Measurement
Part of Building a corporate communications report with its evidence states kept apart
Benchmark research for corporate communications when no industry norm exists
Research comparable communication evidence for an England decision without inventing an industry norm, threshold, result or causal conclusion.
Corporate communications benchmark research should begin with an admission test, not an industry average. For Cedar Vale Engineering Ltd's fictional England site decision, no public figure can be assumed to describe the same organisation, audience, message, channel, period and outcome. This article therefore produces a protocol and an evidence-gap record, not a benchmark.
Define the comparison unit
Write the buyer decision first. Then specify the communication category, authorised source-copy state, recipient population, geography, channel, event, numerator, denominator, unit, reporting period and timezone. A customer-newsletter open definition cannot quietly become an employee-understanding comparator. UK evidence is not an England estimate unless the source publishes that geography.
The ONS says statistical quality includes relevance, accuracy and reliability, timeliness, accessibility and clarity, and coherence and comparability in its quality definition. That framework describes official statistics, but its dimensions are useful questions for admitting external evidence.
Build a source-universe log
Record where the researcher looked, the search date, publisher, dataset or document version and reason for exclusion. Search official statistics, regulator publications, transparent evaluations and buyer-owned historical data separately. Supplier case studies, dashboard estimates and surveys can be logged as third-party claims, but not promoted to independent observations.
For every candidate source, capture:
population | geography | communication job | event | numerator | denominator | unit | period | sample or frame | collection method | weighting | missingness | uncertainty | revision status | funding | link
An empty field is evidence of uncertainty. Do not fill it by inference.
Apply non-negotiable admission rules
Admit a comparator only if its event definition and eligible population match the buyer measure closely enough for the stated decision. Periods, currencies, channels and units must be aligned without concealing transformations. Keep estimates separate from observed records, and retain confidence intervals or other uncertainty exactly as published. If methods changed, break the series.
The UK Statistics Authority's Trustworthiness, Quality and Value explanation notes that quality involves appropriate methods and transparency. It says the framework can support evidence beyond official statistics, but voluntary application does not validate Cedar Vale's eventual comparison.
Use the organisation's past carefully
Buyer-owned history may be more relevant than a public average, yet it still needs the same definition checks. Document changes in message purpose, audience eligibility, channel configuration and source queries. A prior result is an observation under its old conditions, not a target for the next communication.
If a threshold is needed, the decision owner should state the harm or action it governs. It must not be reverse-engineered from a favourable result. Preserve blank fields for baseline, tolerance and stop rule until evidence and authority exist.
Check publication bias before treating the surviving material as a reference set. Public case studies may favour selected outcomes, while failed or inconclusive work may be absent. Record who funded each source, whether the full method and denominator are visible, and whether corrections or withdrawals were searched. An opaque success story can remain in the source log, but it cannot enter the comparable set.
Report the gap as a finding
The output can say that no compatible comparator was admitted as at the research date. It should list the closest excluded sources and the decisive mismatch, such as geography, denominator or supplier-defined event. That is more actionable than a synthetic average because it tells Cedar Vale what evidence to collect next.
HM Treasury's 2026 Magenta Book concerns government evaluation and encourages early evaluation design and careful interpretation. It supplies no private corporate norm. Cedar Vale's named statistical and evaluation reviewers must decide whether a local baseline, a new study or no comparison is proportionate. Until that review, any benchmark field remains blank and the publication verdict remains hold.