Step 1: Write your prompt set
Choose 10 to 20 questions your customers really ask. Include category questions ("best tools for X"), comparison questions ("A vs B"), brand questions ("is [brand] good?") and problem-led questions.
Step 2: Ask each assistant
Run every prompt in several assistants, because they differ in training data and in how they retrieve sources. Use a fresh chat each time to avoid earlier answers influencing results.
Step 3: Record what matters
Keep a simple spreadsheet with one row per prompt and assistant.
- Were you mentioned? Were you recommended or only listed?
- Was the information about you accurate?
- Which sources were cited, such as Reddit threads, review sites or your own pages?
- Which competitors appeared?
Step 4: Act on what you find
If you are missing, look at the sources the assistants cite and ask why you are absent from them. Typical fixes are clearer answer-first content on your site, consistent brand facts across profiles, and authentic presence in the communities the assistants draw on.
Step 5: Repeat on a schedule
Answers change as models and retrieval systems change. Repeat the same prompt set monthly or quarterly so you can compare like with like and see trends.
