AI citation investigation playbook

This playbook is for work shared by discovery, content, catalog, and research teams.

Set the baseline

Choose the answer systems, markets, products, query intents, schedule, and retention rules before collecting results. Use the same capture format for answers and citations.

Steps

1. Create a fixed set of buying questions.
2. Record the answer, cited pages, merchant attribution, and competing products.
3. Check merchant-owned facts against facts on cited third-party pages.
4. Classify the gap: crawl or markup issue, catalog error, missing content, entity confusion, or selection outside the merchant's control.
5. Make one accurate change on a merchant-controlled page or data source.
6. Run the same questions again.
7. Describe the differences without claiming more than the retest shows.

Done when

• The baseline and retest used documented conditions.
• Any edited claim is accurate and helpful to a buyer.
• Mentions, citations, factual accuracy, and placement are reported separately.
• The report makes no promise about future placement or revenue.