A tracked prompt is the unit of measurement for AEO. Each one is a specific question — exactly as a user would phrase it to an AI assistant — that BCited submits on a schedule and inspects the answer for whether your domain appears.
What makes a good tracked prompt
The prompts you want to track are the questions your prospective customers actually ask AI assistants:
- Direct product queries — "best [your category] for [persona]"
- How-to queries — "how do I implement [thing your product does]"
- Comparison queries — "[your category] vs [competitor category]"
- Foundational queries — "what is [domain concept]" — usually the highest-volume
Vague or branded prompts ("tell me about BCited") aren't useful for AEO measurement — you want to know whether the engine reaches for you when the user is NOT looking for you specifically.
How the runs work
Each tracked prompt is dispatched to OpenAI, Anthropic, and Perplexity via AI Gateway. The answer is captured, the citation list is parsed, and the result lands in the aeo_runs table with brand_mentioned, brand_cited, and brand_rank flags. Daily cron (Pro and Agency tiers) fans these out automatically; Free tier users can manually trigger via Run now.
Caps
Free gets 2 prompts (manual only). Pro gets 10 with daily automation. Agency gets 25.