📘 Best Practice: Asking Effective Questions

Knowledge Agents search your team's verified content and generate an answer in plain language, but the phrasing of a question still shapes the result. A specific, well-targeted question pulls from the right sources; a vague one risks a broad or off-target answer. This article covers five habits that make questions more effective, whether you're using search or chat.

Being specific

Add context instead of asking broadly.

• Poor: "Tell me about Workspaces."
• Better: "What are the key features of Workspaces for enterprise customers?"

A vague question tends to return an answer that's too broad or off-target. The more specific the question, the more accurate the answer.

Using keywords instead of general terms

Match the language your content actually uses.

• Poor: "What's the status of our security certifications?"
• Better: "What's the status of our SOC2 certification?"

"Security certifications" is broad enough to pull in unrelated results. Naming the specific certification narrows the search.

Breaking complex questions into simpler ones

Ask one focused question at a time instead of bundling several together.

• Poor: "What are the features, use cases, and competitors of Knowledge Agents?"
• Better:
  1. "What are the key features of Knowledge Agents?"
  2. "What are the primary use cases for Knowledge Agents?"
  3. "Who are the main competitors for the Knowledge Agents feature?"

A single, multi-part question can run out of room in the context window — how much content Guru can reference at once — so part of it may go unanswered. Splitting it up keeps each answer complete.

Avoiding ambiguous pronouns

Name the subject directly rather than relying on "our," "it," or "that."

• Poor: "What is our pricing?"
• Better: "What is Guru's pricing model?"

Guru can't resolve who "our" refers to, which can widen the answer more than intended.

Using consistent terminology

Stick to whichever form — the spelled-out term or the abbreviation — your documentation favors, and use it the same way each time.

• Poor: "How many sources can a KA have?"
• Better: "How many sources can a Knowledge Agent have?"

Even when an abbreviation is common internally, Guru may not always connect it to the full term used across your documentation.