Measurement
Part of PR agencies measurement: reject AVE, keep evidence levels apart
PR benchmarks without an invented average: population, unit, method and comparability
Research PR benchmarks for England without inventing an average: define the population, unit, method, comparability, uncertainty and evidence gap.
This is a method guide, not a benchmark dataset. It sets out how an England PR benchmark is defined and how a buyer judges whether two figures may be compared. No England-wide value for reach, sentiment or return on investment is published here, and none is invented.
What to take away
- No single England benchmark covers reach, sentiment and ROI. A figure without a stated population, unit and denominator cannot be used.
- Draw the England population from a named register, then fix the unit, period and calculation before comparing.
- Classify each source as compatible, partly compatible or incompatible, and pool only the first.
- Keep descriptive comparison separate from any claim that an agency caused an outcome.
- If nothing external matches, publish the gap and build a baseline from your own stable records.
Put the decision before the data
Write the decision the number must support, then work backwards to the evidence. Five steps cover most England cases.
- State the decision and who acts on it.
- Name the population, including England geography and sector.
- Fix the unitlegal entity, agency appointment, campaign or media item.
- Set the period and the collection route.
- Record the calculation, including numerator and denominator.
A legal entity is not a campaign, and a media item is not an audience response. When the unit shifts between sources, the comparison fails.
Where an England population comes from
The ONS business services statistics hub publishes UK industry data on professional services, useful for sizing a population rather than scoring PR performance.
For an England population, the Inter-Departmental Business Register is the sampling frame behind ONS business surveys. SIC 2007 code 70.21 covers public relations and communications activities. The Business Register and Employment Survey reports employee jobs by industry and region, including England regions, each year, with employee jobs as the denominator for the stated period.
Those registers give a defensible population and denominator. They say nothing about reach, sentiment or media outcomes. For the audience definition itself, defining a research population in England shows how to bound a PR population without borrowing a supplier list.
Capture ten fields for every candidate figure, starting with population and denominator.
Evidence fields for each figure
- Populationwho could enter source
- GeographyEngland, UK, other or global
- Periodfieldwork and release dates
- Measureexact event or variable
- Calculationnumerator, denominator, unit, weighting
- Collectioncensus, sample, log, model
- Coveragechannels, languages, publishers
Example: one England question, two candidate sources
An England B2B software team wants a share-of-voice figure for its sector and is offered three sources.
One England question, two candidate sources
Population and unit
- Own monitoring log, England trade titles, fixed query, named date range
- English-language items mentioning the brand, counted per item
- Supplier case study quoting global coverage totals
- Publisher universe and denominator unstated
- Sector report with England respondents
- Respondents, not media items
Verdict
- Own monitoring log, England trade titles, fixed query, named date range
- Compatible
- Supplier case study quoting global coverage totals
- Incompatible
- Sector report with England respondents
- Partly compatible
The partly compatible record can shape study design. It must not enter a pooled average.
Test the meaning before you pool
AMEC's evaluation framework separates outputs, audience responses, outcomes and impact. A coverage count therefore cannot benchmark understanding or reputation change. Share of voice needs a declared source universe and subject rules, and sentiment needs a coding scheme plus a validation statement.
A coverage count measures distribution. It does not measure whether anyone changed their mind.
The Barcelona Principles 4.0 reject advertising value equivalency and require transparency about sources, samples and methods. They still give no England average.
Why an England label is not enough
ONS quality and methodology information lists relevance, accuracy, timeliness and comparability among the dimensions a user should test. Apply them as questions to a private dataset. The ONS does not certify supplier monitoring.
Record every mismatch in population, geography, service and period. Add the calculation, sample and uncertainty limits, then classify. For what a benchmark cannot settle, why allocation is not incrementality explains why a measured value and a caused value are different questions.
Keep causality out of the comparison
HM Treasury's Magenta Book separates process, impact and value-for-money questions in evaluation. The same warning applies to agency reporting: a rising figure beside a campaign is a description, not proof of effect.
When no external figure fits
Most England PR questions end here, and that is a legitimate result. Work through this list before commissioning anything.
- Publish the evidence gap in the report.
- Build an organisation baseline only from stable definitions and collection routes.
- Log every change of supplier, tracking tool, channel or query logic.
- Commission a study with a declared sampling frame and publication protocol.
- Have a qualified statistician approve the design before collection.
If you are choosing a delivery route too, six PR delivery models on one comparison basis compares models on shared terms, keeping the benchmark question separate from the buying question.
Common questions
Are there England-wide PR benchmarks for reach or sentiment?
Not in a form that can be pooled. England coverage and sentiment datasets come from suppliers, each with its own publisher universe and coding rules.
Which England dataset should I start with?
ONS business statistics drawn from the IDBR, using SIC 70.21 for public relations and communications activities. That gives a population and a denominator, but no media or sentiment value.
What makes two PR benchmarks comparable?
Matching population, unit, period, calculation and collection method. One mismatch makes the pair partly compatible at best, and partly compatible sources stay out of pooled averages.
Can I compare England results with a UK or global figure?
Only as context. A UK or global figure usually covers a different publisher universe, sample and period. Label it as context in the report.
