Definition
Degree to which AI-curated performance data in management reviews represents the full scope of management system performance, measured by comparing AI-selected content against a comprehensive performance data set required by Clause 9.1.
The Measurement Gap
ISO 55001 Clause 9.1 requires organizations to determine what needs to be monitored and measured, including the methods and the frequency. Clause 9.3 requires that management review include consideration of the organization's performance, including trends in nonconformities, monitoring and measurement results, audit results, and the performance of external providers. When AI-powered reporting and analytics tools curate the data presented in management reviews, they select indicators based on variance, relevance scoring, or predictive significance. This selection process filters the available performance data into a subset that the tool determines is most important. The failure mode is representativeness bias: AI reporting tools trained on operational data naturally overrepresent asset performance indicators (uptime, cost, reliability) and underrepresent management system effectiveness indicators (governance compliance, competence adequacy, documentation integrity, stakeholder satisfaction). This bias is structural, not malicious. Operational data is more readily available, more frequently updated, and more directly tied to the optimization objectives that AI tools are designed to serve. Legacy management system metrics such as management review completion rates and action item close-out percentages measure whether reviews happened and whether actions were tracked, but they cannot detect that the data presented in those reviews was algorithmically filtered to exclude entire performance categories. The consequence is management making strategic decisions based on an incomplete and systematically biased picture of organizational performance.
Decision Supported:
Is the management review seeing a representative picture of full management system performance, or is AI-curated reporting filtering out categories of performance data that the standard requires?
Formula
Plain Language: Define the complete set of performance data categories corresponding to the required management review inputs specified in Clause 9.3 of the applicable standard. Count how many of those categories have at least one indicator present in the AI-curated management review. Divide by the total required categories.
Notation:
RRS = (Categories_represented / Categories_required) x 100
Companion metric: RRS_depth = (1/C) x SUM(Indicators_presented_c / Indicators_available_c) for c = 1 to C, measuring depth of coverage within each represented category.
Variables:
| Variable | Definition | Unit |
|---|---|---|
| Categories_required | Number of performance data categories corresponding to the required management review inputs specified in Clause 9.3 of the applicable standard. For ISO 55001, Clause 9.3 requires consideration of: status of previous actions, changes in external and internal issues, asset performance information, nonconformities and corrective actions, monitoring and measurement results, audit results, and risks and opportunities. ISO 9001, 14001, and 45001 each specify equivalent required inputs in their respective Clause 9.3. The organization documents its complete category set by mapping the applicable standard's required review inputs. | Count |
| Categories_represented | Number of required categories that have at least one performance indicator present in the AI-curated management review content. | Count |
| C | Total number of represented categories (used in RRS_depth companion metric only). | Count |
| Indicators_presented_c | Number of indicators in category c that appear in the AI-curated management review. | Count |
| Indicators_available_c | Total number of indicators available in category c from all data sources. | Count |
Sign Convention: RRS and RRS_depth are both percentages from 0% to 100%. Higher values indicate more representative reporting. RRS measures breadth (are all categories present?) and RRS_depth measures depth (within each category, how much of the available data is presented?).
Proposed formula, pending validation pilots.