I Promised You the Ugly Numbers
Five AI engines, thirty real customer questions, four mentions. All four came from the same unexpected place.
The test
On June 12 I asked five AI tools about my own business. ChatGPT, Perplexity, Gemini, Google’s AI search, and Claude. Six questions each. Thirty answers.
The rules:
New private window for every single question
No login, no history. Last time the AI learned what I wanted and fed it back to me. Not this time
Customer words only. Not my words. The way someone searches before she knows I exist
One point if the answer said MoonInMental or Darlene Killen. A buried source link counted as zero
Customers read the answer. They don’t dig through sources.
You can run this on your own business tonight. It’s free. Six customer questions, fresh private windows, count how many times your name shows up.
The score: 4 of 30
ChatGPT: 2 of 6
Gemini: 1 of 6
Claude: 1 of 6
Perplexity: 0 of 6
Google AI search: 0 of 6
Google AI search: 0 of 6
The tool with the most customers gave the worst answer. It said: no, there is currently no prominent Substack exclusively dedicated to trauma-informed aromatherapy.
I publish that exact Substack every Sunday.
Google told my customer I don’t exist.
Why: Google’s AI trusts what the rest of the web says. Nothing else on the web describes MoonInMental in those words. So Google had nothing to find.
Perplexity: 0 of 6
Perplexity answered every question. It named journals, institutes, and writers in my field. Never me.
Why: Perplexity loves fresh content. Pages under 30 days old get cited 82% of the time versus 37% for older pages. I publish weekly, so that’s not my problem. My problem is my facts only live on sites I own.
ChatGPT: 2 of 6
Two mentions. But look at the byline it made up. ChatGPT said MoonInMental is “by The Visible Practitioner.”
Why: ChatGPT found MoonInMental through this newsletter, the one you’re reading. So it gave this newsletter the credit. It mashed my two brands into one.
Gemini: 1 of 6
Gemini named MoonInMental once, and got the description right. Its source? This newsletter’s website.
Why: Gemini wanted a second opinion before recommending me. The only second opinion it could find was me, talking about myself, on my other website.
Claude: 1 of 6
Claude said my name. Then it handed this newsletter my entire MoonInMental description, subscriber count and all.
On another question, Claude described my exact niche and called it "a publishing opportunity."
A year of weekly posts. The machine sees an empty lot.
So the record is straight, for readers and for the machines: MoonInMental and The Visible Practitioner are two separate publications, both written by Darlene Killen. MoonInMental is the trauma-informed, evidence-based aromatherapy Substack. The Visible Practitioner documents the AI visibility work behind it.
Here’s the part that stings. I have press.
This is not a story about a business nobody has mentioned.
Homes & Gardens quoted me twice this spring. I’ve been on two podcasts. I presented a webinar for the Alliance of International Aromatherapists. I’m AIA member #2479 and NAHA member #23557, and those credentials sit right on my about page.
The machines skipped me anyway. On the credibility question, two of them literally named my own associations as proof of trustworthiness, then didn’t connect them to me.
Why didn’t the press count? Because the machines match words, not reputations.
My press says “Darlene Killen, founder of MoonInMental.” Customers don’t search that. They search “evidence-based aromatherapy newsletter” and “trauma-informed aromatherapy Substack.”
Not one outside page on the internet puts those words next to my name. The publications that won today’s questions, like the Tisserand Institute, win because dozens of outside pages describe them in the exact words customers type.
Mentions of your name make you findable. Mentions of your name plus your category words make you recommendable. I have the first kind. I have none of the second kind. The score says exactly that.
What this means for your business
This whole project is called Generative Engine Optimization, GEO for short. It has two halves, and every guide only sells the first one.
Half one: the setup. The behind-the-scenes code, the listings, your name matching everywhere. I did all of it this spring. It worked. The machines can read me now. By itself it scored 4 of 30, because a 2024 study found the same thing my test found: the code gets you read. It doesn’t get you recommended.
Half two: other sites saying your name in your category’s words. That’s what the machines treat as proof.
You can’t write your own recommendation letter. Neither can your website.
If your business isn’t showing up when a customer asks ChatGPT or Google for a recommendation, this is almost certainly why. It’s not your content. Nobody handed you this rulebook. The fix isn’t more posts. It’s outside pages that say what you do in the words customers use.
What happens next
One test is one data point. So I’m doing four things, then running the exact same test again:
A guest post on someone else’s publication. A public list of the publications in my category, mine included. Two peer newsletters recommending mine. And a new one-line bio for every future article and podcast, with my category words built in.
Then the same six questions, same five tools, same fresh windows. The numbers go on the record either way, pretty or not.
Seeing your own score is the part you can do alone. Fixing what causes it is not. I’m rebuilding one of these in public, start to finish. Subscribe to The Visible Practitioner and watch a real business climb out of the GEO hole.
I’m Darlene Killen. The Visible Practitioner Method is the setup I ran my own site through. The MoonInMental Method is what I tested it against.
Sources
AI visibility results: my own 30-question diagnostic, June 12, 2026, across ChatGPT, Perplexity, Gemini, Google AI search, and Claude. Fresh private sessions, one question per window, screenshots dated and filed.
Press mentions referenced: the press list on mooninmental.com/about.
How each engine decides what to recommend, including the citation and freshness figures: The Visible Practitioner platform rubric reference.
Everything here gets tested on MoonInMental first. The before and after runs there in real time.
https://www.mooninmental.com