What are CSR metrics?
CSR metrics are defined measures a company uses to understand and manage relevant social, environmental and governance aspects of its work. They may describe resources, activities, conditions, outcomes or stakeholder experiences. A useful metric has a clear purpose, definition, source, boundary and reporting period.
A number alone is not enough. “People reached” might mean unique people, event attendances or estimated audience exposure. Each can be useful for a particular question, but adding them together does not create a meaningful company-wide result.
This guide covers metric examples, a reusable dictionary, calculations, stakeholder input and reporting controls. It is for CSR, sustainability, operations and program teams that need a collection and review process they can maintain across business units and partners.
CSR metrics, KPIs and scores are different
Scroll horizontally to see all columns →
| Term | Meaning | Example |
|---|---|---|
| Metric | A defined measure relevant to the work | Employee participation in a specified volunteering program |
| KPI | A metric selected as important to a strategic or operational objective | The participation measure reviewed against an agreed internal target |
| Composite score | A combination of measures using stated rules and, where applicable, weights | An internal score combining several dimensions of program performance |
| External rating | An assessment under the provider's own methodology | A result that must be interpreted using that provider's definitions and scope |
Do not treat different external ratings as interchangeable, or assume your internal score reproduces them. A composite can hide weaknesses if one strong component offsets another weak one. Retain the component measures, missing-data rules and methodology so a reader can examine the result.
CSR metric examples across eight areas
These areas are a practical organizing list, not a universal reporting standard. Choose measures according to relevant impacts, business activities, stakeholder needs and applicable reporting requirements.
Scroll horizontally to see all columns →
| Area | Examples to consider | Possible sources and definition issues |
|---|---|---|
| Workforce | Representation, retention, development and employee experience | HR records and suitable feedback; define the workforce population and period |
| Community programs | Resources, participation, experience and relevant outcomes | Program and partner records; distinguish people from attendances |
| Environment | Energy, water, waste and greenhouse-gas measures | Operational sources and appropriate calculation methods; define boundaries and units |
| Partners and suppliers | Assessment coverage, corrective actions and relevant practices | Supplier information, documents and review; distinguish a reported action from verified implementation |
| Health and safety | Relevant incident and prevention measures | Appropriate operational records and defined reporting methods |
| Governance and ethics | Policy coverage, training and relevant process outcomes | Governance and operational records; completion alone does not establish effectiveness |
| Customers and products | Access, service experience and relevant product effects | Service records, suitable feedback and specialist evidence where required |
| Stakeholder perspectives | Reported priorities, concerns and experiences | Surveys, interviews and engagement records with clear coverage and context |
Neither system records nor stakeholder surveys are automatically the “easy” source. Meter boundaries, workforce definitions, supplier coverage and survey participation can all create substantial uncertainty. Plan the evidence appropriate to each measure instead of assuming one survey can supply every answer.
How to choose a manageable KPI set
- Start with the decision and relevant impacts. Identify what the company needs to understand or change.
- Review the required evidence. Distinguish internal management needs from a particular disclosure or external assessment.
- Include affected perspectives. Consider who experiences the activity and which voices the current sources omit.
- Balance activity and outcomes. Operational counts help explain delivery but should not substitute for evidence of the intended change.
- Check feasibility and burden. Use suitable existing information and identify where additional collection is justified.
- Assign ownership. Name the source owner, calculation owner and reviewer before the reporting deadline.
Do not choose one KPI per topic merely to make a symmetrical dashboard. Some questions need multiple measures; others need a qualitative explanation or further investigation. The impact strategy guide helps connect priorities to evidence and decisions.
Build a metric dictionary
A metric dictionary is a shared reference for what each measure means and how it is produced. It should be specific enough that two teams can apply the same definition, including what remains unknown.
Scroll horizontally to see all columns →
| Dictionary field | What to record |
|---|---|
| Name and purpose | The measure and the question or decision it supports |
| Definition and boundary | What is included, excluded and covered organizationally |
| Population and unit | People, organizations, events, quantities, percentages or qualitative categories |
| Calculation or assessment method | Formula, numerator and denominator where applicable, or the defined assessment approach |
| Period and frequency | The observation window and when the result is reviewed |
| Source and provenance | System, instrument, document, date and responsible contributor |
| Quality and uncertainty | Missingness, estimates, exclusions and validation steps |
| Version and ownership | Definition version, changes, owner and approval responsibilities |
Not every metric is a formula. A qualitative assessment still needs a clear method and evidence basis. A definition change should not silently rewrite historical results; retain the old version and explain whether a restated series is appropriate.
Worked example: compare volunteering participation
Fictional example. Two business units report participation in a defined volunteering program during the same quarter. Unit A reports 60 participating employees out of 100 eligible employees. Unit B reports 20 out of 25. Both use the same participation and eligibility definitions.
Scroll horizontally to see all columns →
| Unit | Participants | Eligible employees | Participation rate |
|---|---|---|---|
| A | 60 | 100 | 60% |
| B | 20 | 25 | 80% |
| Combined | 80 | 125 | 64% |
The company-wide participation rate is 80 ÷ 125 = 64%, not the unweighted average of 60% and 80%, which is 70%. If the same employees can appear in both units, the company must also resolve overlap before calling the total unique employees.
Participation is an activity measure. It does not establish the program's benefit to the community. That question may need partner perspectives, evidence of useful work completed and a suitable outcome assessment. A participant-satisfaction score answers another distinct question.
Collect stakeholder input for a defined purpose
Stakeholder evidence can identify experiences, concerns and questions that administrative records miss. Decide whose perspective is needed, how they can participate and how findings will be returned or used. Do not equate the largest response count with the most important impact.
Keep source, period and question context with the responses. Use identification only where appropriate; anonymous employee, community or supplier feedback may have a legitimate role. Traceability can mean an inspectable instrument and response set without exposing a named person.
For example, a supplier survey can report whether a practice is claimed to exist. It does not by itself prove the practice is implemented effectively. A review may require documents, observations or other evidence. The collection plan should make that distinction explicit.
How do reporting standards affect metric selection?
Start from the applicable framework or adopted requirement and its actual definitions. A generic CSR metric list is not a complete disclosure checklist, and a field labeled with a standard's name is not automatically aligned.
The GRI Universal Standards work alongside relevant Topic and Sector Standards. For ESRS work, EFRAG's implementation guidance includes materiality, value-chain and datapoint resources and explicitly identifies its guidance as non-authoritative. Check the applicable adopted version and requirements before planning a disclosure.
Do not assume every required measure comes from a stakeholder survey, that every response must be identifiable or that software makes a result disclosure-ready without review. Maintain a requirement-to-source mapping that shows the definition, source, calculation, gaps and responsible reviewer.
Allow local collection while protecting comparability
A business with several sites or partners may need different instruments. Agree a limited shared core for the comparisons that matter, then map local fields to that dictionary. Allow additional local questions rather than imposing irrelevant collection.
Retain organizational, site and reporting-period references where relevant. Stable registration context can be collected once and updated deliberately, while observations such as activity, conditions or outcomes remain dated. Keep incompatible units or definitions separate until a valid conversion or comparison has been established.
When a definition improves, document the change and decide how to handle the historical series. Recalculation may be possible, but only if the underlying evidence supports the new definition. Otherwise explain the break instead of inventing comparability.
Connect quantitative results with reviewed qualitative evidence
Comments and documents can help a team examine a pattern, understand reported experience and identify questions for follow-up. Keep their population and analytical scope clear. A theme among survey respondents is not automatically the reason for a company-wide trend.
Sopact supports connected collection, analysis and governance around relevant evidence. In codebook-based work, the team defines and reviews themes while automated processing applies them across eligible responses and supports reruns after revisions. Reviewers should be able to inspect the relevant source context and uncertainty.
The practical ownership argument is reduced repeated coding, joining and report reconstruction. Use the visual qualitative and quantitative workflow comparison and its illustrative staff-hours model to assess that work. Include configuration, validation and ongoing review in the estimate; specialist environmental or other calculations still need suitable methods.
Use the metric in a scorecard or report
Show the result with its unit, period, boundary, source and important limitation. Separate evidence readiness from performance against a target. A missing supplier update, for example, does not by itself establish poor supplier performance.
If a score combines measures, explain weighting, normalization and missing-data treatment. Keep the underlying results visible. Preserve the approved evidence version for a released report even when operational views continue to update.
Continue to CSR performance for the management process and CSR reporting for communication. Use How to Write an Impact Report and report examples for practical presentation guidance; these do not replace a required disclosure standard.
Watch: keep collection connected to context
This introduction explains the collection and analysis workflow. It is relevant to managing evidence, not a certification of compliance with a particular sustainability reporting requirement.
Frequently asked questions
What is the difference between a CSR metric and a KPI?
A metric is a defined measure. A KPI is a measure selected as particularly important to an objective or decision. Both need clear definitions and evidence.
What are examples of CSR metrics?
Examples include defined workforce, community-program, environmental, supplier, governance and stakeholder-experience measures. Select those relevant to the company's activities and requirements rather than adopting every possible indicator.
Does a shared framework make metrics comparable?
Not automatically. Check definitions, units, populations, periods, boundaries and calculation methods. A common heading can conceal materially different measures.
Should past results change when a definition changes?
Only through a documented decision supported by the underlying evidence. Preserve original definitions and reported versions, explain restatements and identify cases where a new series is needed.
Does stakeholder survey data make a CSR disclosure compliant?
No. Surveys are one possible source. The required evidence and disclosures depend on the applicable framework and requirements, and the result still needs appropriate review.
How does Sopact support CSR metrics?
Sopact supports connected collection, reviewed analysis and governance around relevant records and definitions. Test the required source mapping, revisions, access and reporting workflow; do not assume automatic compliance or specialist calculation capabilities.
Choose your next step
American Heart Association Ventures
Related practice: connecting plans and reported evidence across a portfolio. This is not a CSR compliance certification or validation of the fictional calculations.
Read the customer story →