Skip to main content

Impact Assessment Reference

Scoring Criteria — Detailed Definitions

Frequency (1-5)

How often does this task occur per person (not per team)?

ScoreDescriptionExample
1Monthly or lessQuarterly compliance review
2WeeklyWeekly cost review meeting prep
32-3 times per weekPre-deployment readiness check
4DailyMorning dashboard review
5Multiple times per dayAlert triage (on-call rotation)

Time per Instance (1-5)

How long does one instance of this task take from start to finish?

ScoreDescriptionExample
1Under 5 minutesSingle CLI query, quick check
25-15 minutesReview output, take simple action
315-30 minutesMulti-step investigation, draft response
430-60 minutesDeep diagnosis, cross-system correlation
5Over 60 minutesFull incident response, complex analysis

Error Risk (1-5)

How likely is a human to make a mistake on this task, especially under pressure?

ScoreDescriptionExample
1Almost never — routine, hard to get wrongRun a read-only report
2Rare — occasional minor slipCopy-paste a wrong value
3Occasional — a few mistakes per monthMiss a step in a checklist
4Frequent under pressureSkip a rollback step during an incident
5High risk — incidents happen hereManual database migration with no dry run

Tool Count (1-5)

How many distinct systems, dashboards, or CLI tools does the task require?

ScoreDescriptionExample
1Single toolOne AWS CLI command
2Two toolsCloudWatch + Jira
3Three toolsCloudWatch + Jira + Slack
4Four toolsCloudWatch + Jira + Slack + PagerDuty
5Five or more toolsFull incident response spanning 6+ systems

Quadrant Interpretation Guide

QuadrantFrequencyComplexityROIStrategyDay 3 Capstone?
PRIME CANDIDATESHigh (3-5)High (6-10)HighestAutomate firstYes
QUICK WINSHigh (3-5)Low (2-5)GoodScript first, then agent if neededPossible
ASSIST MODELow (1-2)High (6-10)GoodAI-assisted tool for when it occursPossible (harder to test)
SKIPLow (1-2)Low (2-5)LowKeep manualNo

Complexity score = Error Risk + Tool Count (range: 2-10). The midpoint for "high complexity" is around 6.


Pre-Scored Examples for Calibration

Use these to calibrate your own scoring before you apply it to your tasks.

TaskFreqTimeError RiskToolsTotalQuadrant
Morning CloudWatch alert review533415PRIME
Pre-deploy readiness checklist344516PRIME
Monthly EC2 right-sizing review152311ASSIST
Certificate expiry check41117QUICK WIN
Weekly cost review prep242210QUICK WIN/ASSIST
Password reset21115SKIP
DNS record update11215SKIP

Day 3 Module Reference

Your capstone candidate connects to these Day 3 modules:

ModuleWhat You Build
Module 10: Domain AgentsChoose a track: SRE alert triage, cost analysis, or deployment validation — or bring your own
Module 11: Agent FleetScale your single agent to handle multiple concurrent tasks
Module 12: Event TriggersWire your agent to fire on alarm events, not just on-demand
Module 13: GovernanceAdd approval gates and audit logging to your agent
Module 14: CapstonePresent your finished agent, 30-day adoption roadmap

The better your problem statement from this module, the faster Module 10 goes.