how often they suggest your brand as the one to buy.
Share of voice: your brand's mentions as a share of all brand mentions in those answers.
Accuracy: whether what they say about your brand is correct, checked on questions that name the brand.
For a pharma company, the questions are the ones its business buyers and the public ask, and AI recommendation counts how often an assistant suggests the company as the one to contact. The four measures stay the same.
Answer log
Question: Which antiseptic liquid should I keep at home?
| Assistant | Answer | Brand A named |
|---|---|---|
| ChatGPT | Brand A and Brand B are often named for this need. | Yes |
| Google Gemini | I cannot suggest a brand for this. | Refusal: not counted |
| Perplexity | Brand B is often named for this need. | No |
How we keep the numbers honest
Every answer is recorded and checked.
Every answer is read and counted by a person, and the full answer log is kept, so any count can be checked.
Results are shown as counts first, such as 'named in 0 of 29 answers', with the percentage beside the count, never a percentage alone.
Refusals are recorded and shown, never counted as answers.
Why answers vary
Answers shift for three reasons.
AI companies change their models and how they pick sources.
Web pages themselves change.
Each run can also read different pages.
That is why the same questions are re-tested monthly.
Monitoring only
Monitoring only is the monthly measurement and report on their own. The same buyer questions go to the five AI assistants every month, every answer is read and counted by a person, and you see the change. There is no website or content work.
Questions about measurement
Why can ChatGPT name different brands each time I ask?
AI assistants build each answer fresh, so the brands they name can change from one run to the next. In a January 2026 study, SparkToro and Gumshoe ran 2,961 prompts across ChatGPT, Claude and Google's AI and found the same brand list repeated in fewer than 1 in 100 runs for ChatGPT and Google's AI. Claude was more consistent. So we count how often your brand is named, and re-test the same questions monthly.
Can you promise AI assistants will name our brand?
No one can promise what an AI assistant will say. Each AI company controls what its assistant says, and it can change that at any time. GetPreferred™ works on what makes AI assistants more likely to name a brand. That means a website they can read, facts they can check, and trusted sources that repeat them. We then report what changes, month by month.
Find out which brands AI assistants name in your category.
A thirty-minute call with the founder, then a written baseline: the buyer questions for your category put to five AI assistants, and which brands they name. Free, with no obligation.
