How to Use This Tool
1Review each deal — click any row to expand it. You'll see the deal notes, contact sentiment, company account signals, and the AI's classification rationale.
2Adjust the reason if needed — the AI Lost Reason dropdown is pre-populated from note mining. Override it if you disagree.
3Click Approve — marks the deal as reviewed and includes it in the HubSpot import export. Skipped deals are excluded from the import file.
4Use the filters to prioritize — start with 🔥 Score ≥8 for highest-value re-engagement targets, or ● High conf to batch-approve the most certain classifications first.
5Export & Import — when done, click ↓ Export HubSpot Import. This produces a clean 2-column CSV (Record ID + Lost Reason) for direct HubSpot import.
Understanding Confidence
Confidence reflects how much note evidence exists to support the classification — not the quality of the deal or the certainty of re-engagement.
● HIGH An explicit rep note names the reason directly ("chose Jefferson Electric," "lost funding," "cost overruns caused owner to decline"). Safe to approve as-is.
◑ MED Strong contextual inference — e.g. a batch of deals bulk-closed on the same date indicating a competitive RFP package loss. Reasonable but worth a quick scan.
○ LOW Thin or indirect note coverage. The classification is a best guess from deal name, owner, amount, and timing patterns. Review before approving.
∅ NONE No substantive notes found. Reason is a placeholder only. Do not approve without manual verification in HubSpot.
HubSpot Import Instructions
1Click ↓ Export HubSpot Import. Only Approved deals are included. The file contains exactly two columns: Record ID and Lost Reason.
2In HubSpot: go to Contacts → Imports → Import file. Select Deals as the object type and Update existing records.
3Map Record ID → Deal ID and Lost Reason → Lost Reason (property: closed_lost_reason).
4HubSpot will only update the closed_lost_reason field on matched deals. No other properties are touched.
5After import, verify a sample of deals in HubSpot to confirm the Lost Reason field populated correctly before closing this workflow.
⚠ Note: HubSpot import only updates records where the Deal ID matches exactly. Deals with no match are skipped silently — always check the import results summary for any unmatched rows.
What This Project Does
This tool is the output of an AI-augmented CRM data mining workflow built on top of SES's HubSpot instance. It addresses a specific data quality gap: 138 out of 156 Closed Lost deals in the CIA 2025 Commercial pipeline (88%) had no closed_lost_reason logged. Without this field populated, pipeline reporting, rep accountability tracking, and re-engagement targeting are all blind spots.
How The Data Was Derived
1Deal extraction — All 138 CIA 2025 Closed Lost deals with no lost reason were pulled from HubSpot via MCP API (pipeline 749816422, stage 1089696564).
2Note mining — All notes associated with each deal were pulled and parsed using objectType: notes with associatedWith deal filters. This surfaced rep activity logs, postmortems, bid feedback, and FOIA results.
3Contact sentiment — Primary contacts associated with each deal were pulled and their note histories analyzed for intent signals: energy bills uploaded, qualification notes, proactive inquiries, and engagement status.
4Company account signals — Associated company records were checked for active deal status, account relationship health, and any notes indicating broader account activity (e.g., LG&E's active construction project, DSD Renewables contract near execution).
5AI classification — All signals were synthesized to assign a closed_lost_reason from the 14 valid HubSpot enum values, a confidence level (High/Med/Low/None), a re-engagement flag, and an outreach priority score (1–10).
Derivation from Prior Work
This project is the second phase of a broader CIA 2025 pipeline intelligence initiative. The first phase produced a 64-deal "cliff list" — deals with no activity in 30+ days across active pipeline stages — delivered as both an interactive HTML artifact and an Excel file to flag stalled revenue. That work established the MCP note-mining methodology and deal association patterns reused here. This phase extends the approach to closed lost records, shifting the focus from pipeline health to historical loss classification and re-engagement targeting.
Intended Outcomes
Data Quality
Populate closed_lost_reason on 138 previously blank CIA 2025 deals, enabling accurate pipeline loss reporting.
Re-Engagement
Identify ~30 deals with re-engagement potential (score ≥6, flagged ↑) for targeted outreach by deal owner.
Account Intelligence
Surface account-level signals (active LG&E project, DSD contract) that change the context of individual closed lost deals.
Process Template
Establish a repeatable AI-assisted CRM data enrichment workflow applicable to residential pipeline and future commercial cohorts.
Data Sources & Limitations
All data sourced from HubSpot CRM via MCP API. Classifications reflect note content available at time of analysis (March 2026). Deals with no notes (∅ NONE confidence) could not be meaningfully classified — those should be manually verified before import or left as blank rather than importing a placeholder reason.