ASSmetric by LLM Jesus

What’s your ASS score?

Authority, Sources, Specificity. The three things ChatGPT checks before it recommends a business. We record it doing exactly that for the question your customers ask, then score you on what it found.

  • Recorded, not asked. The score comes from the network traffic behind a real ChatGPT answer: every search it wrote, every page it opened, every business record it read.
  • Not an API. Put the same questions through the API routes most AI-visibility tools use and they agree with the real thing on 1.3% and 1.8% of sources. So we don’t use them.
  • Re-measured, not re-argued. When the model changes, the score is taken again on the new build. Every score names the recording and the build it came from.

Get your ASS score

Free. One real ChatGPT answer, recorded for your question.

That didn’t send. Email support@llmjesus.com with your website and the question, and we’ll run it from there.

Recording takes a real ChatGPT session and is run by a person, so the score usually arrives within two working days. Nothing is sold to anyone. How it’s scored.

Queued for recording.

We’ll ask ChatGPT , record what it does, and send the score to .

While you wait, the public recordings behind the score are at llmjesus.com.

The three pillars

Where a recommended business actually comes from

We traced every recommended business in 82 recorded answers back to whatever first put it there. Each pillar of the score is one of those routes, measured for you on your own question.

AAuthority

Was ChatGPT already carrying you before it searched?

The model writes brand names into its own searches before a single result comes back. If you are one of them, you were in its head already. Scored from whether you are named in the answer, and whether it searched for you by name.

29.8%of recommended businesses were already in the model before it searched82 recorded answers
SSources

Are you on what it reads?

Local business records, the pages it opens, the directories and review platforms it names. Scored only on the channels it actually used for your question: you are not marked down for records it never pulled.

37.0%came out of local business records, and 71.3% of everything it read was a page the recommended brands do not own82 recorded answers
SSpecificity

Does your own site answer what it searched for?

ChatGPT adds words to its searches that your question never had: a year, a price, “official”, a licence, reviews, hours. We read your site for each one it asked for. Scored as answered over asked.

29.0%of its searches check the year, 15.6% ask for prices, 11.8% hunt for the official page186 recorded searches

The number

One score, three parts, and the recording behind it

Each pillar is 0 to 100. The score is the average of the pillars the recording could measure. A pillar the recording could not measure is left blank, never counted as zero.

70–100
In the runningNamed, or on what it reads, and your site answers what it searches for.
35–69
Known, not chosenChatGPT saw you somewhere and recommended someone else. The most fixable place to be.
0–34
Not in the pictureNot in the answer, not in the records, not on any page it opened.

What you get

The score, and everything it rests on

Your ASS scoreThe overall number, the band, and each pillar with what was found.
The traceEvery search ChatGPT wrote, every page it opened, every business record it read, verbatim.
Who it chose insteadThe businesses it recommended, and which route each one came in by.
What to change firstThe Specificity checks you failed, named, with the page we looked at.

The lineage

First you E-E-A-T. Then you ASS.

Google’s quality guidelines gave the web a vocabulary for what a page should be. Answer engines changed the question to what the model does. The ASS score is the next step in that line: same job, different instrument.

2014
E-A-TExpertise, Authoritativeness, Trustworthiness. Introduced in Google’s Search Quality Rater Guidelines. Qualities human raters look for.
2022
E-E-A-TExperience added, December 2022. Still a rater’s checklist: describes the page, does not measure the engine.
2026
ASSAuthority, Sources, Specificity. Three measurements taken from a recording of ChatGPT working on your question. Ours, not Google’s, and re-measured when the model changes.

Nobody can measure “trustworthiness” on a Tuesday. You can measure whether the model searched for you by name, whether you were on the pages it opened, and whether your site answered what it asked. Each pillar maps to a route we watched businesses take into an answer, with the share next to it.

The routes are published, with sample sizes, at llmjesus.com, and the scoring rules are on the methodology page. E-E-A-T is Google’s. ASS is not. ASSmetric is not affiliated with OpenAI or Google.

How we measure, and what would make it wrong