Account Fit Scoring Agent: QA checklist

To be completed by a non-builder reviewer before go-live.

Deployment checklist

• ☐ Scoring weights config written from real closed-won deal data and versioned
• ☐ Seven HubSpot company properties created
• ☐ n8n workflow imported, marked nodes configured
• ☐ Demo-data run matches `demo-data/sample-output.json`
• ☐ Score-distribution QA test passes (no single tier >60%)
• ☐ insufficient_data handling verified for new accounts
• ☐ Weekly Slack digest lands in agreed channel
• ☐ QA checklist completed by a non-builder
• ☐ Rollback tested

Failure-mode verification

• ☐ Verified: Every account scores mid-band does not occur (mitigation: QA checks score distribution; flags if >60% of accounts land in one tier)
• ☐ Verified: New accounts scored zero and buried does not occur (mitigation: Schema requires `insufficient_data` flag path; QA test asserts it fires for accounts <30 days old)
• ☐ Verified: Score changes with no explanation does not occur (mitigation: Every score write includes `account_score_model_version`; changes are diffable)
• ☐ Verified: Readiness score double-counts fit signals does not occur (mitigation: Config documents which signals feed which sub-score exclusively)

Governance verification

• ☐ Agent writes only to its agent-owned fields: `account_fit_score`, `account_commercial_value_score`, `account_readiness_score`, `account_engagement_score`, `account_priority_score`, `account_score_model_version`, `account_score_date` (company)
• ☐ No auto-send, auto-close, auto-route, or auto-create action occurs without the human step described in `agent-spec.md`
• ☐ Confidence/model-version is present on every written record where applicable
• ☐ Demo-data run output matches `demo-data/sample-output.json` in structure

Sign-off

Reviewer name: _______________ Date: _______________ Result: Pass / Fail (attach notes)