the job description is the wrong unit of analysis
job titles average away the variation that matters, and the decisions enterprises now face need a smaller unit than the job.
A board asks for a workforce report. HR produces a deck: engagement scores, a diversity chart, an attrition trend and a slide on culture. The audit committee cannot use it, and usually says so politely by asking for it again next quarter. The gap is not effort. A board report has to survive challenge from people whose job is to challenge numbers, and most workforce reporting was never built for that.
Two things changed recently that make this worth revisiting. ISO 30414 became a requirements standard in August 2025. The European Union finished narrowing CSRD in February 2026. Both change what a defensible disclosure set looks like.
what ISO 30414 requires
The 2018 edition was titled "Guidelines for internal and external human capital reporting". It set out 11 human capital areas and 58 metrics, and it was guidance. Nothing in it was mandatory and nothing in it was auditable, which is why adoption stayed thin outside a small group of German and Japanese firms.
The second edition, published on 24 August 2025, is titled "Requirements and recommendations for human capital reporting and disclosure". ISO's own technical committee describes the change as a move from guidance to compliance-ready reporting, with auditable requirements sitting alongside recommendations. It keeps the 11 human capital areas and introduces a baseline set of 14 metrics that apply to every reporting organisation regardless of size, sector or geography. It extends coverage to human rights, labour relations, productivity, ethics and workforce wellbeing, and adds guidance on data governance, AI and digital tagging (ISO/TC 260, 2025).
Be clear about what the standard does not do. It does not tell you what a good number looks like. A turnover rate of 18 per cent is neither compliant nor non-compliant. The standard fixes the definition and the disclosure obligation, not the target. That is where its value sits. Comparability comes from shared definitions, and most workforce metrics fail on definition long before they fail on data.
what CSRD now requires
The European position moved a long way through 2025 and 2026, and much of the published guidance is out of date.
The Omnibus I Directive was approved by the Council on 24 February 2026, published in the Official Journal on 26 February 2026 and entered into force on 18 March 2026. CSRD now applies to companies with more than 1,000 employees and net turnover above 450 million euros. Norton Rose Fulbright puts the effect at roughly 80 per cent of previously in-scope entities falling out. The threshold for non-EU parent groups rose to 200 million euros. Sector-specific standards were dropped. The planned move from limited assurance to reasonable assurance was dropped as well, so limited assurance remains the requirement (Council of the European Union, 2026; Norton Rose Fulbright, 2026).
On the standards themselves, the European Commission published the delegated act on the revised ESRS on 3 July 2026. Mandatory datapoints fall by more than 60 per cent. The revised standards apply to financial years beginning on or after 1 January 2027, with early adoption permitted for financial year 2026 once the act is in force (EFRAG, 2026).
ESRS S1, the own-workforce standard, still applies where the topic is material under a double materiality assessment. The metrics that survived the cut are the ones with the strongest comparability properties: workforce characteristics by contract type and gender, turnover, collective bargaining coverage, remuneration measures including the gender pay gap, training and skills development, and health and safety.
For a company headquartered in Singapore, Dubai, Riyadh or Mumbai, the direct obligation may not apply. The indirect one usually does. If you supply a European customer that remains in scope, you will be asked for workforce data on their timetable and in their format. The directive also caps what in-scope companies may demand from suppliers with fewer than 1,000 employees, which is worth knowing before the request arrives.
which metrics survive scrutiny
Here is the test an auditor applies, whether or not they say it out loud. A metric survives if five things are true.
- It has a written definition naming the numerator, the denominator and the population, including who is excluded.
- It comes from a system of record, not from a spreadsheet assembled for the report.
- A second person can reproduce it from raw data without asking the first person anything.
- It has a stated restatement policy, so a change in definition is visible rather than silent.
- It has a prior-period comparative calculated on the same definition.
Most workforce metrics fail on the first and the fifth. Headcount is the clearest case. Almost every organisation reports it, and almost every organisation reports a different thing. Some include contractors, some include open requisitions, some use period-end and some use a period average, and the basis quietly changes whenever the HR system does.
The published evidence supports the pessimism. Gibson Dunn reviewed human capital disclosures in S&P 100 annual filings and found that in 2023 only 20 per cent gave a specific employee turnover rate, and 14 per cent gave quantitative talent development figures such as training hours or spend. Although 42 per cent discussed pay equity practices, only 17 per cent disclosed a quantitative pay gap (Gibson Dunn, 2023). These are the largest listed companies in the United States, three years into a mandatory disclosure requirement. That requirement is principles-based, adopted in 2020, and the resulting lack of comparability was the objection raised at the time by two of the regulator's own commissioners (Harvard Law School Forum on Corporate Governance, 2021).
a three-tier disclosure set
Build the report in three tiers and be explicit about which tier each number belongs to. This is the part worth taking away.
Tier one, the assured spine. Eight to twelve metrics held to the same control standard as financial data. Headcount by employment type and geography on a stated basis. Joiners and leavers with a defined turnover formula. Voluntary and involuntary separations reported separately. Total workforce cost. Gender distribution by management level. Gender pay gap, mean and median, with the population defined. Recordable work-related injuries and days lost. Collective bargaining coverage. These are the numbers you take into the audit committee and defend line by line.
Tier two, the managed set. Metrics with named owners, written definitions and a source system, but no external assurance. Internal fill rate, days a critical role stays vacant, time to productivity, training investment per full-time equivalent, regretted attrition, span of control. These support management decisions and inform the board without carrying the same evidentiary weight.
Tier three, the narrative. Engagement results, culture commentary, programme descriptions and strategy. Present these as context. Never let a tier three number carry the primary claim in a board paper, because the method is self-reported, the responding population is self-selected and the result is not comparable to any other organisation.
One rule governs movement between tiers. Nothing moves up until it has a definition, an owner, a source system and two consecutive periods reported on the same basis. Metrics move up only. Moving one down requires a written explanation in the paper, which is a useful piece of friction.
four questions to prepare for
Before the meeting, write out the answers to four questions for every tier one metric. Where does this number come from, and who has the ability to change it. What population does it cover, and who is excluded. What was the figure last period on this exact definition. What would cause you to restate it.
If you can answer those four for eight metrics, you have a board-ready report. If you cannot answer them for any, you have a deck.
how to package it
Keep the board page to one page with three columns: the number, the same number a year ago on the same definition, and the decision it informs. Add a short list of what you do not measure and why. Naming the gaps is the strongest credibility signal available, because it tells the audit committee you know where the edges of your data are. Committees distrust completeness far more than they distrust a stated limitation.
Most of the burden here is a definitions burden, and definitions live in systems. Getting a tier one metric assured means enforcing the definition where the data is captured, rather than reconstructing it at quarter end from four extracts and a spreadsheet. That is the work platforms such as entomo's workforce analytics exist to do, and it is unglamorous work that pays off only under challenge.
The bar is not high. A board wants a small number of metrics it can trust, on a consistent basis, with the gaps named honestly. Most workforce reporting fails that test by trying to show everything.