Skip to main content
A prompt is a realistic question a customer might ask an AI assistant. Onsomble sends each active prompt to the AI assistants and regions configured for a scan, records their answers, and measures how your business appears. For example, a commercial insurance Site might track:
  • “What does business interruption insurance cover?”
  • “Which insurance broker is best for a small retailer in Auckland?”
  • “How do I get a commercial insurance quote?”
Each question represents a different customer need. Together, the prompts form the sample that Onsomble uses to measure your AI discoverability.
Your results describe the prompts, AI assistants, regions, and completed scans you measured. They do not represent every question someone could ask about your market.

Why prompt choice matters

Every active prompt contributes to your scan results. The complete set determines which customer situations those results can describe. If your Prompt Library contains only questions that name your brand, the results show how AI assistants respond when your brand is already part of the question. They do not show whether the brand appears in broader discovery or recommendation questions. If the library contains only broad research questions, the results do not describe how your business appears when customers compare providers, check whether an option is suitable, or prepare to make a decision. A useful prompt set covers the parts of the customer decision that matter to the business:
  • what product or service they are considering;
  • which Persona they belong to;
  • what they are trying to achieve;
  • where they are in their decision;
  • which market or region changes the answer.
See Build a prompt strategy for the full framework.

A prompt captures intent and context

A keyword names a topic, such as commercial insurance. A prompt gives the AI assistant a goal and enough context to answer:
  • Goal: compare providers, understand a risk, check proof, or take a next step.
  • Context: the product, customer, market, constraint, or situation that changes the answer.
Commercial insurance identifies a subject. “Which commercial insurance broker is best for a two-store retailer in Auckland?” identifies what the customer wants and the context that matters. Write prompts as natural questions. Minor rewrites usually add less coverage than a genuinely different customer need or context.

Keep a stable core

Use a stable core set of prompts across scans so changes over time remain easier to interpret. Add or update prompts when the business changes or you identify a meaningful coverage gap, rather than rewriting them in response to one result.

What’s next

Build a prompt strategy

Choose the products, customers, stages, markets, and intents worth measuring.