What records show how a company repaired a failed process?
Records that, read in order, show what went wrong, why, what was changed and whether the change worked. Each stage has a typical document, and the gaps between them are where most chains break.
Quality and reliability practice already names these stages. Google's Site Reliability Engineering book defines a postmortem as a written record of an incident, its impact, the actions taken to resolve it, the root causes and the follow-up actions to prevent recurrence. ASQ defines a root cause as the factor that caused a nonconformance and should be permanently removed through process improvement. A company that follows either discipline will usually have the middle of the chain. The ends, the raw failure records and the measured outcome, are what distinguish a complete repair from a report.
The annotated record chain
Illustrative: a fictional 340-person third-party logistics warehouse that repaired a rising mis-pick rate. The record names are generic; the systems are plain-text examples.
| Stage | Record | Typical system | What it proves | Common gap |
|---|---|---|---|---|
| 1. Initial error | Customer claims and mis-pick exception tickets, with timestamps | Helpdesk or warehouse management system | The failure, its frequency and who noticed | Tickets closed without a cause code |
| 2. Investigation | Root cause analysis linking mis-picks to similar SKUs in adjacent bins | Quality system or shared drive | Why it happened, with evidence | Analysis held in slides, not linked to tickets |
| 3. Intervention | Corrective action record, slotting change request, interim double-scan rule | CAPA log, change management in Jira or ServiceNow | What was changed, by whom, when | Interim fixes never recorded |
| 4. Revised policy | Picking procedure version 4 replacing version 3, with change history and training sign-offs | Document control or wiki | The new rule and when staff adopted it | Old version overwritten, no history |
| 5. Validated outcome | Effectiveness check: mis-pick rate for the 90 days before and after, closure approval | Reporting tool or quality system | That the fix worked, measured | Check skipped after the rate improved |
A reviewer reading the chain can follow one decision from symptom to verified result. That is different from a library of procedures, which shows only the current rule, and from a ticket archive, which shows only symptoms. The root cause analysis records page covers stage 2 in depth; this page is about the links between all five.
Why a repair chain is different from ordinary process data
It contains a counterfactual. The company tried one way, measured it failing, changed it and measured again. Most operating records describe a process that works; a repair chain describes the boundary where it stopped working and what moved it back.
AI data vendors describe related needs in their own terms. Appen, for example, lists trajectory analysis and failure-mode taxonomies among its agent evaluation services. That is a vendor's description of its offering, not evidence that any buyer wants a particular company's records, and it says nothing about price. Whether a given company's repair chains match current dataset demand is something only SourceX's review can answer. For how step-by-step records of correct work differ, see golden trajectories.
How to spot a complete chain in conversation
You do not need to see a single record. Ask these questions and note the answers in general terms:
- When something goes wrong repeatedly, where is it logged, and does each entry get a cause code?
- Who writes the analysis, and is it linked back to the original tickets?
- Are corrective actions tracked with an owner and a closure date?
- When a procedure changes, is the previous version kept with a change note?
- After a fix, does anyone check the metric again, and where is that check recorded?
- Roughly how many repairs like this happen in a year, and over how many years?
| Answer pattern | What it suggests | Note to make |
|---|---|---|
| Ticketing, CAPA log, document control and reporting all named | Likely complete chains | "Five-stage chains in named systems, about N repairs per year" |
| Analysis exists but lives in slides or email | Middle of the chain only | "RCA in documents, links to tickets unclear" |
| Procedures overwritten in place | Stage 4 missing history | "Current procedures only, no version trail" |
| No effectiveness checks | Outcome unverified | "Fixes made, results not measured" |
Teams that run on-call or incident response often keep the strongest chains; the incident and on-call field guide shows where to look, and quality assurance workflows covers inspection-led repairs.
Who owns the records and who can authorize a discussion?
Ownership usually splits by stage. Failure records sit with operations or customer service; analysis and corrective actions with quality or engineering; procedures with document control; outcome reports with finance or operations leadership. The COO or head of quality can often speak for the whole chain, but the CEO is the one who can authorize a licensing conversation and sign.
The company needs 50+ full-time employees at peak (contractors excluded), a documented operating history, rights to license the records and an authorized contact. Companies in the U.S., Canada and the EU qualify, as do companies in the UK, Australia, New Zealand, India, Singapore, South Africa, Latin America, the UAE and Saudi Arabia when at least 70% of the data is in English. Records that mainly describe customers' own data, or that customer contracts keep confidential, may not be licensable; the rights review decides.
What stays unknown until SourceX reviews it?
Whether there is current demand for the record types, what the rights review allows, how much de-identification is needed, what the company is offered, and therefore any reward. Partners never ask for or handle raw records; discovery stays at metadata level. For broader context on why companies are paid for operating data, see AI companies pay for data and the training data marketplace overview. More guides are on the guides hub.
If a license does complete, any reward is calculated on what SourceX actually pays the company. Illustrative: a $400,000 payout would mean a $70,000 reward; a $120,000 payout, $56,000. SourceX pays it separately, and it never reduces the company's payment.
Next step
Refer a company. No cap on rewards. Default reward on a company's first qualifying transaction: the greater of $50,000 plus 5% and 15% of Actual Supplier Payout, with no cap; further qualifying transactions first agreed within 12 months of the first Closing pay 15%. Eligibility and payment conditions apply, including the USD 100,000 Minimum Transaction, and a signed Schedule A or addendum can differ. A lead, meeting or signature alone does not trigger payment.
If an owner describes a repair they can trace from first ticket to measured result, register as a partner, check the company against who qualifies, and read the program terms before you introduce the executive who owns the chain.