Benchmark research for corporate communications when no industry norm exists

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.

What to take away

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](/corporate-communications-cost-guide-in-england) | 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; 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 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.

Before you act

Common questions

Why is an industry average unsuitable for corporate communications benchmark research?

No public figure can be assumed to describe the same organisation, audience, message, channel, period and outcome. The article says research should begin with an admission test, and it produces a protocol and evidence-gap record instead of a benchmark.

What should a researcher record for every candidate source?

Capture the population, geography and communication job. Also record the event, numerator, denominator, unit and period. Then add sample or frame, collection method and weighting. Finish with missingness, uncertainty, revision status, funding and link. An empty field is evidence of uncertainty and must not be filled by inference.

How should buyer-owned historical data be treated?

It may be more relevant than a public average, but 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.