Operations
A blank measure record for buyer-owned PR agency service standards and anti-gaming tests
A blank measure record for buyer-owned PR agency service standards, with fields, evidence, clocks, exclusions and anti-gaming checks to complete yourself.
PR agencies service standards need buyer evidence, not borrowed response times. The templates below contain no target, tolerance or benchmark. An organisation operating in England should fill them from its risk, baseline, contract and tested capacity, then have the relevant owner approve each measure.
What to take away
- Buyers must fill service standards from their own risk, baseline, contract and tested capacity, not borrowed benchmarks.
- Each measure needs a complete record covering event, population, counts, clock, exclusions, evidence, ownership, uncertainty and control.
- Do not compare percentages with different denominators or elapsed time with different clocks.
- Test whether a standard could reward the wrong behaviour, such as logging fewer challenges to improve a correction measure.
- Stop reporting a measure if the denominator cannot be reconstructed or the source loses integrity.
Use a complete measure record
Every standard needs these fields:
Complete measure record fields
- Service eventobservable state change
- Eligible populationwork items entering measure
- Numerator and denominatorexact counts
- Unitpercentage, time, count
- Clockstart, stop, pause, timezone
- Exclusionspre-agreed omitted cases
- Evidencesource system, field, query version
Field / Blank definition to complete
- Service event
- The observable state change being measured
- Eligible population
- Which work items enter the measure
- Numerator and denominator
- Exact counts or quantities and their relationship
- Unit
- Percentage, elapsed working time, count or another defined unit
- Clock
- Start, stop, pause events, timezone and working calendar
- Exclusions
- Pre-agreed cases omitted and the reason
- Evidence
- Source system, field, query version and retained artefact
- Ownership
- Service owner, evidence owner and decision authority
- Uncertainty
- Missing records, late entry, ambiguity and known bias
- Control
- Target, tolerance, response, stop rule and review date
Do not compare percentages with different denominators or elapsed time with different clocks.
Candidate standards
Factual approval completeness
Event: a work item reaches release; buyers define eligible items and required evidence. The numerator could be released items with a signed factual owner, claim sources and exact approved version; the denominator could be all eligible releases.
Candidate standards compared
Factual approval
- Event
- Release reached
- Evidence
- Signed owner, sources
- Target
- Blank
- Authority
- CAP substantiation
Contact-control
- Event
- Outreach record
- Evidence
- Purpose, source, checks
- Target
- Blank
- Authority
- ICO right-to-be-informed
Correction-queue
- Event
- Challenge enters queue
- Evidence
- Client evidence
- Target
- Blank
- Authority
- Factual authority
Leave target and exception policy blank. CAP's substantiation rules support evidence for objective advertising claims within scope, not a universal PR quality percentage.
Contact-control completion
Define the population of intended outreach records. Evidence may include purpose, source, recipient type, channel, transparency and suppression checks. The ICO's right-to-be-informed guidance says that the source should be identified specifically where personal data came from elsewhere. It neither sets a processing target nor proves that outreach is permitted.
Correction-queue handling
The event might start when a challenge enters the controlled queue and stop when the factual authority accepts containment and every affected client-controlled version is addressed. Set pause rules for missing client evidence. Do not include a publisher's independent correction time in the agency clock unless the contract clearly separates the agency's action from the publisher's decision.
Accessible-format acceptance
Use the exact artefact and required task population. The numerator may count versions that pass buyer-specified tests; the denominator is all versions that require those tests. Record assistive technologies, tester, defects and retest version. Government inclusive communication guidance provides public-sector prompts, not a private universal SLA.
Incident and recovery evidence
Measure a defined supplier-assurance and recovery state, not absence of incidents. The NCSC's supply-chain guidance supports ongoing reporting, review and contractual assurance. Buyer fields should identify the alert source, containment authority, evidence captured, recovery test and restart approval. The required time and tolerance remain blank.
Commercial reconciliation and exit
Define which invoice lines must match authorised units and accepted artefacts, and which exit items require archive, export, access removal, deletion evidence or continuity transfer. Unknown or disputed lines stay visible. A low operational score cannot waive a legal, privacy, rights, accessibility or security failure.
Guard against a measure becoming the job
Test whether the standard could reward wrong behaviour. A short drafting clock may encourage work to begin before the evidence pack is complete, a high release count may favour volume over relevance, and a correction measure may look better if staff do not log challenges.
Where gaming is plausible, keep a balancing indicator and an audit sample. Let the buyer's consequence analysis set the response.
Review a standard when scope, channel, supplier, system, law, risk or evidence query changes. If the denominator cannot be reconstructed or the source loses integrity, stop reporting the measure until it is repaired and re-baselined.
Before you act
- Fill every field in the measure record from your own evidence.
- Have the relevant owner approve each measure.
- Check that denominators and clocks match before comparing.
- Add a balancing indicator and audit sample where gaming is plausible.
- Review the standard when scope, channel, supplier, system, law, risk or query changes.
- Stop reporting if the denominator cannot be reconstructed.
Common questions
What fields must every service standard include?
Every standard needs a service event, eligible population, numerator and denominator, unit, clock, exclusions, evidence, ownership, uncertainty and control. The control field covers target, tolerance, response, stop rule and review date. Buyers must complete these fields from their own risk, baseline, contract and tested capacity.
How can a measure reward the wrong behaviour?
A short drafting clock may encourage work to begin before the evidence pack is complete. A high release count may favour volume over relevance. A correction measure may look better if staff fail to log challenges. Keep a balancing indicator and an audit sample where gaming is plausible.
When should a standard be reviewed or stopped?
Review a standard when scope, channel, supplier, system, law, risk or evidence query changes. If the denominator cannot be reconstructed or the source loses integrity, stop reporting the measure until it is repaired and re-baselined. The buyer's consequence analysis sets the response.
