Try Prompts in X

Problem

500K users visited the Copilot Prompt Gallery website (see above) every month, but the traffic lead to a dead end.

Users had to manually copy prompts, navigate to the correct Microsoft 365 app, and paste before submitting, a 6-step workflow with significant drop-off at each stage.

Prompt usage was a leading indicator of Copilot retention: users who tried prompts were 70% more likely to return. That made the funnel drop-off a direct retention and growth problem.

Solution

I led a cross-product initiative to transform the Prompt Gallery website from a static content site into an interactive activation surface by introducing a one-click “Try in X” experience.

Each prompt got a "Try in X" button that opened the right app (Copilot Chat, Word, Excel, or PowerPoint) in a new tab with the prompt pre-populated — ready to add context and submit (see cover image above)

The solution reduced the user journey from 6 steps to 4:

This resulted in:

• 70% website → in-app activation rate of high intent users

• 99% reliability in landing users in-app

• 50% reduction in latency (time from click to seeing the prompt pasted)

• Established a scalable foundation for future cross-surface experiences

My Role & Takeaways

• Led a cross-functional initiative across 5 product and engineering teams to align on a shared goal of reducing cross-application friction (Cross-Functional Leadership, Product Ownership)

• Influenced partner teams to prioritize latency and reliability as user-facing product metrics, not just engineering concerns (Stakeholder Management, Systems Thinking)

• Diagnosed and resolved performance bottlenecks by instrumenting the handoff flow and parallelizing app initialization, reducing latency by 50% (Analytical Thinking, Technical Collaboration)

• Advocated for user trust by prioritizing 99% reliability and addressing edge cases like read-only file failures (User-Centric Decision Making, Quality at Scale)

• Navigated tradeoffs by prioritizing activation over feature expansion (e.g., no auto-submit, no desktop support) to maximize speed and impact (Product Judgment, Prioritization)

  
  

Long Term Impact: This work enabled embedded links to be used in marketing initiatives and blog posts that empowered customers to try custom prompts from anywhere.