What are CSR metrics?
CSR metrics are defined measures a company uses to understand and manage the social, environmental and governance effects of its work, from its own workforce and operations to the community programs its foundation funds. A useful metric names its purpose, definition, source, boundary and reporting period.
This guide is for CSR and corporate foundation teams that fund community partners (grantees) and need their results to add up. The trouble is rarely a missing metric; it is one word counted three ways. “People reached” can mean unique people, attendances or estimated audience, and adding them together produces a number nobody can defend.
THE SHORT VERSION
- Spend and volunteer hours measure what the company put in; add at least one metric for what changed for the people your grants reach.
- Write every shared metric once in a dictionary and give it to every grantee before the first report, so a number from one partner means the same as a number from another.
- Add up only the numbers that share a definition, and show the rest beside the total with the question you have asked.
What is the difference between a CSR metric, a KPI and a score?
A metric is any defined measure; a KPI is a metric chosen because it matters to a specific objective; a score combines several measures under stated rules. External ratings sit apart, because each provider applies its own method.
Scroll horizontally to see all columns →
| Term | Meaning | Example |
|---|---|---|
| Metric | A defined measure relevant to the work | Grantee trainees placed in a job within 90 days of exit |
| KPI | A metric chosen for an objective | That placement count, reviewed against the grant’s goal |
| Composite score | Measures combined by stated rules and weights | An internal grantee score across several metrics |
| External rating | An assessment under a provider’s own method | Read only with that provider’s definitions and scope |
External ratings are not interchangeable. A composite can let one strong component hide a weak one, so keep its parts and missing-data rules visible.
Why aren’t grant spend and volunteer hours enough?
Spend and volunteer hours describe what the company gave, not what changed, so a CSR report built only on them cannot answer the board’s real question: did the money help anyone? They belong in the report as inputs, next to at least one outcome.
The course’s running example shows the ladder. Picture a fictional regional workforce fund, standing in for a company foundation’s workforce program, that funds four job-training partners, A to D. Its agreement with them names five metrics, and only the last three say anything about change.
Scroll horizontally to see all columns →
| Level | CSR example | What it answers |
|---|---|---|
| Input | Dollars granted; employee volunteer hours | What the company gave |
| Output | Participants enrolled (unique people); completed training | What grantees delivered |
| Outcome | Placed in a job within 90 days of exit | What changed for people |
| Outcome, lasting | Same job at 12 months; starting wage by track | Whether the change held and how deep it was |
A useful test for any metric on your list: if the number doubled, would anyone’s life be better? Volunteer hours can double with no change for anyone; placements that last a year cannot. For the full distinction, see output vs outcome.
What are examples of CSR metrics?
CSR metrics usually fall into eight areas, from workforce and environment to community programs and stakeholder perspectives, and each area has its own definition traps. Treat the list as an organizing aid, not a reporting standard; choose by your company’s actual effects and requirements.
Scroll horizontally to see all columns →
| Area | Examples to consider | Definition trap |
|---|---|---|
| Workforce | Representation, retention, development, employee experience | Which workforce, which period |
| Community programs | Grant dollars, participants, outcomes for people served | People versus attendances |
| Environment | Energy, water, waste, greenhouse-gas measures | Boundaries, units, calculation method |
| Partners and suppliers | Assessment coverage, corrective actions, practices | A reported action is not a verified one |
| Health and safety | Incident and prevention measures | Defined reporting method |
| Governance and ethics | Policy coverage, training, process results | Completion is not effectiveness |
| Customers and products | Access, service experience, product effects | Specialist evidence where needed |
| Stakeholder perspectives | Priorities, concerns, experiences | Who answered, and who did not |
What should a CSR metric dictionary for grantees include?
A metric dictionary is one shared page per metric that says who counts, what qualifies, when, from which source and how it may be added up, written by the funder and sent to every grantee before the first report. It is what lets a CSR team add partner numbers without guessing.
Scroll horizontally to see all columns →
| Dictionary field | What to record |
|---|---|
| Name and purpose | The measure and the decision it supports |
| Definition and boundary | Who or what is included and excluded |
| Population and unit | People, organizations, events; count or percentage |
| Calculation | Formula, numerator and denominator |
| Period and collection point | The window and the form the value comes from |
| Dimension and standard | Which of the Five Dimensions it answers; IRIS+ code if one fits |
| Missing-value rule | No answer is unknown, never zero |
| Owner | Who approves a change to the definition |
One row from the fictional fund shows the shape: total clients enrolled, meaning unique trainees in the reporting period, not sessions; tagged How much and Who in the Five Dimensions of Impact; mapped to IRIS+ PI4060, “Client Individuals: Total”; added across partners.
When a definition changes, keep a short change log as a team practice: the date, old and new wording, who approved it and which reports it affects. A 30-day and a 90-day follow-up are different metrics, so give the new one its own row. The chapter A shared data dictionary: metric, dimension, standard has a full entry and a starting template.
Here is what a dictionary like that protects in a finished community-investment review: committed, paid and spent dollars kept apart, outcomes kept by program, and participations never counted as people.
PROMPT · PASTE INTO CLAUDE, CHATGPT OR YOUR AI TOOL
Below are the CSR metrics we ask our grantees to report, as written in our grant agreements. For each metric, draft one dictionary row: plain definition (who counts, what qualifies, when), unit, collection point, missing-value rule, owner, and whether it is an input, output or outcome. Rules: - Use only what our agreements say. Where they are silent, write "ASK: [question for the grantee]". - Do not invent numbers, targets or benchmarks. - If a metric is named but never defined, write "not in our data". - Flag any two metrics that could be counting the same people twice. [PASTE METRICS FROM GRANT AGREEMENTS]
How do you add up CSR metrics across grantees?
Add only the numbers that share a definition, and put every other number beside the total with a one-line reason and the question you have asked. A total is honest only when every part means the same thing.
The fictional fund’s placement metric shows why. Partners A, B and D counted placements within 90 days of exit: 42, 31 and 27, a total of 100. Partner C reported 55 counted within six months, so its number is held until C confirms its 90-day count.
Reporting 155 (the 100 plus C’s 55) would look better and mix two measures; the first board member who asks “within how long?” will find it. The same rule covers rates: a company-wide rate is total placed divided by total completers across the grantees that match, never the average of each grantee’s rate.
The short Sopact Sense introduction below, AI Data Collection With Context, covers the collection side: gathering data with its context so your own team can manage and analyze it. Clean collection at the source is what lets a shared definition hold when the totals are added.
In Sopact Sense, each grantee’s data can sit in its own folder, where that team’s AI Assistant sees only its folder and the CSR owner sees the aggregated results; every line of an answer links to a record you can open. The chapter Roll up and benchmark portfolio results covers held numbers and outside benchmarks.
How do you choose a manageable CSR KPI set?
Start from the decision the company needs to make, pick the fewest metrics that inform it, and name an owner for each before the reporting deadline. Five well-defined metrics every grantee can collect beat twenty that half of them skip.
- Start with the decision. What does the company need to understand or change about its giving?
- Separate management needs from disclosure. An internal KPI and a required datapoint may need different evidence.
- Include affected people. Ask which voices your current sources leave out.
- Balance activity and outcomes. Keep counts that explain delivery, and add at least one outcome.
- Check the grantee’s burden. Use what they already collect before adding a new form.
- Assign owners. Name who owns the source, the calculation and the review.
The impact strategy guide helps connect priorities to the few metrics that matter, and the CSR reporting guide covers how to report them once they are agreed.
How do GRI and ESRS affect which CSR metrics you use?
When a reporting standard applies, start from its own definitions; a generic metric list is not a disclosure checklist, and a field labelled with a standard’s name is not automatically aligned. Keep a mapping from each requirement to its definition, source, calculation, gaps and reviewer.
The GRI Universal Standards work alongside Topic and Sector Standards. For ESRS, EFRAG’s implementation guidance covers materiality, value chain and datapoints, and describes itself as non-authoritative. Check which adopted requirements apply to you before planning a disclosure.
The video below shows the dictionary side of this: one set of definitions whose rows map to IRIS+, the SDGs and ESRS, instead of a spreadsheet rebuilt for each framework. Watch for where each framework label attaches: to a definition every grantee already counts by.
What can CSR metrics not prove?
Outputs are not outcomes, and a rise in an outcome after a grant is not proof the grant caused it. Placements can climb because the local job market improved; a claim of cause needs a comparison, such as earlier cohorts counted the same way.
Much grantee evidence is self-reported follow-up, and the people who answer may differ from those who do not, so report unknowns as unknown. A supplier survey shows a practice is claimed, not that it works. If an AI tool drafts your dictionary or report, check it against the records; your team decides what the evidence supports.
Start with one grantee metric this cycle
Pick one outcome metric that every grantee in one program reports, and make it add up before you touch the rest.
- Choose the outcome your board asks about most, such as placed in a job within 90 days of exit.
- Write its dictionary row: definition, unit, collection point, missing-value rule and owner.
- Send the row to every grantee in the program before the next report is due.
- When reports arrive, mark each number as matching the definition or not.
- Add the matches, show the rest beside the total with one line each, and send each grantee its question.
After the first cycle you have one CSR metric with a written definition, a total you can defend, and a short list of grantee questions that improves the next report.
Frequently asked questions
What is the difference between a CSR metric and a KPI?
A metric is any defined measure, such as grantee trainees placed in a job within 90 days of exit. A KPI is a metric the company has chosen because it matters to a specific objective, usually reviewed against a target. Both need the same written definition, source and owner; calling a metric a KPI does not make its number more reliable.
What are examples of CSR metrics?
Common examples span eight areas: workforce representation and retention, community program outcomes, energy and emissions, supplier assessment coverage, safety incidents, ethics training, customer access and stakeholder perspectives. For community investment, go past grant dollars and volunteer hours to at least one outcome for the people served, such as placements that last 12 months.
How do you measure CSR program outcomes across grantees?
Agree the outcome metric and its definition with every grantee before the first report, including the time window and how missing answers count. When reports arrive, compare each number with the definition, add only the matches and show the rest beside the total. Unique people, a fixed window and a missing-value rule do most of the work.
Does a shared framework make CSR metrics comparable?
Not on its own. Two grantees can use the same framework heading and still count different people, periods or units. Comparability comes from a shared definition: who counts, what qualifies, when, from which source and in what unit. Check those before adding or comparing any two numbers, whatever framework label they carry.
Should past CSR results change when a definition changes?
Only by a documented decision the underlying evidence supports. Keep the original definition and the results reported under it, record the change in a change log, and restate earlier periods only if the records allow it. Otherwise start a new series and explain the break, rather than presenting two definitions as one trend.
How does Sopact support CSR metrics?
Sopact Sense gives each person or grantee a persistent ID from the first form, keeps each grantee’s data in its own folder with the CSR owner seeing aggregated results, and links every line of an AI Assistant answer to a record you can open. It does not make a disclosure compliant; that still needs review.
