AXO-8 — AI visibility test recording worksheet
https://axo-8.com/en/insights/how-to-check-chatgpt-visibility
Version: 20 October 2026

SET THE TEST
Company:
Service:
Target buyer:
Market:
Question language:
Systems and visible model labels:
Search mode:
Sign-in state and account context:
Location settings (write unknown if not controlled):
Test date and time zone:
Exact discovery questions (do not name your company):
Brand-accuracy questions (record separately):
Planned repetitions per question and system:

ONE RECORD PER OBSERVATION
Question ID:
Exact question:
Question group: discovery / named-brand accuracy
System and visible model:
Search mode and sign-in state:
Date, time and time zone:
Repetition number:
Status: complete / unavailable / failed / not run
Full first answer or saved capture reference:
Company mentioned: yes / no / not assessed
Recommended as a provider to consider: yes / no / not assessed
Description accuracy and specific errors:
Other named providers:
Visible source labels:
Opened source URLs and final destinations:
Does each source support the claim?:
Context or conditions that changed:
Specific next inspection:

COMPARE COMPLETED DISCOVERY ANSWERS PER SYSTEM
Planned observations:
Completed observations:
Unavailable, failed or not-run observations:
Completed answers mentioning the company:
Completed answers recommending the company:
Answers with a checked relevant citation:
Mention coverage = mentioning answers / completed discovery answers.
Keep named-brand checks out of the discovery denominator.
Keep systems separate. An unavailable run is not an answer with zero mentions.
Record source labels separately from verified final URLs.
These are measures of this sample, not market share or recommendation probability.

BUSINESS OUTCOME — TRACK SEPARATELY
Attributable visits:
Received enquiries (exclude tests and duplicates):
Qualified conversations:
Offers and sales:
Unknown attribution:

REPEAT
Keep the exact original questions, available systems, settings and scoring rules.
Record publication dates and changes before the next observation.
Mark model, search-mode, location and access changes as comparison limits.
A before/after change does not establish what caused it.
The free AXO-8 scanner checks web/search signals; it does not run these prompts.