Plumbing and drain service call transcripts with booking notes, 19 firms

Summary

Customer call transcripts and the booking notes CSRs wrote from them, from 19 US plumbing and drain cleaning companies. Each call is paired with the job it created, so the set shows how a caller's description of a problem becomes a job type, a priority and an arrival window. Audio is dropped at collection, and callers, addresses, phone numbers and payment details are removed or replaced with placeholders.

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: 3.3Mcall transcripts, jobs and emails
Size: 11.9 GB
Tokens, estimated: 1.7B
Years: 2021 to 2026
Firms: 19

Owner profile: 19 US residential and light commercial plumbing and drain companies, 6 to 85 staff each, Southeast, Midwest and Pacific Northwest
Industry: Field services
Data type: Support and service
Headcount: 6 to 85 staff per firm
Region: United States (Southeast, Midwest, Pacific Northwest)
Languages: English, Spanish
Formats: JSONL, Parquet
Updated: 2026-10-06

Sources

System	Kind	Records
ServiceTitan call transcripts	Call transcripts	1.38M calls
ServiceTitan bookings and jobs	Field operations	1.02M jobs
Google Workspace mail	Email	860K messages

Good for

training booking agents on real service-call dialogue, from transcripts only
classifying call intent, job type and urgency from the first minute of a call
evaluating whether booking notes capture what the caller actually said
measuring how CSRs handle reschedules, cancellations and repeat callers

Fields

Field	Type	Description
call_id	string	Unique call id within the pool
firm_code	string	Placeholder for the company
started_at	timestamp	Call start time
duration_s	int	Call length in seconds
language	enum	en, es or mixed
transcript	array<string>	Speaker-labeled turns, de-identified
call_intent	enum	new_service, existing_job, reschedule, cancellation, billing or other
booking_note	string	Note the CSR wrote on the job
job_id	string	Job created or updated by the call
job_type	enum	drain_cleaning, water_heater, leak_repair, sewer_camera, fixture and others
priority	enum	emergency, same_day or scheduled
outcome	enum	booked, not_booked, callback_requested or existing_job

Example record

{
  "call_id": "PL19-CALL-0920144",
  "firm_code": "[COMPANY_16]",
  "started_at": "2025-02-03T07:02:00-08:00",
  "duration_s": 214,
  "language": "en",
  "transcript": [
    "CSR: Thanks for calling [COMPANY_16], this is [PERSON_51], how can I help?",
    "CALLER: Kitchen sink is backing up into the other side and now the dishwasher won't drain.",
    "CSR: Any water coming up in a tub or shower, or just the kitchen?",
    "CALLER: Just the kitchen. The disposal hums but doesn't spin.",
    "CSR: I have a tech near you between 10 and 12 today. Is this still [ADDRESS_3]?"
  ],
  "call_intent": "new_service",
  "booking_note": "Kitchen sink backing into 2nd bowl, DW not draining, disposal humming and jammed. No other fixtures affected. Likely kitchen branch clog, bring small cable machine.",
  "job_id": "PL19-JOB-0781230",
  "job_type": "drain_cleaning",
  "priority": "same_day",
  "outcome": "booked"
}

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.
First look
One buyer gets 2 to 4 weeks to make an offer before anyone else sees the dataset.

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