SnowRock / Company profile
Pump service data: company profile
A complete worked example of the profile a buyer reviews before requesting a sample.
Illustrative example. The company and all dataset figures below are fictional. This is not an available dataset or a SnowRock client.
October 8, 2026
Company and proposed use
The example business is a regional industrial pump service company with an in-house repair shop and field technicians. Its customers use pumps in manufacturing and water-handling operations. It services multiple manufacturers rather than one product line.
The proposed buyer is developing an AI assistant that helps a technician locate relevant prior repairs. The proposed use is training and offline evaluation; automatic control of equipment is outside the scope.
Data inventory
| Data | Illustrative volume | Format and connection |
|---|---|---|
| Work orders | 8,400 closed jobs, 2021–2025 | CSV; one job_id per completed job |
| Service reports | 6,100 reports | PDF; linked by job_id and equipment family |
| Parts and labor | 31,000 line items | CSV; linked by job_id |
| Test results | 2,900 completed test sheets | PDF or CSV; linked by job_id |
| Customer messages | Excluded from initial sample | Rights and personal-data review required |
What a usable job contains
A connected example begins with the reported symptom, identifies the pump family and operating conditions, records what the technician checked, shows the repair and parts used, and ends with a test result or documented outcome. More files do not make an incomplete job complete.
| Field | Meaning | Treatment |
|---|---|---|
| job_id | Connects the files for one repair | Replace the source identifier with a stable sample identifier |
| equipment_family | Pump type and model family | Keep only the detail the buyer needs |
| symptom_text | Problem described before repair | Review for names, locations and customer details |
| diagnosis_and_action | Checks, finding and repair performed | Retain sequence; flag copied boilerplate |
| test_result | Measured or described outcome | Include test method and units when recorded |
Initial quality review
In this fictional example, the company reports that later jobs are easier to connect than older files. No audit has been completed. SnowRock would ask for a small, approved sample and calculate coverage before promising a delivery size.
- Count jobs with both a clear symptom and a recorded outcome.
- Check whether one job_id connects reports, parts and test sheets.
- Separate repeat visits from duplicated files.
- Flag scanned PDFs that need text extraction.
- Keep a record of exclusions rather than silently removing difficult cases.
Permissions and exclusions
The company’s owner and operations lead would identify who can approve a license. Customer contracts, manufacturer material, employee information and third-party service arrangements need review for the proposed use. A general nondisclosure agreement does not grant training rights.
- Do not include customer contact details, signatures or site-access instructions.
- Exclude drawings, manuals and other third-party content unless the rights permit the proposed use.
- Remove jobs whose customer terms prohibit the proposed sharing.
- Agree the sample’s permitted use, storage and deletion before delivery.
Proposed sample and acceptance
The illustrative first sample contains 30 approved jobs: 10 routine seal/bearing repairs, 10 diagnostic jobs and 10 repeat-visit or uncertain-outcome cases. The mix tests whether the data covers more than easy repairs. It is a proposed evaluation sample, not a delivery commitment.
- The buyer confirms that all 30 sample identifiers resolve to their supplied files.
- A reviewer can locate the symptom, action and available outcome for each job.
- Missing outcomes and uncertain diagnoses are explicitly marked.
- The sample includes a field dictionary and an exclusions list.
- The buyer returns specific gaps before either side discusses a larger delivery.
Commercial next step
If the sample fits, the company and buyer agree the license scope, permitted uses, delivery work, acceptance rules and payment schedule. The buyer pays the company for the license. SnowRock’s introduction or sourcing fee is separate and agreed in advance. No price is assigned to this fictional dataset.
Next action: hold a scoping call with the buyer’s data lead and the company’s authorized decision maker. Confirm the target task, sample permissions and who will prepare the files.