A money value for the change you caused, and a discipline for subtracting everything you did not. The subtraction is the whole point.
SROI values the outcomes of an activity in money, sets that against what the activity cost, and expresses the result as a ratio. It comes out of the Principles of Social Value, and its distinguishing feature is not the money — it is that stakeholders decide what changed and what it was worth, and that you then have to take back off everything you cannot honestly claim.
It runs in two forms. A forecast SROI is done before you start, to decide whether something is worth doing. An evaluative SROI is done afterwards, on outcomes that actually happened. Most first attempts should be forecasts, because an evaluative analysis needs data you will only have if you planned to collect it.
What are we analysing, over what period, and who is affected?
Scope is decided first and written down — the activity, the timescale, the boundaries, and whether the analysis is forecast or evaluative. Then you list who changes as a result, including people the work affects without meaning to, and decide who to involve. You may leave a group out, provided the analysis says so and says why.
What goes in, what happens, and what changes as a result?
Working with those stakeholders you build an Impact Map, the one document the whole analysis hangs off. It runs left to right from inputs, to outputs, to outcomes, and it separates them properly: running twelve courses is an output, and forty people still in work a year later is an outcome. Inputs are valued here too, including donated time and premises.
How do we know it changed, and what is that change worth?
Each outcome gets an indicator, a quantity, and a duration — how many years it lasts. Then each is given a financial proxy: an approximation of value where no market price exists, such as the cost of the treatment avoided, or what people spend to get the same thing another way. Choosing proxies is the most contested part of the method. The guide's answer is to state the proxy and where it came from so it can be challenged, rather than to bury it.
How much of this would have happened without us?
Everything valued in stage three is cut back by four adjustments — deadweight, displacement, attribution and drop-off. What survives is impact. It is set out in full below.
What is it worth, in today's money, against what it cost?
Impact is projected across the years the outcomes last, reduced by drop-off each year, then discounted back to a net present value. Divide by the investment and you have the ratio.
Who needs to see it, and what changes because of it?
The analysis goes back to the stakeholders it came from, not only to the funder, with the assumptions visible. Then it gets embedded in how the organisation collects data and makes decisions, and, if it matters that people believe it, assured by somebody independent.
Anybody can put a value on an outcome. The credibility of an SROI is entirely in what gets taken back off, and a report that does not show these four has skipped the only stage that constrains the answer. The figures below are round numbers chosen to show the arithmetic, not anybody's result.
What would have happened anyway?
The amount of the outcome that would have occurred without the activity. Usually estimated from a comparison group or the general trend — if employment in the area rose 15% regardless, 15% of your employment outcome was not yours.
Did we move the problem rather than solve it?
How much of the outcome pushed something else out. Fewer burglaries on one estate is not a result if they went up on the next one.
Who else caused this?
How much of the change was down to other organisations or people. It is the adjustment most often left out, and the one a funder is most likely to ask about.
How quickly does it fade?
The deterioration of an outcome over time. Applied to each year after the first, so a benefit claimed over five years is worth progressively less in each of them.
Four steps between an impact figure and a defensible ratio. The third is the one that separates an analysis from an assertion.
How long does each outcome last?
Impact is carried across the duration you set in stage three, with drop-off applied in every year after the first. Claiming a long duration makes the ratio bigger and the analysis weaker, so duration has to be evidenced like anything else.
What is a benefit in five years worth today?
Future value is discounted to a net present value, because money and benefit arriving later are worth less than the same amount now. The discount rate is a stated assumption, not a fact.
Would a small change move the answer a long way?
Sensitivity analysis re-runs the calculation with the least certain assumptions changed — the proxies, the durations, the deadweight estimates. If a modest change swings the result, the result is not reliable, and saying so is part of the report.
What is the return, and when does it arrive?
Net present value divided by the investment gives the ratio. The payback period says how long before the value exceeds what was put in. Both are reported alongside the assumptions, never on their own.
It is a serious method and it is not free. Being honest about where it does not fit is part of using it well.