AgentSkillsCN

engagement

深入分析用户参与度——包括会话时长、每次会话的操作次数、每日参与模式,以及页面浏览速率。

SKILL.md
--- frontmatter
name: engagement
description: Analyze engagement depth - session duration, actions per session, daily engagement patterns, and pageview velocity

Engagement & Session Analysis

You are a Customer Success analytics assistant analyzing how deeply users engage with Atlan.

Parameter Collection

Parse $ARGUMENTS for a domain. Ask for what's missing:

  1. Domain (required): "Which customer domain? (e.g., acme.atlan.com)"

  2. Focus (optional, default "overview"): "What aspect of engagement?"

    • sessions - Session duration trends (monthly + daily, with avg/median)
    • actions - Average actions per session per month
    • daily - Daily engagement matrix (event count distribution) and pageview trends
    • overview - All of the above (default)
  3. Start date (optional, default 3 months ago)

  4. Include workflows? (optional, default: no): "Include workflow/automation events? These system-generated events are excluded by default since they're massive volume noise from automated processes."

    • If yes: Before executing, remove the AND ... NOT LIKE 'workflow_%' filter from TRACKS queries in the SQL.
    • If no (default): Execute as-is (workflow events are already filtered out in the SQL files).
    • Do not ask this question unless the user mentions workflows — just use the default (exclude).

SQL File Mapping

FocusSQL File PathParameters
sessions (monthly)~/atlan-usage-analytics/sql/03_engagement_depth/session_duration.sqlSTART_DATE, DOMAIN
sessions (daily)~/atlan-usage-analytics/sql/03_engagement_depth/session_duration_daily.sqlSTART_DATE, DOMAIN
actions~/atlan-usage-analytics/sql/03_engagement_depth/actions_per_session.sqlSTART_DATE, DOMAIN
daily matrix~/atlan-usage-analytics/sql/03_engagement_depth/daily_engagement_matrix.sqlSTART_DATE, DOMAIN
daily pageviews~/atlan-usage-analytics/sql/03_engagement_depth/avg_pageviews_per_user_daily.sqlSTART_DATE, DOMAIN

Parameter Substitution

  • {{DOMAIN}}'acme.atlan.com' (single-quoted)
  • {{START_DATE}}'2025-11-13' (single-quoted date)

Execution

  1. Read the SQL file(s) from the paths above
  2. Replace {{START_DATE}} and {{DOMAIN}} with collected values
  3. Execute via mcp__snowflake__run_snowflake_query
  4. For "overview", run all queries sequentially

Important Context

Sessions are derived using 30-minute inactivity gaps (no Amplitude session IDs available in the data). Single-event sessions are excluded from duration calculations. Sessions longer than 8 hours are filtered as outliers.

Presentation

sessions

Show monthly trend of avg/median session duration. Convert seconds to minutes for readability. Explain that median is more reliable than mean (resistant to outliers). Benchmarks:

  • Median < 3 min = superficial usage (likely just logging in)
  • Median 3-10 min = moderate engagement
  • Median > 10 min = deep, productive sessions

For daily view, show recent 14-day trend. Flag days with unusually low/high session counts.

actions

Show avg and median events per session per month. Benchmarks:

  • < 5 events/session = low engagement (browsing only)
  • 5-15 events/session = moderate (active exploration)
  • 15 events/session = deep engagement (power usage)

daily

Show the engagement matrix: how many users fall into each bucket (0, 1-4, 5-9, 10-19, 20+ events) per day. Identify if most users cluster in the low buckets (shallow) or spread across (healthy distribution). Show avg pageviews per user trend.

Engagement Quality Assessment

Synthesize a 1-paragraph assessment combining session depth (duration), breadth (actions per session), and consistency (daily patterns). Rate overall engagement quality as Strong/Moderate/Shallow with specific evidence.