Skip to main content

Performance reporting scenario

Regularly sharing performance status with executives and stakeholders is a key output of monitoring operations. WhaTap already provides APM reports (daily/weekly/monthly) and integrated reports (multi-project), so this guide focuses on which cadence maps to which audience — how to make reporting a routine.

Who this guide is for
  • Team leads, SREs, and engineering managers
  • Engineers who need to share performance status regularly with executives or product teams
  • Teams wondering "does anyone actually read our weekly metric-list report?"

Prerequisites​

The 3-report set​

Combine three reports with different purposes and cadences. Avoid the "one-size-fits-all" report that ends up serving no one well.

ReportCadenceAudienceCore question
Weekly ops reportWeeklyDev team / SRE (internal)What happened last week, what do we handle this week?
Monthly executive reportMonthlyLeads / executivesHow is performance affecting the business and where should we invest?
Quarterly performance retroQuarterlyWhole teamTrends, incident patterns, improvement metrics

Weekly ops report​

Make it the most frequent but the lightest. Familiarity with the same format each week is what matters.

Source data​

  1. In Report, select Weekly report → query the previous week
    • Summary: Apdex, TPS, average/max response time, error rate
    • Mid-week peak and anomaly windows
  2. Capture last week's events from Event history
  3. Screenshot directly from the team Flexboard/dashboards

Report skeleton​

Weekly performance report — YYYY-Www

1. One-line summary (Good · Fair · Watch)
2. Key metrics (WoW delta)
   - TPS, Apdex, avg response, error rate
3. Events that fired (top 3–5)
4. Actions this week
5. References (Flexboard URL, postmortem docs)
Emphasize week-over-week

"Avg response 180ms" is meaningless. "Avg response 180ms (+15% WoW)" is an interpreted number. A report is interpretation, not raw numbers.

Monthly executive report​

Executives don't care how the numbers moved — they care how it impacted the business.

Source data​

  1. In Report, query Monthly report
  2. For multiple products/projects → use Integrated reports (see APM integrated report)
  3. Collect links to incident postmortems

Report skeleton​

Monthly performance report — YYYY-MM

1. Business impact summary
   - "One incident affected 10,000 users for 3 minutes"
   - "Payment API avg response improved 20% MoM"
2. KPI trends (charts)
   - Monthly Apdex, error rate, availability trend (3–6 months)
3. Major incidents / improvements (1–2 lines each)
4. Next month's risks / investment proposals
   - "DB capacity expected to hit 80% — expansion required"
Translate engineering terms into customer language

Not "GC pause increase" but "risk of user-perceived latency." An executive report that uses business terms — not WhaTap terms — actually gets read.

Quarterly performance retro​

A report for the whole-team retrospective meeting each quarter. Improvement proposals are the core.

Source data​

  1. Compare three monthly reports across the quarter in Report
  2. Full list of incident postmortems in the quarter
  3. Use integrated reports to summarize trends across products/projects

Report skeleton​

Quarterly performance retro — YYYY-QN

1. Trends (3-month movement)
   - TPS, response time, error rate, availability
2. Incident pattern analysis
   - Category-level incident frequency
   - Recurring incidents hint at unresolved root causes
3. Performance improvement outcomes
   - Which change improved which metric by how much
4. Proposals for next quarter (top 3)

Automation tips for reporting​

Pin Flexboard URLs​

Instead of taking screenshots every time, attach the filtered Flexboard URL in the report so readers can pull the latest data themselves when needed.

Bookmark event history​

Bookmarking Event history with a week-scope filter dramatically shortens weekly report prep.

Lock in templates​

Pin the report skeleton as a team wiki template and copy weekly/monthly. Just maintaining the format accelerates consumption.

Use AI Chatbot for interpretation and draft copy​

The hardest part of a report is "finding meaning in the numbers and turning them into sentences." The WhaTap AI Chatbot shortens this process with natural-language prompts.

  • "Apdex dropped from 0.85 to 0.78 last week — possible causes?" — metric interpretation
  • "Summarize this week's event list one line each for executives." — draft copy
  • "Could increased heap memory alerts be a leading indicator?" — context judgment

AI output is a draft; the final call is yours. But not starting from a blank page cuts the reporting overhead substantially.

Auto-generate reports via MCP​

WhaTap MCP lets AI agents like Claude Code or ChatGPT access WhaTap data directly. With MCP, you can delegate the report generation itself to an AI.

  • "Summarize last week's APM key metrics into the weekly report template" → AI fetches data → markdown draft
  • "Create a month-over-month error rate chart and interpretation" → comparison automation
  • Specify the executive report's tone as a prompt for consistent voice

Details: MCP usage guide

Review AI-generated reports

Reports created by MCP or the chatbot must have their data source and interpretation verified by a human. Executive reports especially require exact numbers — accept AI as a draft, but verify final figures directly in WhaTap.

Verify​

  • Each report (weekly/monthly/quarterly) has a dedicated audience and cadence
  • Every metric in the report is annotated with period-over-period change and interpretation
  • Executive reports are written in business impact, not engineering jargon

Once reporting becomes a routine, monitoring data has a channel out of the team. If incident response is "reaction," reporting is "propagation."

Next steps​

Last updated on