GEO & AEOCase study
AI Visibility Baseline: 0 Mentions in 2,034 Non-Branded Answers
By Vamshi Krishna, FounderPublished
In October 2026 we tracked Cognova AI in six AI visibility tools and reviewed 2,532 AI answers. Across the 2,034 answers to questions that did not contain our name, AI never genuinely recommended us. Two tools still showed headline visibility above 15%, because those numbers included prompts with our brand name in them. This page is our baseline; we will update it after we fix our site and re-measure.
- Client
- Cognova AI (our own brand)
- Industry
- AI consulting and development
- Size
- New brand, website relaunched October 2026
Results
2,532
AI answers reviewed across six tools
0 of 2,034
Genuine mentions of Cognova AI in non-branded answers
20.7%
Peec AI's default headline visibility, all from branded prompts
2 of 6
Tools whose automatic setup measured a market we do not sell in
The challenge
We had just relaunched our website and wanted to know a simple thing: when buyers ask ChatGPT, Google's AI answers or Gemini about AI voice agents and AI consulting, does AI mention us? We also wanted to know whether the AI visibility tools our clients pay for would answer that question honestly when set up the way most buyers set them up, with the default onboarding.
A new brand is the hardest test for these tools. The true answer was likely close to zero, so any tool that reported a healthy score had to be counting something other than real recommendations.
What we built
We signed up for six tools on their standard trials: Peec AI, Scrunch, Analyze, Writesonic, Rankscale and SE Ranking. In each one we accepted the suggested prompts and competitors, let it collect answers, and then compared its dashboard with the AI answers underneath. We read the answers through each tool's MCP connection where the trial allowed it, and from the dashboard screens where it did not.
For every tool we split the prompts into branded (the question contains our name or domain) and non-branded (it does not), counted how often AI named us in each group, and checked the competitor list against the brands the answers actually named. Peec AI ran for six days; the other five tools completed one collection run each.
“We expected AI to ignore a brand this new. What surprised us was how many dashboards told us it did not. That gap is exactly what our clients need to see before they spend money on fixes.”
How we ran the test
We tracked Cognova AI between October 22 and October 28, 2026. Each tool used its own default onboarding: the prompts, engines and competitors it suggested, with no edits from us. We then read the stored AI answers and counted mentions ourselves, separately for branded and non-branded prompts.
Results by tool
| Tool | Engines | Prompts (branded) | Answers reviewed | Default headline | Non-branded mentions | Setup issue we found |
|---|---|---|---|---|---|---|
| Peec AI | ChatGPT, AI Overviews, Gemini | 96 (20) | 2,304 over 6 days | 20.7% visibility, 97% share of voice | 0 of 1,821 | Tracked 7 enterprise AI firms, not the voice-agent rivals AI named |
| Scrunch | ChatGPT, Perplexity, AI Overviews, Copilot | 5 (0) | 20 | 0% | 0 of 20 | Competitor list missed Synthflow, the rival named most |
| Analyze | ChatGPT, Perplexity, Google AI Mode | 15 (0) | 90 | 1% | 1 of 90, not genuine | 5 prompts named three rivals; our one mention called us ‘the user organization’ |
| Writesonic | ChatGPT, Gemini, AI Overviews | 30 (5) | 90 | 15.56% on the home widget | 0 of 75 | Prompts drifted into market-research software |
| Rankscale | ChatGPT, Google AI Mode, Copilot | 6 (0) | 18 | 0% | 0 of 18 | 2 prompts named a rival; one company split across several names |
| SE Ranking | ChatGPT | 10 (0) | 10 | 0% | 0 of 10 | Prompts about analytics tools; rivals added under the wrong domains |
Analyze’s single mention came from an answer that described Cognova as ‘the user organization’ evaluating the platforms, a sign our own account setup reached the AI with the question. We do not count it as a genuine recommendation.
Limits of this baseline
This is one brand, measured on trial accounts, mostly with a single run per tool. It shows how each tool behaves with default settings, not how accurate it can be after careful setup. Prompt sets differed between tools, so the per-tool numbers are not directly comparable with each other.
What happens next
We are fixing the issues this baseline exposed on our own site, starting with the technical changes in 7 Technical Fixes That Help AI Search Engines Cite Your Website. We will re-measure the same non-branded prompts, compare them with prompts we did not work on, and update this page with the results.
How Cognova AI can help
We run this same baseline for clients before any GEO work starts. We audit the prompts, competitors and brand-name matching in whatever AI visibility tool you use, measure what AI actually says on the questions your buyers ask, and split the result into branded and non-branded so you know the real starting point.
Then our GEO & AEO Consulting team fixes the causes and re-measures the same prompts, alongside prompts we left untouched, so you can see whether the work moved the number. Book a call if you want your own baseline.
Frequently asked questions
What is an AI visibility baseline?
An AI visibility baseline is a record of how often AI assistants mention your brand before you change anything. It should cover the questions buyers ask without your name in them, list the engines and dates, and note which competitors AI names instead. Without an AI visibility baseline, you cannot tell whether later improvements came from your work or from normal day-to-day variation in AI answers.
Why did AI visibility tools show a score above zero if AI never recommended us?
The AI visibility tools that showed a score above zero included branded prompts, questions that already contained our name or domain. AI almost always mentions a brand the question names. In Peec AI, those branded prompts produced the whole 20.7% headline. In Writesonic, they produced a home page figure above 15%. Filtered to non-branded prompts, every tool showed zero genuine mentions.
How long should an AI visibility baseline run?
Run an AI visibility baseline for at least a week of daily checks before you change your site or content. AI answers vary from run to run, so a single day can mislead in either direction. Our Peec AI baseline covered six days; the other tools had one run each, which is enough to spot setup problems but not enough to measure a trend.
Can I trust the default settings in an AI visibility tool?
Check them first. In our baseline, the default settings in AI visibility tools caused most of the errors: branded prompts lifted headline scores, competitor lists missed the rivals AI named most, and two tools generated prompts for markets we do not sell in. An hour spent reviewing prompts, competitors and brand-name matching makes any AI visibility tool far more reliable.
Sources
Cite this article
Vamshi Krishna. (2026, October 29). AI Visibility Baseline: 0 Mentions in 2,034 Non-Branded Answers. Cognova AI. https://cognovaai.com/insights/ai-visibility-baseline-0-mentions-in-2-034-non-branded-answers
Permanent link: cognovaai.com/insights/ai-visibility-baseline-0-mentions-in-2-034-non-branded-answers. Quoting a passage? Select it and use “Copy quote” to get a link that jumps to that exact text.