Disclosure focus
The practical focus is on how broadly training is covered across the organisation. A useful question is whether the average reflects all employees and all relevant parts of the business, or only selected sites, functions, or programmes. The reporting should make clear the scope used so readers can judge whether the number represents training across operations or only a more limited set of activities.
This LRA educational guidance supports disclosure preparation. For the exact requirements, always refer to the official GRI source.
Prepare
What to collect before drafting
Tick each field as the evidence lands. Owners are the teams that usually hold the source files.
-
Gender split
HR / People AnalyticsCapture the gender categories used for the workforce disclosure, using the same definitions and labels as the underlying people data.
Evidence HRIS workforce extract, diversity reporting definitions, and any self-identification guidance used to classify staff.
Watch for Categories do not align with employees’ self-identification or HRIS codes, so the reported split cannot be reconciled to the source population.
-
Worker category
HR / People OperationsCapture the employee groupings used for reporting, with each person assigned to the correct workforce category under the organisation’s own classification rules.
Evidence HRIS employee master data, workforce segmentation rules, and payroll or contract records used to assign categories.
Watch for A definition mismatch means contractors, fixed-term staff, or other groups are coded differently in HRIS and the disclosure, so the category totals do not reconcile.
-
Training hours average
L&D / HR AnalyticsCapture the average number of training hours completed by employees during the reporting period, based on the same employee set and time window used in the training records.
Evidence Learning management system reports, training attendance logs, and the calculation workbook showing the average and the employee count used.
Watch for Period and population are mixed, so the average is calculated from a different employee set or time window than the training source records.
Before you start
- Have you decided which staff groups you will include, and whether you are splitting the figures by gender and by job level or type?
- Are you using the same reporting period for everyone, and is the training count based on that period only?
- Can you trace the training hours back to a source record, such as attendance logs, learning system data, or signed records?
- Have you fixed what counts as training in your organisation, so the same definition is used across all teams and locations?
- Will the average be worked out from employee-level training hours, rather than from a team total or another proxy?
How to prepare it
Request the data
Request the training-hours data from People Analytics
Translate the disclosure into an internal business question — then adapt it to your organisation's own language.
What training time did each employee group record during the reporting period, split by gender and employee category?
Use your organisation’s own labels first, then map them to the reporting categories. For example, if you use job family, grade, population, or workforce segment internally, ask for those terms and translate them later into the disclosure wording. Keep the request in the language the data owner already uses.
Weak request
Please provide the GRI 404-1 data showing the evidence needed for GRI 404:GRI 404-1, split by gender and employee category.
Why it fails: It uses framework language rather than the organisation’s own terms, and it does not tell the owner which system, population, grouping labels, or calculation basis to use. That makes the response harder to prepare and harder to check.
Better request
Please send the training-hours extract for [period] from [system], using your usual workforce group labels. Include gender, employee category, total training hours, average hours per employee, the population covered, the calculation method, and any exclusions or assumptions. We will map your labels later. This is a possible LRA training template only; please adapt it to your organisation and check the official source before sign-off.
Formal email template
Subject: Request for training-hours data for sustainability reporting Hi [Name], Could you please share the training-hours extract for [reporting period] from [source system], using the workforce groups you normally report internally? We need the data broken down by: - gender - employee category / workforce segment - total training hours recorded in the period - average hours per employee for each group Please also include: - the population covered - the definition of training used in the extract - the calculation method - the extract date and source system - any exclusions, assumptions, or data quality notes If your team uses different internal labels, please send those as-is and we will map them later. This is a possible LRA training template only; please adapt it to your organisation and check the official source before sign-off. Thanks, [Your name]
Short Teams / Slack version
Hi [Name] — could you send the training-hours extract for [period] from [system], split by your usual workforce groups plus gender? Please include the method, population covered, exclusions, and extract date. We’ll map your internal labels later. This is a possible LRA training template only; please adapt it to your organisation and check the official source before sign-off.
Industry examples
Retail
Context. A store-based workforce with full-time, part-time, and seasonal staff recorded in an LMS and HR system.
Adapted request. Please send the learning-hours extract for [period] from [LMS/HR system], split by your usual staff groups such as store, warehouse, and head office, plus gender. Include total learning hours, average hours per person, the population covered, and any exclusions such as agency staff or one-off induction sessions. This is a possible LRA training template only; please adapt it to your organisation and check the official source before sign-off.
Example response. Returned table shows: gender; staff group; headcount; total learning hours; average hours per person. Notes state that agency staff were excluded and that induction was included for new starters only.
Manufacturing
Context. A site-based workforce with production, maintenance, and office teams, where training is tracked through a learning platform and local spreadsheets.
Adapted request. Please provide the training-time extract for [period] from [system], using the site’s normal workforce categories such as production, maintenance, and support teams. Split by gender and include total hours, average hours per employee, the calculation method, and any items not counted, such as toolbox talks or external conferences. This is a possible LRA training template only; please adapt it to your organisation and check the official source before sign-off.
Example response. Returned table shows: gender; workforce category; headcount; total training hours; average hours per employee. Notes explain that toolbox talks were excluded and that only completed courses in the learning platform were counted.
Draft your disclosure
Notes that turn data into a disclosure
LRA training templates — adapt them to your organisation, and check the official source before sign-off.
Method note
Define the employee groups and gender categories used, and state that the figures show the average training time per person during the reporting period.
Context note
These figures show how much learning time different parts of the workforce received on average, helping readers see whether development time is spread evenly or concentrated in particular groups.
Fluctuation statement
If the averages moved materially from the prior period, explain whether this was driven by changes in headcount mix, training availability, programme design, or other operational factors.
Content index entry
GRI 404-1 Average hours of training per year per employee — [location / page] / [notes]Download Centre
Preparation tools & forms
Download preparation tools for GRI 404-1 with free verified email access for 24 hours. Source-report links remain public.
Check
For each claim, check the evidence
We built the coverage figure from the employee records we held for the reporting year, and we split the workforce by sex and by job group using our internal HR coding.
Risk The assurer may test whether the grouping logic was applied consistently, whether any staff were left out, and whether the categories used in the published figure match the source records.
Evidence to check HR headcount extract for the reporting period; workforce coding guide or data dictionary showing sex and job-group fields; reconciliation from source records to the published figure; sample employee files or system screenshots confirming the assigned codes.
For the disclosed workforce breakdown, we used the same cut-off date and the same employee population across all tables, and we excluded only people outside the defined reporting boundary.
Risk The assurer may probe boundary decisions, duplicate counting, treatment of leavers and joiners, and whether the same population was used consistently across the disclosure.
Evidence to check Reporting boundary memo; population definition used for the disclosure; dated headcount report; joiner/leaver logs; reconciliation showing how the final population was derived; evidence of any exclusions and the reason for them.
We calculated the training-hours figure from attendance logs and learning-system records, then checked the total and the average before publication.
Risk The assurer may question whether the underlying hours were complete and accurate, whether the average was calculated correctly, and whether the published number was rounded or adjusted without support.
Evidence to check Training attendance records; learning-management-system export; calculation workbook showing total hours and average; rounding policy or calculation note; review sign-off from the preparer and checker; exception log for missing or corrected records.
Where records were incomplete, we used documented estimates or replacements only after review, and we kept a clear audit trail from the original source to the final figure.
Risk The assurer may test whether estimates were justified, whether substitutions were applied consistently, and whether the audit trail is strong enough to support the published result.
Evidence to check Data-quality issue log; estimate methodology or replacement rule; approval evidence for any manual adjustments; version history of the calculation file; source-to-report traceability schedule.
Before release, we ran a final consistency check against prior-period data, internal management reports, and the narrative in the report so the published figure agreed with the supporting schedules.
Risk The assurer may look for unexplained movements, mismatches between tables and narrative, and weak review controls before sign-off.
Evidence to check Pre-publication review checklist; variance analysis versus prior period and management reporting; final proof of the report; sign-off emails or approval form; evidence of any corrections made after review.
Evidence pack to prepare
Common reporting gaps
- The information is presented without a date or as-at point.
- The scope or boundary of the statement is left undefined.
- Key terms are used inconsistently across the report.
- Material changes since the previous period are not disclosed.
- Assertions are made without supporting detail or a source record.
- Boilerplate is used that does not actually answer what is asked.
Where drafts go wrong
Mistakes to avoid and where judgement is needed
The left column is what preparers get wrong while collecting; the right column is where you will have to make and explain a judgement.
Mistakes to avoid
- Wrong ownerChasing the learning team alone can miss the payroll, HRIS, or local people lead who actually holds the headcount and training records.
- Framework language firstAsking for the metric in reporting jargon can confuse the business owner, so they send a different training summary than the one used internally.
- Scope left vagueIf you do not pin down which worker groups are in or out, teams may include contractors, interns, or leavers by mistake.
- Period basis mixed upUsing a calendar-year training log for a financial-year disclosure can pull in the wrong months and distort the result.
- Counting basis not alignedCombining course hours, booked hours, and completed hours in one file can make the average inconsistent across sites or systems.
- Source labels stripped outCopying figures into a new sheet without the original field names or system tags makes it hard to trace what each number came from.
- Groups merged too earlyRolling all staff into one pool before separating the agreed people groups can hide differences between categories that should stay apart.
- Evidence details missingSaving only the final number, without the extract date, source file, or owner, leaves no way to check how the figure was built.
- No sign-off trailIf the draft is not reviewed by the data owner before submission, errors can pass through without anyone confirming the numbers.
Judgement needed
- Set the people base before you calculate the averageDecide which workers sit in the headcount for the period, explain any inclusions or exclusions at the margins, and keep the same basis across the training total and the people count.
- Handle joiners and leavers on a consistent time basisChoose whether to count only time actually employed in the period or to annualise part-year service, then state that approach so the average is not distorted by mid-year starts or exits.
- Agree one training-hour definition across countries and functionsWhere local records treat learning time differently, map those differences to one group-wide rule, describe the conversion used, and note any country-specific exceptions.
- Decide how to treat outsourced, agency, and other non-standard workersIf some labour sits close to the workforce boundary, explain whether those people are inside or outside the calculation and why that treatment matches your reporting scope.
- Use a clear rule for acquisitions, disposals, and restructuresState whether the period includes only the business as owned for part of the year or a full-year view for acquired or sold operations, and explain any restatements or exclusions.
- Choose measured data or a reasonable estimate where records are incompleteIf learning logs are missing or inconsistent, disclose the estimation method, the source data used, and the extent of any judgement applied.
- Round in a way that does not change the storySet and disclose a rounding rule for hours and averages, and make sure the rounded figures still reconcile sensibly to the underlying totals.
- Protect personal data by grouping small populationsWhere a small team or a narrow category could identify individuals, combine groups or suppress detail as needed, and explain the aggregation approach used.
Examples
How it reads on the page
Synthetic, written by LRA — not from a company report, not text from any standard.
During the year, we tracked learning time by gender and by staff group, and we report the average number of training hours per person for each slice.
- Women: managers 18 hours, specialists 14 hours, support staff 10 hours.
- Men: managers 16 hours, specialists 12 hours, support staff 9 hours.
This example shows how a reporter can break learning time down by gender and employee group, while keeping the figures internally consistent and clearly labelled as illustrative.
We measured development time across our workforce and present the mean hours per employee, split by gender and job level.
- Women: senior leaders 22 hours, supervisors 15 hours, production operatives 11 hours.
- Men: senior leaders 20 hours, supervisors 13 hours, production operatives 10 hours.
This example demonstrates a second plausible way to present the same information, using a different sector and different staff categories while still showing the average training time by gender.
Quality ladder
- Basic
We averaged 12 training hours per employee this year.
- Better
We averaged 12 training hours per employee this year, split by gender and staff group.
- Best
We averaged 12 training hours per employee in the reporting period, split by gender and staff group, and the figure was higher than last year because we added more role-based learning.
Check your understanding
Scenarios to work through
A preparer is compiling the learning and development note for a group with three staff segments: senior managers, office staff, and field staff. The training log shows different totals for women and men in each segment, and some people joined part-way through the year.
A company has a mix of permanent employees, fixed-term staff, and agency workers. The HR system records training for everyone, but the reporting team is unsure whether to combine all of them into one average.
The training register shows 120 employees in one category. Ninety employees each completed 8 hours of training, while 30 employees completed none. The team is unsure whether the zero-hour group should be left out because it makes the average look lower.
A preparer has separate spreadsheets for women and men, but one category has only two employees and the other has 200. The team wonders whether it is acceptable to report only the larger group because the smaller one is easy to identify.
Real reports
How 3 companies report it
Examples of full and partial reporting practice. These are evidence-led reviews, not exact disclosure templates to copy.
Indra Sistemas, S.A.
Partial
Firstsource Solutions Limited
Partial
Temenos AG
Partial| Field | Indra Sistemas, S.A. | Firstsource Solutions Limited | Temenos AG |
|---|---|---|---|
| Gender split | Not found | Not found | Not found |
| Worker category | Not found | Not found | Not found |
| Training hours average | Not found | Not found | Not found |
Ask the Knowledge Hub AI Assistant about this disclosure
Get practical answers for your reporting context.
Check your text
Paste one or two paragraphs. The fragment is used for this check only and is not stored in your conversation.
Peer comparison uses examples on this card, not prevalence in the report corpus.
Advice, not assurance
Framework references
Relevant GRI requirements and related disclosures
GRI
GRI 404-1
within GRI 404: Training and Education
Related & explore
More in GRI 404 →Browse full catalogue →Search all disclosures →
FAQ
Questions this page answers
The page says to prepare three core datapoints: gender split, worker category and average training hours. Use those as the starting set for your data request and check they are available for the same reporting period and scope.
Use it as a working checklist to move from the plain-language explainer to the datapoints, then to methodology, evidence and draft output. It is designed to help you prepare the disclosure rather than just describe it.
The page is set up to help you define what population is included, how worker categories are treated and how the average training hours are calculated. Keep those choices consistent across the datapoints and explain them clearly in the draft.
The page is aimed at sustainability/ESG managers, HR or data owners, and assurance reviewers, so ownership should sit with the person who can confirm the source data and methodology. In practice, that usually means one named owner for the data and one reviewer for assurance readiness.
The page includes an evidence pack with five items and five assurance claims to verify. Use those to show the source data, the calculation approach and the checks that support the final figures.
The page says there are five claims to verify, each with a claim, risk and evidence prompt. Use them as a control list to test whether the disclosure is supported, where it could go wrong and what documents prove it.
The page lists common reporting gaps and mistakes so you can check for missing scope, unclear methodology or weak evidence before drafting. Use that section as a pre-submission quality check.
The page includes draft-output support with visualisation ideas, narrative starters and a GRI content-index line. Use those to turn the prepared data into a short, readable draft and then tailor the wording to your organisation.
The Download Centre includes a Prep & Assurance workbook in .xlsx format. Use it to organise the datapoints, evidence and assurance checks before you finalise the draft.
The Download Centre also includes a printable Library Card in .pdf format. It is a quick reference aid for the disclosure, useful when you want the key points in one place while you work through the data and evidence.
The page notes ESRS S1 (Own Workforce) as the closest correspondence, so the same underlying data may be reusable. Treat that as a practical link rather than assuming the reporting asks are identical.
More questions this page can help with