How to benchmark social value without losing sight of local need

Benchmarking can show where social value performance differs. It cannot explain the difference on its own.

A credible comparison combines consistent measures with local context, community knowledge and transparent methodology. The most useful benchmark leaves you clearer on what to investigate, what to improve and where future investment may create the greatest benefit.

What is social value benchmarking?

Social value benchmarking is the process of comparing social value performance against a defined reference point. That reference point might be another supplier, a group of similar organisations, a previous year, a sector average or a set of comparable places.

Used carefully, benchmarking can help an organisation:

• understand whether performance is improving
• identify projects or suppliers that may need additional support
• set realistic targets
• find examples worth investigating
• compare activity across a portfolio
• strengthen board and public accountability
• improve future commissioning and investment decisions

The term covers several different types of comparison. Organisations may benchmark activities and outputs, outcomes, monetary social value such as an SROI ratio, or measurement maturity. Each answers a different question. A comparison of volunteering hours says little about whether the volunteering met a local priority. A comparison of outcome values can also mislead when projects use different assumptions, beneficiary groups or evidence standards.

Benchmarking and ranking are different

A benchmark provides a reference point. A ranking implies that higher or lower performance can be interpreted consistently across everyone being compared.

Two projects can report the same number of job outcomes while working with people who face very different barriers. Two suppliers can report different financial values because one uses more conservative assumptions. A rural service may cost more to deliver because participants are dispersed and transport is limited.

The UK Government's Local Outcomes Framework offers a useful example. It is designed to support comparison between local authorities, while its statistical-neighbours model groups areas with similar contextual characteristics. The accompanying guidance presents comparison as a way to support improvement and understand progress, rather than as a simple national ranking.

What can reasonably be compared?

Similar contracts or projects may be reasonable peers when they share a similar purpose, comparable scale and duration, a similar delivery model, related beneficiary groups, comparable geographic conditions, consistent outcome definitions and equivalent evidence requirements. Perfect matches are rare. The aim is to identify the differences and decide whether they are significant enough to change the interpretation.

The same programme over time can reveal trends, changes between participant cohorts and the effects of improvements to delivery. Check whether the measurement approach has changed, as a stronger survey process, a different financial proxy or a longer follow-up period can alter the reported result even when delivery is broadly similar.

Similar organisations may be compared using characteristics such as sector and purpose, turnover or contract value, workforce size, delivery model, geographic reach and maturity of measurement systems. A national company delivering major infrastructure contracts will have different opportunities and resources from a local SME or community organisation.

Places with comparable contexts are not always the same as neighbouring places. Geographic proximity does not make two areas suitable peers. The statistical-neighbours approach compares places using a set of contextual indicators, providing a more balanced reference group than a national average or a list based only on location.

The six conditions for a fair social value comparison

1. Comparable questions

Begin with the decision the benchmark needs to support. "Which project generated the largest reported value?" is a different question from "Which approach produced the strongest outcome for the intended population?" The first may help describe a portfolio. The second requires a closer look at beneficiary need, outcome quality, evidence and cost.

2. Consistent outcome definitions

Small differences in definitions can produce large differences in results. Employment measures may refer to a job created, a person entering work, employment sustained for 13 or 26 weeks, progression into better-paid or more secure work, or an existing job safeguarded. These measures should remain separate unless the comparison clearly explains how they have been reconciled.

3. A shared time period

Agree which period is being compared: delivery period, contract year, financial year, point of follow-up or expected duration of the outcome. A programme reporting employment starts at completion cannot be compared directly with one that follows participants for six months.

4. Equivalent evidence standards

Evidence may range from a supplier declaration to an independently verified evaluation. A comparison should show where the evidence differs. A higher figure supported by weaker evidence should not automatically be treated as stronger performance.

5. Transparent methodology

Where outcomes are given a financial value, check that the comparison uses consistent financial proxies, price and inflation years, deadweight, attribution, displacement, leakage, duration and drop-off assumptions, and rules for avoiding double counting. These assumptions influence the result and should be visible alongside the headline figure.

6. Relevant contextual information

Local conditions can affect the need for an intervention, the resources required to deliver it and the results that can reasonably be expected. Relevant context may include employment and economic inactivity, deprivation and household income, health and disability, transport and digital access, housing conditions, rurality, demographic change and existing public and voluntary-sector provision. Context should be selected in relation to the outcome being studied.

A practical process for benchmarking social value

Step 1: Define the decision

Write down what the comparison will inform. Examples include selecting a supplier, reviewing a grant portfolio, improving an existing programme, setting future targets or understanding variation between places. This keeps the exercise focused and helps determine how much evidence is proportionate.

Step 2: Choose a sensible comparison group

Select peers using explicit criteria. Consider purpose, scale, population, delivery model, geography and evidence maturity. Record why each project or organisation is included. If the group is broad, divide it into more meaningful subgroups.

Step 3: Agree common and local measures

A useful framework can contain three layers: core measures collected consistently across the portfolio, locally relevant outcomes reflecting the needs and priorities of each place, and learning measures used where a project needs to test or understand something specific. This structure provides enough consistency for comparison while leaving room for local relevance.

Step 4: Create a data dictionary

Define each field, unit and evidence rule. Include the outcome definition, beneficiary criteria, calculation method, evidence required, reporting frequency, exclusions, treatment of missing data and ownership. A shared label is not enough if suppliers interpret it differently. The data dictionary creates the practical basis for a fair comparison.

Step 5: Add local data

Local data helps establish the starting point and explain why apparently similar interventions may produce different results. Place Explorer provides free place-based profiles for UK postcodes and local authority areas, drawing together official information on deprivation, Census 2021 demographics, employment, economic conditions and social cohesion.

Use the data selectively. For an employment programme, labour-market conditions, skills, transport and deprivation may be relevant. For a loneliness intervention, age, health, household composition and community assets may be more useful.

Step 6: Add community voice

Published data shows patterns. Residents and service users can help explain how those patterns are experienced and what may be missing. Two areas may have similar levels of economic inactivity. In one, residents may identify limited public transport as the main barrier to work. In another, affordable childcare, poor health or a shortage of suitable jobs may be more important.

For guidance on collecting community feedback well, see how to design effective, low-friction stakeholder surveys. Community engagement should be proportionate and accessible. It should also close the loop by explaining how people's contributions were used.

Step 7: Compare distributions as well as averages

Averages can hide variation. Where the data allows, examine the range, the median, quartiles, outliers, results by geography, results by beneficiary group, and confidence and completeness. A portfolio average may look strong while one group experiences consistently weaker outcomes. Distributional analysis can make questions of fairness and inclusion more visible.

Step 8: Investigate the differences

A benchmark identifies a difference. The next step is to understand it. Ask whether delivery was genuinely stronger, whether beneficiary groups were comparable, whether local conditions affected the result, whether one organisation collected better evidence, whether different proxies or assumptions were used and whether an outlier is caused by a data-quality problem. Qualitative evidence, practitioner knowledge and community feedback are particularly valuable at this stage.

Step 9: Record what changes

Benchmarking should lead to a decision or a further question. Possible actions include changing programme design, offering targeted support to suppliers, adjusting evidence requirements, commissioning further evaluation, revising targets or redistributing funding. Record the decision, the evidence used and any uncertainty. This turns benchmarking into a learning process rather than a reporting exercise.

Worked example: two employment programmes

Programme A and Programme B each report that 30 people entered employment. At output level their performance appears identical. A fuller comparison shows the following differences.

Programme A worked with recently unemployed adults in an area with a high vacancy rate and good public transport. Employment was confirmed at start only, using a supplier declaration without deadweight assessment, and included mixed hours and contract types.

Programme B worked with people out of work for more than two years in an area with a low vacancy rate and limited transport. Employment was sustained for 26 weeks and confirmed through payroll and participant follow-up, with deadweight estimated using local evidence, mostly covering Living Wage roles with guaranteed hours.

Programme A may still be effective. The available evidence simply does not support the conclusion that the two results are equivalent. A fair benchmark would separate employment starts from sustained employment, include job quality, identify the participant groups and account for different labour-market conditions.

Worked example: comparing supplier portfolios

A commissioner finds that Supplier X reports twice as much social value per pound spent as Supplier Y. Before drawing a conclusion it checks whether both suppliers report contract-specific activity, whether the same outcomes and proxy values are used, whether evidence is equivalent and whether the contracts operate in similar places.

The review finds that Supplier X included company-wide volunteering and used gross values without adjustments. Supplier Y reported only contract activity, followed up outcomes and applied conservative assumptions.

The original benchmark remains useful. It revealed a difference worth investigating. Once the methodology is reconciled, the commissioner can compare performance more credibly and improve its reporting requirements for both suppliers. For guidance on how suppliers should structure their evidence, see how suppliers can evidence social value in public sector bids.

Common benchmarking mistakes

Comparing different methodologies without reconciliation. If methods differ, explain the differences or recalculate the results using a common approach. See what is the difference between SROI and the TOMs framework for a practical guide to how different approaches define and value social change.

Treating the most common measure as the most relevant. Measures are often popular because they are easy to collect. Frequency does not establish local importance or outcome quality.

Treating missing evidence as zero without explanation. A zero may mean that no activity took place, that evidence was unavailable or that the organisation did not collect the measure. These should be distinguished.

Rewarding easy-to-count activity. High volumes of workshops, volunteering or donations may look impressive while providing little evidence of sustained change.

Ignoring negative and displaced outcomes. An intervention may shift a problem elsewhere, exclude some groups or create unintended harm. A credible comparison makes these effects visible where they are material.

Creating league tables from uncertain data. Rankings can imply a level of precision that the evidence does not support. Use ranges, sensitivity analysis and contextual notes where appropriate.

Setting targets from the top performer without testing feasibility. The leading result may depend on different resources, circumstances or reporting methods. Investigate how it was achieved before using it as a target.

Download the benchmarking checklist

Before you compare any two social value results, use the worksheet below to check whether the comparison is fair and to record what you find. It covers whether the results are truly comparable and guides you through the factors that may explain any difference.

Download the Social Value Benchmarking Checklist