Keep outcome evidence, financial proxies and assumptions connected. Calculate a defensible SROI ratio, review new records, and explain what changed.
To calculate SROI as new evidence arrives, connect each outcome to its source record, apply documented financial proxies and contribution adjustments, and divide the present value of adjusted outcomes by the investment. Keep the scope and calculation rules versioned. When a record changes, review its effect before releasing an updated ratio. Program leads, evaluators and funders can use this method to maintain a calculation they can explain, rather than rebuild it from a fresh spreadsheet export.
“Live” describes how the evidence is maintained. It does not mean every new response is a verified outcome, every change should alter the published ratio, or a higher ratio proves the program improved.
A trustworthy calculation lets a reviewer follow the ratio back to the outcomes, assumptions and records behind it. A dashboard can refresh quickly and still carry weak evidence.
Keep three things distinct: what you observed, how you valued it, and how much contribution you claimed. For example, a participant’s reported wage increase is an observation. Annualizing that increase requires working-hours and duration assumptions. Estimating the share attributable to the program requires additional evidence. Store each separately.
The broader method is described in Social Value International’s Guide to SROI. This lesson focuses on keeping the calculation current and reviewable. For the full introduction, see the SROI framework and calculator.
Write down whose changes you will value, over which period, against which investment. Use that same scope whenever you refresh the result.
For a workforce program, specify whether the analysis covers one intake cohort or all participants in a year. Define employment, retention and earnings consistently. Include material negative outcomes and people who leave the program; excluding them can make the result look better without changing the work.
Give each outcome a source, an observation date and a valuation basis. Label borrowed estimates so a reader can distinguish measured change from assumptions about its monetary value.
Use a stable participant ID to connect intake, services, case notes and follow-up. A shared identifier links records; it does not establish that the program caused the change. Keep the relevant consent, access permissions and retention rules with the data workflow.
| Record component | What to store | Example or rule |
|---|---|---|
| Observed outcome | Baseline and follow-up, dates, units and source | Hourly wage before and after the program; distinguish self-report from verified payroll. |
| Monetary value | Proxy or valuation method, source, geography and year | If annualizing wage gains, state hours worked and the benefit period. |
| Contribution | Deadweight, attribution and displacement assumptions | Record each adjustment separately, with the evidence and uncertainty behind it. |
| Duration | Benefit period, drop-off and discount rate | Do not assume a year-one benefit continues indefinitely. |
| Review status | Complete, awaiting evidence, not applicable or disputed | A missing follow-up is an evidence gap, not a confirmed zero outcome. |
Where an outcome matters but no defensible monetary proxy exists, report it in its own units and explain the exclusion from the ratio. Do not invent a value to complete the table. Use the value-map walkthrough and financial-proxy guide for the underlying decisions.
Estimate how much outcome value should be credited to the work, then test how different reasonable assumptions affect the result. Participant feedback can inform these estimates, but a response such as “mostly because of the program” does not establish a universal numerical attribution factor.
| Adjustment | Question it answers | What to make explicit |
|---|---|---|
| Deadweight | What would have happened anyway? | The comparison, benchmark or other basis for the estimate. |
| Attribution | How much did others contribute? | Other services, employers, family support or contextual influences. |
| Displacement | Did a gain replace or reduce value elsewhere? | Who else may have experienced a loss or foregone benefit. |
| Drop-off | How do benefits fade in later years? | Evidence or assumptions about persistence over time. |
| Discounting | What are future benefits worth in present terms? | The rate, timing convention and reason for choosing them. |
Check for double counting. A wage gain, a tax contribution and a benefit saving may overlap depending on the stakeholder perspective and scope. They should not automatically be added together. See deadweight and attribution for a closer treatment.
SROI ratio = present value of adjusted outcomes ÷ value of inputs. A 3:1 result means an estimated $3 of social value per $1 invested within the stated scope. It is not $3 in cash returned or proof of causation.
The following example is hypothetical and matches the hero video. It assumes one outcome, no displacement, and benefits confined to the initial period, with no later benefits to discount. A complete study must justify those assumptions.
| Calculation | Working | Result |
|---|---|---|
| Gross outcome value | 100 people × $2,000 per outcome | $200,000 |
| Remove deadweight | 20% of $200,000 is excluded | $160,000 remains |
| Remove others’ contribution | 25% of the remaining $160,000 is excluded | $120,000 remains |
| Divide by investment | $120,000 ÷ $40,000 | 3:1 estimated SROI |
Now change one assumption. If deadweight is 40% rather than 20%, the adjusted value becomes $200,000 × 0.60 × 0.75 = $90,000. With the same $40,000 investment, the ratio is 2.25:1. The evidence about contribution changed; the program did not necessarily become better or worse.
For a multiyear model, calculate each period’s benefits, apply the relevant persistence and contribution assumptions, and discount future amounts consistently. Show the base case and sensitivity range beside the result. The ratio walkthrough explains the calculation in more detail.
A new record should update the evidence trail first. Recalculate affected values after the identity, units, dates and review status pass your checks. Keep the prior reporting version available.
| Incoming information | First check | Action |
|---|---|---|
| A follow-up wage response | Correct participant, date, currency and hourly or annual units? | Update the observed outcome; flag discrepancies for review. |
| A follow-up is overdue | Is the outcome unknown or explicitly reported as unchanged? | Show the coverage gap. Do not label missing evidence as verified failure or success. |
| A revised financial proxy | Does its population, geography and year fit the study? | Test the effect in a new model version before adopting it. |
| A mentor note challenges an outcome | Does it contradict a claim used in the calculation? | Route the discrepancy to a reviewer and retain both sources. |
| A reviewer approves a correction | Are the rationale and affected results recorded? | Save the updated version with the evidence cutoff and change log. |
Track evidence coverage beside SROI: how many follow-ups are due, how many arrived, and which outcomes remain unresolved. If you produce a provisional estimate, label it and explain how missing information is treated. Avoid interpreting a ratio that rises as responses arrive as proof of better performance.
Keep cost per outcome separate. It answers a different question. If your outcome is employment retained for 90 days, divide the relevant program cost by the number of people meeting that definition. Report the cohort, period and evidence coverage; do not substitute enrollment or completion for retention.
You can build the initial value map and calculation in a spreadsheet. The recurring challenge is maintaining connections across forms, documents, case notes and follow-up records while keeping definitions and access consistent.
In Sopact, organize those sources around the participant, program and observation date. Configure the analysis around your approved outcome definitions, proxies and adjustments. Use source-linked review to investigate gaps and compare calculation versions. Confirm the configuration against a manually checked sample before relying on it for reporting.
AI can help extract candidate evidence, surface conflicting records and explain a calculation. Reviewers still decide whether an outcome is supported, a proxy is appropriate and a contribution claim is defensible. Keep sensitive participant details available only to authorized roles.
Try this review prompt with a value map and calculation you can inspect:
Review this SROI model using only the supplied records, scope, proxies and adjustment rules. For each outcome, show the observation, source, monetary value, contribution adjustments and review status. Flag missing evidence, inconsistent units, overlapping benefits and assumptions without a cited basis. Reproduce the base-case arithmetic and test the supplied sensitivity ranges. Do not invent missing values or infer causation from a linked record. List the decisions a human reviewer must resolve before publication.
Share the ratio with the information needed to question it. A single number without scope, sources or assumptions is difficult to use responsibly.
Start with one outcome this week. Connect its baseline and follow-up, identify a defensible valuation basis, and reproduce the calculation by hand. Only then extend the model to additional outcomes or recurring updates.
It means the underlying evidence and calculation can be maintained as records change. Your workflow should distinguish a working estimate from an approved reporting version. Decide which changes can be processed routinely and which require review, then save the assumptions and evidence cutoff with each released result.
A missing response means the outcome is unknown. Treating it as zero may be a conservative modeling assumption in a particular analysis, but it is not an observed zero. Explain the choice, report response coverage and test how alternative treatments affect the result. Retain the full population defined in the study scope.
Participant accounts help explain what contributed to change. They should be interpreted alongside the study design and other available evidence. Converting “mostly” into a fixed percentage is an assumption requiring justification, not a general SROI rule. Record the reasoning and test a reasonable range rather than presenting the percentage as a measured causal effect.
No. A ratio can rise because of a broader scope, a larger proxy, different duration assumptions or lower estimated deadweight. Compare like-for-like model versions and explain the drivers. Across programs, differences in participants, outcomes and valuation methods can make headline ratios unsuitable for a simple performance ranking.
Do not add individual ratios. First check whether the scopes, periods, valuations and contribution rules are compatible. Where aggregation is justified, combine the appropriately adjusted outcome values and corresponding inputs, then calculate the aggregate ratio. Use the portfolio SROI guide to examine those conditions.
Keep the estimate provisional and show the specific gap: a missing observation, unsuitable proxy, uncertain contribution estimate or calculation error. Assign someone to resolve it. You can still use non-monetized outcome evidence to improve the program while the valuation work continues; a premature ratio is not a prerequisite for learning.
Bring one outcome, its baseline and follow-up records, your investment total, and your proxy assumptions. Explore how to keep the calculation connected and reviewable.
Explore Case Management →
