What should you prepare
When should you use this guide?
Workflow
1
1. Create a ChatGPT question set
Split recommendation, comparison, definition, and buying-review questions in Prompt Status.
2
2. Classify brand appearance in Prompt Analysis
Record whether the brand is first, one of many, only described, or absent.
3
3. Check spelling and description accuracy
Review variants such as HaloX and Korean brand names; move incorrect descriptions into source work.
4
4. Connect competitors and evidence
Send competitor-heavy questions to Strategy Map and weak evidence questions to Citation Tracking.
5
5. Report by question type
Avoid “appears/does not appear”; report comparison, definition, and buying questions separately.
How do you verify the result?
ChatGPT mention check table
What commonly blocks teams
Is it a bug if GSC/GA4 numbers and HaloX scores do not match exactly?
Is it a bug if GSC/GA4 numbers and HaloX scores do not match exactly?
No. GSC and GA4 describe search traffic and user behavior, while HaloX AVI, citation rate, and source visibility describe how the brand and sources appear in AI answers. They support the same operating decision but use different measurement units.
Should the same question set be reused forever?
Should the same question set be reused forever?
Keep baseline questions, but add separate sets when products, competitors, or buyer questions change. Do not overwrite the baseline; split the comparison axis so weekly reports can explain the cause of movement.
Where should the guide output be recorded?
Where should the guide output be recorded?
Move evidence into Weekly Report and Exports, and convert actions into Content Production briefs or Site Audit issues. In meetings, report the question group, source gap, and next action together, with the score.
Related docs
AI Visibility Index
Read the core AI visibility score.
Prompt Analysis
Inspect answers and citations by question.
Content Production
Turn findings into briefs.
Weekly Reports
Package changes for meetings.