onboarding-tutor

Onboarding Tutor

Guide new users through Fullstory MCP — teach the mental model, run their first queries, and explain how to use each skill effectively.

When to Use

• "I'm new to Fullstory MCP — how do I use this?"
• "Teach me how to ask questions about our product data"
• "What can I ask Claude about Fullstory?"
• "Walk me through my first analysis"
• "What skills should I know about?"
• "A new PM joined — onboard them to Fullstory"

Mental Model

Fullstory MCP turns your product data into a conversation. The key concepts:

1. Segments = groups of users ("who")
2. Metrics = measurements ("what" and "how much")
3. Sessions = evidence ("why")

You don't need to know SQL or analytics. You ask questions in natural language. Claude figures out which tools to use.

Workflow

Step 1: Welcome and assess

Ask the new user:

• "What's your role? (PM, engineer, designer, support, exec)"
• "What kind of questions do you want to answer?"
• "Have you used Fullstory's UI before, or is this your first time?"

Tailor the onboarding to their role:

• PM: Metrics, funnels, feature adoption, experiments
• Engineer: Errors, API monitoring, deploy validation, session debugging
• Designer: Session review, mobile analysis, a11y, scroll depth, heatmaps
• Support: Session search, customer 360, user journey, bug reproduction
• Exec: Dashboards, weekly digests, benchmarks, revenue impact

Step 2: Run a "hello world" query

Start with something simple and guaranteed to work:

"Let's start with the simplest question: how many users visited your site yesterday?"

Walk through what happens:

1. Claude builds a metric: fullstory:build_metric(query="page views yesterday", output_type="single_number")
2. Claude computes it: fullstory:compute_metric(metric_id)
3. Claude presents the answer: "12,340 page views yesterday."

Explain that every question follows this pattern: build → compute → present. More complex questions might add: search first → build → segment → compute → validate → investigate sessions.