GEO Lab #03 · S3-CHATGPT-V1-R2 · Original response

This page presents one selected response from GEO Lab #03. It adds a smaller reading and citation entry point; it does not replace the full archive or collect a new answer.

The OpenAI arm uses the Responses API with web search and does not represent consumer ChatGPT Search. Grok records identify their actual consumer-interface mode and available model label.

Recorded conditions

Record ID
S3-CHATGPT-V1-R2
Platform
OpenAI API + web search
Question / condition
V1
Repeat
2
Model label
gpt-5.6-sol
Visible mode
Not independently recorded
Reasoning
medium
Submitted / first recorded time
2026-11-02T09:30:13.472155+00:00
Completeness
Recorded body complete
Search
Actual search activity observed; exact controls and evidence remain in the raw record.

Original prompt

I'm a solo SaaS founder with no GEO budget. How can I tell whether an AI visibility score for ChatGPT and Grok is trustworthy?

The agreed prompt is reproduced exactly. Any recorded difference in the actual submitted text or search UI is retained in the raw record and full appendix.

Original answer

**Short answer:** don’t trust a single “AI visibility score.” Trust the **underlying observations** only if they’re transparent, repeatable, and relevant to real buyer prompts.

ChatGPT and Grok can produce different answers depending on whether web search is active, prompt wording, location, conversation context, and model version. ChatGPT may rewrite searches and use approximate location; OpenAI also warns that search results and citations can be incomplete, outdated, or incorrect. Grok’s API does not access real-time information unless Web Search or X Search is enabled. ([help.openai.com](https://help.openai.com/en/articles/9237897-chatgpt-search))

## Ask these seven questions

A trustworthy vendor should answer all seven:

1. **Which exact product and model were tested?**  
   “ChatGPT” isn’t enough. Ask whether it was logged-in ChatGPT, Search mode, API, or another mode—and which Grok model/interface.

2. **Was live search enabled?**  
   Search-enabled and training-knowledge answers are fundamentally different tests. Citations should be captured when available. ([help.openai.com](https://help.openai.com/en/articles/9237897-searching-the-web-with-chatgpt))

3. **Can I see the actual prompts and full responses?**  
   Reject black-box scores that provide only a number or chart.

4. **How many independent runs were performed?**  
   One run per prompt is essentially anecdotal. I’d want at least **three fresh-chat runs per prompt**, preferably on different days.

5. **What exactly counts as visibility?**  
   These should be reported separately:
   - Brand mentioned
   - Brand recommended
   - Position in the answer
   - Linked or cited
   - Description accurate
   - Sentiment positive/neutral/negative

6. **Were prompts genuinely buyer-oriented?**  
   “What is Acme?” measures recognition.  
   “Best invoicing tool for a two-person US consulting firm” measures commercial discovery.

7. **Does the score include uncertainty?**  
   “Mentioned in 9 of 30 runs” is useful. “Visibility score: 76” without raw counts is not.

## Red flags

Be skeptical if:

- The methodology is proprietary.
- You cannot export prompts and responses.
- Every prompt includes your brand name.
- Competitor and category prompts are missing.
- Results come from one run on one day.
- Search mode, location, model, and date are undisclosed.
- Mentions and citations are treated as equivalent.
- The score moves dramatically after tiny website changes.
- The vendor guarantees higher rankings.
- It claims to represent “ChatGPT visibility” while testing only an API model.

## A free DIY validation test

Create **15 prompts**:

- 5 category prompts: “Best tools for ___”
- 5 problem prompts: “How can I solve ___?”
- 3 comparison prompts: “Alternatives to [competitor]”
- 2 high-intent prompts: “What should a [specific customer] buy for ___?”

Then:

1. Run each prompt **three times in fresh conversations** on ChatGPT and Grok.
2. Keep search settings consistent.
3. Record the date, interface, model if shown, answer, citations, and your brand’s position.
4. Calculate:
   - **Mention rate:** mentions ÷ total runs
   - **Recommendation rate:** genuine recommendations ÷ total runs
   - **Citation rate:** responses citing your domain ÷ search-enabled runs
   - **Accuracy rate:** accurate descriptions ÷ mentions
5. Repeat monthly using the same prompt set.

A simple spreadsheet is enough. Your baseline might look like:

| Metric | ChatGPT | Grok |
|---|---:|---:|
| Mention rate | 8/45 | 5/45 |
| Recommendation rate | 3/45 | 2/45 |
| Domain citation rate | 2/45 | 1/45 |
| Accurate descriptions | 7/8 | 3/5 |

That is more informative than “AI visibility: 42.”

## Best purchasing rule

Only pay for a visibility tool if it saves enough manual work **and** its results substantially match your DIY sample. Compare 10–15 identical prompts. If the tool reports strong visibility while your controlled tests rarely show your company, ask the vendor to explain the discrepancy before trusting it.

The most trustworthy product is not necessarily the one with the fanciest score. It is the one that lets you inspect every prompt, response, model, search setting, date, citation, and calculation.

Recorded sources

These are the source roles recorded in the public dataset. A retrieval or source-panel item is not automatically an answer-body citation. A URL supplied in the prompt and repeated by an answer does not prove content access. No destination has been repaired or independently validated here.

Recorded answer-body links (2)

Retrieval-only / source-panel inventory (14)

Evidence and qualifications

This is the existing redacted public record. It has not been rewritten for this page. Earlier privacy edits and collection qualifications continue to apply. Editorial text is English; a few original captures retain non-English UI badges or source-panel date labels. An answer’s empirical claims are platform output, not verified findings of this archive.

Body review

{
"reviewed_at": "2026-11-02T09:31:30.426176+00:00",
"reviewer": "assistant",
"complete_body_read": true,
"coherent_ending": true,
"annotation_and_explicit_url_coverage_checked": true,
"external_claims_verified": false,
"independent_human_double_coding": false
}

Existing public-text privacy policy

{
"body_changed": true,
"policy": "Only actual URL tracking, signed/session URL parameters and necessary identifying information are removed; original wording, including analytics parameter examples outside URLs, remains.",
"annotation_positions": "Original position indexes omitted because URL/privacy edits can change character locations."
}

Prompt SHA-256: e927e70f9860279797a292b031efab8dca60a13a2936d0a237c5c2c015170469
Answer SHA-256: 63b8cc69ccafdb1cdda1f7cc203103f4ec36cac1ec11c6b59cf3d0ae471a99d6