How to Improve Product Profiles by Testing Signal Criteria with the Pursuit MCP

Last updated: September 25, 2026

When to Use This

Use this when you want to try new signal criteria for a product profile and see the signals they return before you change the live profile.

Before You Start

  • The Pursuit MCP is connected to your AI assistant (see Using Pursuit with AI Assistants)

  • Admin access to the product profile, if you plan to save changes

  • Context on why customers buy the product, such as won-deal notes, RFPs, or call summaries (optional, but it produces better criteria)

Steps

  1. Open a new conversation in your AI assistant with the Pursuit MCP connected.

  2. Ask for a baseline: "Preview signals for the [Product] profile." The assistant returns sample signals for the current criteria.

  3. Share context on buyer pain. Paste your notes, or ask the assistant to search Pursuit documents for accounts that bought this product in the past and what led to the purchase.

  4. Tell the assistant to iterate without saving. Example: "WITHOUT SAVING THE PROFILE, test signal criteria derived from this context until you have 10-15 criteria that each return multiple hits per month."

  5. Ask for a summary at the end: "Print the profile that worked and the changes you recommend."

  6. Review each proposed criterion and its sample signals. Cut any criterion whose signals don't indicate demand for the product.

  7. Save the approved criteria. Ask the assistant to save the profile, or paste the criteria into the product profile yourself.

  8. Run Tune signals on the saved profile to confirm quality with your own answers.

Tip: Keep each criterion to one buying condition, such as "council approves funding for a utility billing replacement." Stacked conditions return fewer hits and are harder to tune.

Warning: Always include "without saving" in your prompt when you're experimenting. Without it, the assistant may update the live profile your team sees in Radar.

What Success Looks Like

You have a list of 10-15 signal criteria, each backed by sample signals that show real demand for the product, ready to save.

Common Issues

Problem: A criterion returns zero or one hit per month. Fix: Ask the assistant to broaden the wording or to drop a qualifier, then preview again.

Problem: Sample signals are off-topic. Fix: Give the assistant more buyer context, such as the documents that preceded past purchases, and ask it to rewrite the criteria from that.