AI Citation Baseline Audit (Onboarding)
A checklist with 7 steps: Freeze the client's keyword universe into two tiers before any tracker is connected.
By InnovaAI ResearchPublished
What are the steps?
AI Citation Baseline Audit (Onboarding)
- 01
Freeze the client's keyword universe into two tiers before any tracker is connected
Tier one holds 20 to 50 commercial-intent terms the client already ranks for; tier two holds 15 to 30 prompt-style queries buyers actually type into assistants. Lock both lists in a shared doc with a version date so month-over-month comparisons stay honest.
- 02
Capture a manual baseline of AI answers for every tier-two prompt on the day the retainer starts
Run each prompt in at least two assistants and screenshot the raw response. This becomes the reference point when a tracker later reports a visibility score, and it costs an account manager roughly 90 minutes.
- 03
Record which third-party sources the assistants cite for each prompt
Note whether the answer leans on review sites, forums, vendor pages, or news. Citation sources tell the delivery team where to earn placement, which is more actionable than a bare mention count.
- 04
Configure the tracking tool with the same prompt wording used in the manual baseline
Paraphrasing prompts between the manual pass and the tool setup breaks comparability. Paste the exact strings, including punctuation and location modifiers.
- 05
Set the reporting cadence and volatility threshold in writing with the client
Agree on weekly or daily checks and define what counts as a material move, for example three or more positions on a tier-one term. Anything below the threshold stays out of the client-facing deck.
- 06
Assign one owner for search-engine data and one for AI-platform data
Splitting ownership prevents the common failure where traditional rankings get reviewed monthly while AI citation drift goes unnoticed for a quarter. Name both people in the onboarding doc.
- 07
Publish a one-page baseline summary to the client within five business days
Include tier-one positions, tier-two AI mention rates, and the named citation sources. Agencies that hand over a baseline early have a defensible starting line when the first algorithm update lands.