Finding grant benchmarks that match your geography, period and cost basis

A benchmark is useful only when the compared measure means roughly the same thing. Match geography, period, population, intervention, quality and cost basis before using another programme's number to judge an England small business project.

What to take away

Define the question

State whether the benchmark will test price, unit cost, delivery time, reach, output quality or outcome. Write the internal measure specification first. Research cannot find a comparable value for an undefined metric.

Decide the acceptable range of comparability. A board planning estimate may tolerate more uncertainty than a formal claim that one programme outperformed another.

Set inclusion criteria

Record the required geography, business size, sector, intervention type, delivery model, period and unit. State exclusions. Search England-specific official material first, then UK evidence with a clear scope note, followed by global material only when it remains relevant.

Do not silently treat UK, Great Britain and England as interchangeable. Check the population behind every table.

Prefer primary records

Use official statistics, published evaluations, programme reports, audited accounts and first-party methodology. Record title, publisher, URL, publication date, data period, table or page and access date.

The government's Evaluation Registry guidance explains the registry for planned, live and completed central government evaluations from April 2024. That makes it a discovery source for applicable government work, not proof that a particular study matches a business project.

The Magenta Book provides a framework for judging process, impact and value-for-money evidence. Read the actual evaluation method and limitations, not just its headline.

Name England and UK sources precisely. Useful named sources include the ONS UK Business Counts, the ONS Business Register and Employment Survey and the ONS Annual Survey of Hours and Earnings. HMRC's R&D tax credit statistics break claims down by scheme, year and region.

Grant records include Innovate UK projects published through Gateway to Research and GrantNav, run by 360Giving, which lists UK grants by funder, area, date and amount. English councils publish grants to voluntary and community organisations under the Local Government Transparency Code 2015.

Normalise with care

Align currency date, VAT treatment, nominal or real prices, project duration, participant definition and cost scope. Separate full economic cost, public funding and recipient contribution.

For a cost-per-output comparison, confirm that both numerator and denominator cover the same activity and time. Do not compare approved budget in one programme with paid cost in another.

Worked example, with illustrative amounts. Source A records an approved grant of £120,000 including VAT at the 20% standard rate over 24 months. Source B records a paid grant of £100,000 excluding VAT over 12 months. 20 gives £100,000 excluding VAT, or £50,000 per year across 24 months. Source B equals £100,000 per year.

The normalised values differ by a factor of two, which the unadjusted comparison hides.

Assess evidence quality

Check sample size, missing data, selection, counterfactual, sensitivity and conflicts. A large programme can still publish a weak causal claim; a small study can provide useful process evidence without estimating impact.

The government's Quality in Policy Impact Evaluation guidance distinguishes observed outcome change from evidence that an intervention caused it. Preserve that distinction when importing a result.

Build a benchmark range

Where several comparable records exist, present their distribution and context rather than selecting the most favourable figure. Explain outliers and avoid a calculated average when underlying definitions differ.

When no defensible benchmark exists, say so. Use the project's own baseline, target assumptions and later actuals, clearly labelled, instead of manufacturing market authority.

Document the research

Keep the search date, terms, sources screened, criteria, extraction, conversions and reviewer. Schedule an update when the benchmark is time-sensitive. Quote no number without its period and source.

Present the result for review

Build an evidence table with one row per source. Show the original measure, normalised measure, conversions, comparability judgement and limitation. Keep rejected records in a separate log so another reviewer can see why they were excluded.

Use a range when evidence varies and explain what drives it. If one record differs because it includes capital costs and another excludes them, do not hide the distinction inside an average. State which point or range informed the project target and why.

Ask a second person to reproduce at least one extraction and every material conversion. When the source changes or disappears, retain the permitted record, retrieval date and enough citation detail to verify what was used.

Before you act

Common questions

Why should I avoid treating UK, Great Britain and England as interchangeable?

The article warns that these terms cover different populations, so you must check the population behind every table. Using a UK figure for an England small business project may mislead because the underlying geography and population differ.

What should I do if no comparable benchmark is available?

The article says to use the project's own baseline, target assumptions and later actuals, clearly labelled. This keeps the comparison honest and avoids false precision.

How should I handle differences in cost scope between programmes?

Separate full economic cost, public funding and recipient contribution. Do not hide distinctions inside an average. State which point or range informed the project target and why, and explain what drives the variation.