Complete AI Training
Sign inGet my AI kit

Your job's AI kit

Get your AI kit

Tell us who you are and what you do. We show you your kit right away and email you the link: skills, prompts, AI agents, MCP servers and courses for your job.

500+ jobs ready, and we make a kit for any other job. No payment needed to look.

Share

AI agent for product analysts

KPI Anomaly Investigation Agent

Explain every unusual KPI move within hours

KPI Anomaly Investigation Agent: what goes in, what the agent does and what you get

What it does

When a key metric drops overnight, product teams panic, and analysts spend hours finding out it was a tracking change. Each morning this agent checks key product metrics against their expected range for that day of the week. When one moves outside the range, it investigates in order: was data late or incomplete, did an event definition or app release change tracking, is the change limited to one platform, country or segment, and does it line up with a launch or outside event. After each step it checks whether the cause is found. If not, it moves to the next split. It drafts a short note with the cause, evidence and impact. The analyst confirms the note before it goes to the product team. Edge case: a drop only on the newest app version points to a tracking change first.

How it works

Follow the arrows from top to bottom. The orange dashed arrow is the loop: when a check fails, the agent goes back and tries again.

Start and resultWhat it doesA check on its own workWaits for your OKGoes back and retries
Yes, continueApprovedNo 1 STARTS WHEN Daily data load complete 2 USES A TOOL Compare metrics with expected daily ranges 3 USES A TOOL Check data freshness and completeness 4 DOES Check release notes and event schema changes 5 DOES Split by platform, version, country and segment 6 CHECKS THE RESULT Is the cause clear and supported by the splits? If not: try the next split or check the launch calendar.Back to step 5. 7 USES A TOOL Draft the anomaly note 8 YOU APPROVE Analyst confirms before sharing 9 RESULT Note posted and logged
Read the steps as a list
  1. Daily data load complete
  2. Compare metrics with expected daily ranges
  3. Check data freshness and completeness
  4. Check release notes and event schema changes
  5. Split by platform, version, country and segment
  6. Is the cause clear and supported by the splits?If not: try the next split or check the launch calendar. Back to step 5.
  7. Draft the anomaly note
  8. Analyst confirms before sharingThe agent waits here for your OK.
  9. Note posted and logged

How it decides

It rules out data and tracking problems before looking for real user behavior changes.

  • Rule out data delays first
  • A drop on only the newest version points to tracking
  • Notes go out only after analyst review

Make it yours

Every agent is a starting point. You choose these settings for your own situation.

  • Metrics watched
  • Range sensitivity
  • Split order
  • Who receives notes

What keeps you in control

It always asks you first

  • Sharing the note with product teams

Hard limits

  • Read-only data access
  • Never changes metric definitions

It stops when

  • Done: cause found and shared
  • Stop: data pipeline down, alert data engineering

Set it up

We guide you through the set-up, step by step

Members get the full set-up guide for this agent. No technical skills needed: you copy, paste and upload.

10 minto set it up in your AI
5 AIsChatGPT, Claude, Copilot, Gemini, Grok
  • One set of instructions to paste into your AI, with the clicks for ChatGPT, Claude, Microsoft 365 Copilot, Gemini and Grok
  • The agent then walks you through connecting your own data, one source at a time
  • A downloadable copy with the flow chart, the rules and the full guide
Get access to this agent

An example run

What happensOn Tuesday, daily checkouts fell 18%. Data was complete, so the first cause check failed and the agent moved to platform splits. The drop appeared only on Android version 5.2, released the night before. Release notes showed a renamed checkout event. The agent drafted a note saying the drop was tracking, not users, and the analyst confirmed it.

More agents for product analysts