The free AI search audit nobody is running
The 8-step exercise that tells you more than most AEO dashboards.
👋 Hey, I’m George Chasiotis. Welcome to GrowthWaves, your weekly dose of B2B growth insights—featuring powerful case studies, emerging trends, and unconventional strategies you won’t find anywhere else.
This note is brought to you by Minuttia.
According to our survey of 599 marketers:
Over 80% of organizations are actively addressing AI search.
And with over half of respondents expecting their AEO budget to rise in the next 12 months, if you’re not investing in AI search, you risk falling behind.
That’s where Minuttia comes in.
Minuttia helps B2B companies dominate search through a suite of human-first, high-touch, value-driven services:
Google and AI Search Strategy
Content Audits
Human Content Creation
Digital PR
If you want to win in AI search before it becomes table stakes, request an intro call with us.
We’ve already helped 50+ B2B SaaS companies. You could be next.
I posted something on LinkedIn last week that I did not expect to get the reaction it did.
An 8-step exercise. Importantly, no tools are required!
Just ChatGPT in incognito, a simple question, and a spreadsheet.
The post got more saves and DMs than anything I have shared in weeks. (Surprisingly, it wasn’t as successful in terms of engagements and impressions, but that’s ok.)
And those DMs told me something I had suspected but never confirmed: most companies have never checked what AI search engines actually say about them.
They track visibility. They measure citations. They pay for dashboards.
But they have never opened ChatGPT, typed “What is [their brand]?” and read what comes back.
So, let’s see what the post was about and how you can use what I’m about to share to identify perception gaps for your brand.
The tracking tools gold rush
We published a report on AI search tracking tools back in December 2025.
At the time, we found the category had already become a commodity. Hundreds of tools, most doing some version of the same thing: track your brand’s mentions across AI search engines.
Since then, the number has grown. By my last count, at least a few hundred dedicated solutions exist in this space. Every week, I see a new one launch on LinkedIn or drop into my inbox.
I get it. Measurement matters. In the AEO survey Kevin Indig and Minuttia ran with 599 respondents, 40.6% said their single biggest challenge is a lack of reliable measurement tools and attribution.
They are right. Measurement is a problem.
But not because we lack tools! But because the tools measure the wrong thing.
The variance problem
Here is where it gets uncomfortable.
SparkToro ran a study with 600 volunteers who ran 12 prompts across ChatGPT, Google AI Overviews, and other AI engines. A combined 2,961 runs.
The finding: ChatGPT returned the same brand list less than 1% of the time for identical prompts. The same list in the same order? Less than 0.1%.
Nearly every response was different. Different brands, different order.
If you are tracking AI search visibility using point-in-time snapshots, you are measuring noise. There is signal buried in the variance, but most tools present the data as if it were stable and reliable.
It is neither.
Author’s Note: I am digging into the mechanics of why prompt tracking breaks down in next week’s GrowthWaves note.
Visibility is not the same as perception
I wrote about this in my note on Perception Deviation.
Your brand has a ground truth. How it creates value, who it serves, what it stands for.
AI search engines form their own perception of your brand based on the content they retrieve from the web. That formed perception may or may not match your ground truth.
The gap between the two is what I call Perception Deviation.
Tracking visibility tells you whether AI mentions your brand. Tracking perception tells you whether AI understands your brand correctly.
The second question is harder to answer. It is also far more important.
If ChatGPT recommends your product to the wrong audience, that is not a visibility win. If Perplexity describes you as something you stopped being two years ago, citations are meaningless. If Gemini gets your pricing model wrong, every user who reads that answer starts the relationship with a misconception.
Visibility without accuracy is a liability.
An example: Meilisearch
I want to make this concrete.
Minuttia has the privilege of working with Meilisearch for a while, so I have an informed view of both their ground truth and their formed perception. (Full disclosure: they are a Minuttia client.)
Ask any major AI search engine “What is Meilisearch?” and you will almost certainly get some version of this answer: Meilisearch is an open-source search engine.
That is not wrong. Meilisearch started as an open-source project. The GitHub repository has over 58,000 stars.
The open-source roots are real.
But the answer is incomplete.
Meilisearch isn’t just an open-source search engine.
Their enterprise customers include Hugging Face and Louis Vuitton.
The product includes AI-powered hybrid search and a managed cloud platform. It has moved well beyond its open-source origins.
Calling Meilisearch “an open-source search engine” in 2026 is like calling Slack “an IRC replacement.” Technically rooted in truth. Missing everything that happened since.
That is perception deviation in action.
And here is the thing:
AI search engines did not invent this perception. They (somewhat) inherited it.
From every blog post, every listicle, every Stack Overflow answer, every GitHub discussion that described Meilisearch using those three words.
The AI reflects what the internet says. If the internet says “open-source search engine” in 200 places and “AI-powered enterprise search platform” in 20, the AI will synthesize accordingly.
You cannot fix the AI’s answer by optimizing for the AI. You can fix that by finding the sources and changing what they say.
The exercise
Here is the process I shared on LinkedIn:
Open ChatGPT in incognito. No login, so personalization does not skew results.
Ask a brand question. Start with “What is [your brand]?”
Run it 50 times. Each in a new chat to capture response variance.
Log every answer and every source it cites.
Rank the sources by how often they appear.
Flag every gap between what AI says and what you know to be true.
Go to the top-cited sources. Whether you own them or not.
Re-run monthly. Track whether the gaps are closing.
This will not give you everything a proper tracking tool would. But it costs nothing, and it shows you where you actually stand.
Most companies have not even done this much.
Questions worth asking
“What is [brand]?” is the starting point.
But you can and should go further.
Here are some questions that surface different types of perception deviation:
Run each one 10 to 20 times. Log the answers.
Compare them to your ground truth, and you’ll start noticing certain patterns.
Where the perception lives
Once you have identified the gaps, the next question is: where do you go to close them?
I outlined part of the framework in my note on Perception Deviation, but the short version is that AI perception gets formed across four layers:
Brand experience
Owned perception
Multi-channel perception
Brand mentions
Start with what you control. That could include:
Your website
Your product messaging
Your social profiles and directory listings
If your homepage still leads with language that no longer reflects your current positioning, AI search engines will anchor on what your own site says.
Which means you need to take action…
G2 reviews. Reddit threads. YouTube videos. Community forums and conference talks.
These are the sources AI engines cite most frequently. They are also the ones most companies ignore when thinking about AI search optimization.
None of this is AI search optimization in a narrow sense. It is brand management.
The AI is the mirror. You fix the reflection by fixing the source.
Final Thoughts
Hundreds of AI search tracking tools exist. Some are good. Most is redundant.
But before you invest in any of them, do this simple exercise.
Open ChatGPT and ask what it knows about your brand.
Read the answer carefully. Compare it to what you know to be true.
If the answer is wrong, a dashboard will not fix it. The source material will.
The best thing you can do for your AI search perception right now costs nothing and takes an afternoon.
Be one of the companies that has actually done it.
Sorry for missing last week’s note! I had a heavy travel schedule, and there were not enough hours for me to sit down and write something for you.
Research Disclaimers and Limitations
GrowthWaves and its author are not sponsored by or compensated by any company mentioned in this note. This is independent editorial analysis and does not constitute investment, financial, or legal advice. The author may have relationships with, work with, or hold equity in companies referenced; however, no content in this piece was influenced, commissioned, or incentivized by any such relationship. AI tools were used as a research assistant in the preparation of this piece. All claims are sourced and linked throughout.
Sources
SparkToro, “NEW Research: AIs are highly inconsistent when recommending brands or products”
GrowthWaves x Growth Memo, “The State of AEO: What 599 Marketers Told Us About AI Search”
GrowthWaves, “AI Search Tracking Tools Report”
GrowthWaves, “Perception Deviation: The most important metric you’re not tracking”
GitHub, Meilisearch repository