book, words, reading, text, dictionary, education, learning, dictionary, dictionary, dictionary, dictionary, dictionary
Photo by lucasgeorgewendt on Pixabay

Measurement

A blank announcement dashboard built on a data dictionary and change control

Build a blank reporting dashboard for one England announcement with reproducible definitions, evidence lineage, corrections and decision controls.

A useful dashboard for an England announcement is blank until its definitions and evidence pass review. It does not begin with a colourful total. Begin with a decision, then expose the delivery lane, source record and limitation behind every displayed measure.

The framework below was prepared on 6 September 2026. No buyer records, supplier account or live report was accessed. Every value, threshold, service level and comparison field is intentionally empty. It is a reusable design template, not a live dashboard, downloadable tool, example result or industry standard.

What to take away

  • Complete the data dictionary first, with one row per measure, so labels such as pickup or reach carry an operational meaning.
  • Use text states as well as measures so an absent number is never displayed as zero.

Establish three data layers

The raw layer holds supplier responses, newsroom events, authorised outreach logs, paid-placement records, monitoring observations and finance entries. Add source and ingestion timestamps, source-copy ID, route, schema version and retention decision.

Three dashboard data layers

  1. Raw layer: supplier responses, logs, finance entries
  2. Transformation layer: versioned query, eligibility, exclusions
  3. Presentation layer: fields for one named decision
  4. Each layer carries source and ingestion timestamps
  5. Changed rule creates a new version

The transformation layer holds a versioned query or calculation. Record inputs, eligibility, exclusions, deduplication, timezone, missing-value treatment, owner, reviewer and release date. A changed rule creates a new version.

The presentation layer selects only fields needed for a named decision. ONS describes statistical quality in terms that include relevance, accuracy, timeliness and coherence and comparability. Those dimensions can be used as challenge questions here, although this private dashboard is not an official statistic.

Complete the data dictionary first

Create one row per measure. Do not accept a label such as "pickup", "reach" or "engagement" without its operational meaning.

Data dictionary required fields

  • Decision and responsequestion, pass or fail action
  • Event and populationobservable state, eligible records
  • Calculationnumerator, denominator, unit
  • Time basisclock start, stop, timezone, period
  • Evidence lineagesystem, export, query, versions
  • Data treatmentdeduplication, missingness, exclusions
  • Accountabilityowner, reviewer, retention, deletion

Dictionary field requirements

Decision and response
[owner's question; action if gate passes or fails]
Event and population
[observable state; records eligible to enter]
Calculation
[numerator; denominator; unit]
Time basis
[clock start; clock stop; timezone; reporting period]
Evidence lineage
[system; export; query; schema and code versions]
Data treatment
[deduplication; missingness; exclusions; reconciliation]
Accountability
[evidence owner; reviewer; retention and deletion]
Interpretation
[uncertainty; buyer-set tolerance; recheck trigger]

Worked example rows for two measures, national pickup and reach:

National pickup vs Reach

National pickup

Decision and response
Did the launch secure at least one national pickup? If yes, brief the board; if no, extend outreach.
Event and population
National outlets on the approved list.
Calculation
Count of outlets publishing the release; denominator: eligible national outlets.
Time basis
Publication timestamp to 7 days after release; Europe/London.
Evidence lineage
Monitoring export; query version 3; schema 1.
Data treatment
Deduplicate by outlet and URL; exclude missing timestamps.
Accountability
Communications lead; reviewer: data manager; retain 24 months.
Interpretation
Compare against buyer-set tolerance; recheck if coverage changes.

Reach

Decision and response
Did reach meet the buyer-set tolerance? If yes, keep the plan; if no, recheck coverage.
Event and population
National outlets on the approved list with audience data.
Calculation
Unique users reached; denominator: eligible national outlets' total audience.
Time basis
Publication timestamp to 7 days after release; Europe/London.
Evidence lineage
Monitoring export; query version 4; schema 1.
Data treatment
Deduplicate by outlet and article URL; exclude missing audience data.
Accountability
Communications lead; reviewer: data manager; retain 24 months.
Interpretation
Compare against buyer-set tolerance; recheck if audience data changes.

Keep newsroom publication, manual outreach, technical transmission, paid placement, monitoring and independent editorial publication as distinct route values. A supplier's "accepted" status stays a supplier-defined label until contract and endpoint evidence establish its scope.

Assemble the blank decision view

Use text states as well as measures. That prevents an absent number being displayed as zero.

Blank decision view panels

Panel

Source-copy authority
Not entered
Buyer newsroom
Awaiting record
Manual outreach
No approved data
Technical transmission
Supplier evidence absent
Paid placement
Order not supplied
Monitoring
Query not registered
Independent publication
No verified observation

Current state

Source-copy authority
Unreviewed
Buyer newsroom
Unassessed
Manual outreach
Unreconciled
Technical transmission
Definition open
Paid placement
Classification open
Monitoring
Coverage unknown
Independent publication
Not evaluated

Quality flag

Source-copy authority
Decision
Buyer newsroom
Response
Manual outreach
Route
Technical transmission
Investigate
Paid placement
Hold
Monitoring
Define
Independent publication
Record only

Owner action

Source-copy authority
Buyer newsroom
Manual outreach
Technical transmission
Paid placement
Monitoring
Independent publication

Blank decision view

PanelCurrent stateEvidence cutoffQuality flagOwner action
Source-copy authority[not entered][timestamp][unreviewed][decision]
Buyer newsroom[awaiting record][timestamp][unassessed][response]
Manual outreach[no approved data][timestamp][unreconciled][route]
Technical transmission[supplier evidence absent][timestamp][definition open][investigate]
Paid placement[order not supplied][timestamp][classification open][hold]
Monitoring[query not registered][timestamp][coverage unknown][define]
Independent publication[no verified observation][timestamp][not evaluated][record only]

Do not add an overall success score. One favourable route cannot compensate for unresolved authority, advertising, privacy, security or correction evidence. For paid placement, apply CAP's recognition requirements to identifiable marketing, not performance.

Surface quality and correction work

Place freshness, missingness, duplicate treatment, reconciliation status and query version beside the relevant output, not in a hidden note. Create a correction queue with issue ID, affected measure and period, original view, evidence, owner, decision, replacement version, downstream recipients and closure state.

When personal data appears in an outreach or response record, the ICO's accuracy guidance says organisations should make the source and status clear and consider challenges to accuracy. The privacy owner must determine the actual UK GDPR route; the dashboard itself does not do so.

For the reporting environment, apply NCSC SaaS guidance to access control and activity logging. Security monitoring remains separate from communications reporting, but unauthorised changes to the reporting environment still need a correction record.

Verify current ICO, CAP and NCSC guidance on their official websites before relying on any cited requirement.

Apply change control

Every release needs a dashboard version, dictionary version, source cutoff, approver, open limitations and distribution list. Trigger review when a supplier renames a field, an endpoint population changes, a query is edited, a correction affects a past decision, or the intended use expands.

Hold the view if any displayed measure lacks a reproducible definition, complete population, denominator where required, source version or accountable owner. Publish internally only when the decision owner can trace every statement back to retained evidence. External publication needs fresh statistical, privacy, advertising, security, finance and editorial review; this draft remains on hold.

Before you act

  • Name the decision and the action if the gate passes or fails.
  • Create one dictionary row per measure before display.
  • Keep each route value distinct, including paid placement and monitoring.
  • Show quality flags and correction status beside the output.
  • Record dashboard version, dictionary version, cutoff and approver for each release.

Common questions

What belongs in the transformation layer?

The transformation layer is defined above under Establish three data layers.

Why avoid an overall success score?

See Assemble the blank decision view above. Each panel keeps its own state, quality flag and owner action.

When should the view be reviewed or held?

See Apply change control above for the review triggers and hold conditions.

More in Measurement