Start with the decisions the data must support.
Define whether the immediate job is work dispatch, PM control, failure trending, cost reporting, migration, or asset renewal. The minimum useful fields differ. A register built only for accounting may not contain maintainable-item hierarchy; a work order adequate for billing may not contain a failure mode.
Separate completeness, validity, consistency, and accuracy.
A field can be populated but invalid, valid but inconsistent with the data dictionary, or internally consistent but physically wrong. Use the register completeness checker only as a first screen, then test uniqueness, accepted values, relationships, dates, and a sample against tags, drawings, manuals, and field observations.
Keep identity stable and hierarchy changeable.
Use the asset ID builder to draft readable codes from stable components. Do not encode owner, condition, or temporary process state. Maintain parent-child relationships in dedicated fields so reorganizing a hierarchy does not require changing every identifier.
Capture the event at closeout.
The IAEA notes that maintenance work orders and equipment test records are useful failure-data sources, while incomplete component identification, failure detail, and timing degrade the resulting database. Review a sample with the closeout quality checker, then show technicians how the information changes planning or reliability decisions.
Separate problem, cause, and remedy.
One dropdown cannot reliably answer what was observed, why it happened, and what was done. Draft three controlled lists with the failure code taxonomy builder. Include a governed unknown value instead of forcing an unsupported cause, define each code, pilot it on historical work, and retain mappings when the taxonomy changes.
Treat migration as a controlled engineering change.
Preserve a source export, approve field mappings, test representative edge cases, reconcile counts and critical fields, sample dates and open work, establish rollback, and name the acceptance owner. The migration readiness planner turns these gates into a phased remediation list; it does not replace system validation or local change control.
Run a repeatable assurance loop.
- Declare the population, required fields, accepted values, and owner.
- Profile completeness and validity without altering the source.
- Prioritize gaps that block current decisions.
- Correct records with traceable change control.
- Sample against independent evidence.
- Measure recurrence and revise forms, lists, and training.
The UK Government FM Asset Data Standard likewise calls for asset-register structure, regular sampling, change control, completeness and consistency checks, documentation, ownership, and active use of the data.
Public sources
- UK Government, Facilities Management Standard 002: Asset Data — public requirements and guidance on asset-register structure, data quality, assurance, ownership, and use.
- IAEA TECDOC 1922 — public discussion of equipment reliability data sources and raw-data quality problems.
- U.S. Defense Logistics Agency, MIL-HDBK-2155 record — public FRACAS scope emphasizing disciplined use of failure and maintenance data for corrective action.
These sources establish why data structure and assurance matter. ReliabilityBench scoring thresholds and draft fields are declared screening conventions, not claims of compliance with any source.