Choosing corporate communications measures with complete definitions and blank thresholds

Corporate communications key metrics should be admitted because they answer a named decision, not because software happens to display them. In the fictional Cedar Vale Engineering Ltd case, the question is whether authorised groups received the approved information about a proposed England site consolidation and whether an evidence gap requires correction or further research. No release or result is claimed.

The UK government's communication functional standard distinguishes outputs, outtakes and outcomes in its campaign requirements. Use those categories as a filing discipline only. They do not provide Cedar Vale with private-company benchmarks.

Four useful measure families

Readiness measures test control before release. Examples are approval completeness, rights evidence status and accessible-format readiness. Each is a pass, fail or unresolved state against buyer-owned evidence, not a percentage score.

Delivery measures record controlled technical events. An authorised version published to a named channel is different from a message dispatched, accepted by an endpoint or opened under a supplier's definition. The data contract must retain those distinctions.

Audience measures concern understanding, recall, sentiment or ability to act. They require an appropriate research design. A click, receipt or potential-audience estimate cannot substitute for a response from the defined population.

Organisational measures concern the business or service decision, such as correctly routed enquiries. They need an eligible population and operational source. A movement after publication is an observation; whether communication contributed or caused it is a separate evaluation question.

Complete the card before collecting

For each candidate measure, write:

• purpose and authorised decision;
• exact event and eligible population;
• numerator, denominator and unit, or a reason no ratio applies;
• England, UK or other geographic coverage;
• period, clock start and stop, and timezone;
• inclusion, exclusion and repeat-event rules;
• source system, export and query version;
• missingness, estimation and known uncertainty;
• evidence owner, access control and retention route;
• blank buyer threshold, response and review trigger.

If one field is unknown, mark it unresolved. The ONS quality dimensions address official statistics, but their focus on relevance, accuracy, timeliness and comparability provides a useful challenge for a corporate measure. It does not certify the result.

Reject seductive substitutes

Do not rename a supplier's estimated audience as reach, count a system receipt as understanding, or divide events from one population by people from another. Avoid combining employee, customer and journalist activity merely to make a larger total. A rate is meaningful only when its denominator had the opportunity to produce the numerator.

Personal-data fields require their own necessity test. The ICO's data-protection principles cover purpose limitation, minimisation, accuracy, storage and security, and the page carries a DUAA review notice. Qualified review must determine the actual processing; adding a dashboard field does not settle that question.

Set the response before seeing the result

A metric is operational only when an owner has written what happens next. A missing approval record should stop release. A delivery-data gap might pause interpretation and open a source investigation. An inaccessible chart should trigger an alternative presentation and retest. A change in source definition should break the series rather than be hidden.

Keep thresholds blank until Cedar Vale supplies a justified baseline, risk tolerance and decision date. The final card should say whether the measure is descriptive, attributed by rule, modelled by a third party or intended for causal evaluation. That label prevents a tidy number from acquiring a stronger meaning than its evidence allows.