Commercial roofing contractor, leak calls to warranty claims, 2011 to 2025

Summary

Fifteen years of records from one US commercial roofing contractor doing low-slope replacement, repair and maintenance. Leak calls, roof inspection reports, CRM opportunities, warranty files, internal meeting transcripts and the mail around them show the full cycle from a first leak call to a reroof and a warranty claim years later. People, buildings and manufacturers are replaced with placeholders, amounts are removed, and the archive closes at the end of 2025.

Collected to order. We collect it from US companies for your order. Scope, volume and timing are confirmed in writing before you sign.

Pricing on request

Request scope and pricing

At a glance

Planned scope. Volume, sources and timing are confirmed in writing before you sign.

Records: 5.4Memails, CRM records, documents and transcripts
Size: 15.8 GB
Tokens, estimated: 1.8B
Years: 2011 to 2025
Firms: 1

Owner profile: One US commercial roofing contractor, 50 to 150 staff, low-slope replacement and service
Industry: Construction
Data type: Company operations archive
Headcount: 50 to 150 staff
Region: United States
Languages: English
Formats: JSONL, Parquet
Updated: 2026-10-11

Sources

System	Kind	Records
Google Workspace mail	Email	5.2M messages
HubSpot record exports	CRM	79K deals and service tickets
Shared drives (inspection reports, proposals, warranty files)	Shared files	139K documents
Zoom transcripts (2020 to 2025)	Call transcripts	4.4K meeting transcripts

Good for

training intake agents that turn a leak call into a scheduled inspection
extracting roof condition, defects and recommended repairs from inspection reports
evaluating reasoning over warranty terms against later leak history on the same roof
forecasting which service customers convert to replacement projects

Fields

Field	Type	Description
record_id	string	Unique record id
record_type	enum	leak_call, inspection, opportunity, warranty, email or transcript
roof_id	string	Placeholder for the roof section, stable across all records
created_at	timestamp	Creation time with the original UTC offset
roof_system	enum	tpo, epdm, mod_bit, bur or metal
text	string	De-identified narrative or body text
findings	array<string>	Defects and conditions found, for inspections
next_action	string	Recommended or scheduled next step
linked_records	array<string>	Calls, opportunities and warranty files for the same roof

Example record

{
  "record_id": "RF1-INS-0028841",
  "record_type": "inspection",
  "roof_id": "[ROOF_1873]",
  "created_at": "2021-04-08T11:15:00-04:00",
  "roof_system": "tpo",
  "text": "60 mil mechanically attached TPO, about 42,000 SF, installed 2013 per warranty file. Leak reported over aisle 14 after the last storm.",
  "findings": [
    "Open seam 18 in at RTU-7 curb flashing, inside corner not welded",
    "Ponding about 1.5 in deep at NE drain, strainer missing",
    "Wet insulation confirmed by core cut, about 300 SF around RTU-7"
  ],
  "next_action": "Temp patch today. Send core photos to [COMPANY_12] for warranty review before replacing wet insulation.",
  "linked_records": [
    "RF1-CALL-0119420",
    "RF1-WAR-0003310"
  ]
}

One record, de-identified. Placeholders such as [PERSON_12] and [COMPANY_4] stay the same across the whole dataset, so threads still read correctly.

Licenses available

Exclusive
One buyer. Nobody else can license it during the exclusive term, usually 2 years.
Non-exclusive
Licensed to more than one buyer.

What ships with it

Datasheet: what it is, what it is good for, verified size, sources, years, format and limits.
Provenance file: the owner shown by code, staff notice and consent status, exclusions, mapped to the EU AI Act Article 53 training-data summary and California AB 2013.
Manifest with a hash for every file.
QC report and de-identification report.
Your own copy, with unique canary records.

How it is de-identified

Default exclusions removed, files flattened to text, identifiers found in three layers and replaced with consistent placeholders, and the result measured against a 99% bar for each identifier type in each source.

The full method