Key takeaways
- Matching text is a candidate signal, not proof of identity.
- Source values and merge decisions should remain auditable.
- Event records require different tests from people or property identities.
Source record
5 cited sources
Last verified
2026-09-04
Table of Contents
Duplicate records make rental portfolio reporting unreliable, but an aggressive merge can be worse than a duplicate. Two work orders with similar language may describe a recurring failure. Two vendor names may refer to related but legally distinct entities. This review asks what evidence is strong enough to link records without erasing meaningful differences.
Research design and scope
The review compares five public sources covering data quality, identity resolution, record linkage, internal control, and privacy. Sources were checked September 4, 2026. The resulting framework applies general evidence principles to rental operations. It is not a statistical validation of a particular software product and does not establish legal identity, ownership, tenancy, or liability.
Four dimensions of a defensible match
Identity evidence answers who or what the record concerns. Context evidence covers property, unit, account, contract, or workflow. Time evidence asks whether records coexist or describe a sequence. Provenance identifies the system or person that supplied each value. A strong match uses several dimensions rather than a single normalized name.
| Record pair | Useful evidence | Reason to pause |
|---|---|---|
| Property to property | Internal ID, verified address, unit structure | Similar street text alone |
| Vendor to vendor | Legal name, approved vendor ID, tax record reference | Shared phone number alone |
| Work order to work order | Unit, location, symptom, time, linked repair | Repeated keywords |
| Person to person | Approved unique identifier and verified source | Name and email similarity |
Normalization should preserve original source text. Converting “Street” and “St.” to a common comparison value can surface candidates, but the source values still matter for contracts, mail, and reconciliation. The merge record should show the reviewed pair, evidence, decision, reviewer, and date.
The property identity resolution research examines property-specific evidence. The property address normalization review turns the concept into a daily control.
Why events need a different rule
Entities persist; events occur. A later maintenance report about the same unit and symptom can be a duplicate intake, a continuation, or a new failure. Time, prior closure, changed condition, and new evidence determine the relationship. Systems should link related events without forcing them into one record.
Automated similarity scores can prioritize review. They should not silently merge high-consequence records. Threshold performance also depends on the tested data. A rule tuned on clean addresses may perform poorly on unit identifiers or multilingual names.
Limitations
The sources describe general controls and data practices. They do not provide a universal matching threshold for rental portfolios. Available identifiers, local rules, system design, error costs, and record sensitivity vary. Manual review can also make mistakes, so sampling and reversible decisions remain necessary.
Evidence-led conclusion
A defensible duplicate decision connects identity, context, time, and provenance. It retains original values, records the reviewer, and allows a mistaken merge to be reversed. For rental operations, the most important distinction is between linking related records and deleting one of them.
Published September 4, 2026.
Sources and verification dates
- National Institute of Standards and Technology, Digital Identity Guidelines, checked September 4, 2026.
- U.S. Census Bureau Statistical Quality Standards, checked September 4, 2026.
- National Archives, records management guidance, checked September 4, 2026.
- GAO Green Book, checked September 4, 2026.
- FTC business privacy and security guidance, checked September 4, 2026.