Operations
Part of Running email software with a durable record for every campaign
Email service standards a buyer can own, with a clock and an exclusion list
Define buyer-owned email service measures with a population, unit, clock, exclusions, evidence, threshold method, owner and escalation route.
Email marketing software service standards should define how the buyer will measure and govern its own operation. A supplier's uptime term, a team's response target and a recipient outcome are different things and need separate records.
This England-focused method was prepared on 5 September 2026. It supplies no benchmark, promised turnaround, staffing ratio or acceptable defect rate. Qualified operational, financial, legal, privacy, accessibility, cyber and commercial owners must set thresholds from the actual risk and capability.
What to take away
- A supplier uptime term, a team response target and a recipient outcome are different things needing separate records.
- Every measure needs a definition, population, unit, clock and exclusions, plus evidence and ownership.
- An empty field means the measure is not ready, so do not fill it with an industry average.
- Thresholds come from harm, legal or contractual duty, tested capacity and the exposure owners accept.
- If the data source fails, report the measure unavailable rather than counting missing events as success.
Write a complete standard card
Every measure needs these fields:
Standard card fields
- Definitionevent or condition measured
- Populationall eligible records or events
- Unitcount, proportion, elapsed time
- Clocktimezone plus start, pause, stop
- Exclusionsnamed cases and approver
- Evidencesystem, query, log, retention
- Threshold methodrisk, obligation, capacity
- Ownership and escalationmeasurer, actor, warning, breach, stop, recovery
Field / Required entry
- Definition
- the event or condition being measured
- Population
- all records or events eligible for the measure
- Unit
- count, proportion, elapsed time or other stated unit
- Clock
- timezone plus exact start, pause, restart and stop events
- Exclusions
- named cases and who may approve them
- Evidence
- system, query, log version and retention location
- Threshold method
- risk, obligation, tested capacity and contract evidence used
- Ownership
- person who measures and person who can act
- Escalation
- warning, breach, stop and recovery route
An empty field means the measure is not ready. Do not fill it with an industry average that has a different population or clock.
Worked example: suppression propagation
The card below is illustrative and is not a benchmark. Threshold values are set by the accountable owner from harm, duty and tested capacity.
Field / Example entry
- Definition
- an authorised objection recorded in the source system until every in-scope destination confirms the new state
- Population
- all suppression changes sent from the source system to in-scope marketing destinations
- Unit
- count and proportion of changes confirmed within the clock
- Clock
- Europe/London; starts at source system objection timestamp; pauses while a destination is unavailable; stops at final destination confirmation
- Exclusions
- destinations under legal hold; changes reversed by an authorised compliance request; approver role named in the card
- Evidence
- suppression_events table; confirmation_state field; query
SELECT suppression_event_id FROM suppression_events WHERE confirmation_state <> 'confirmed'; retention location set by privacy owner - Threshold method
- harm of unlawful contact, ICO preference guidance, tested propagation time and exposure the owner accepts; no published operating target
- Ownership
- operations owner measures; data protection lead can act
- Escalation
- warning at owner-set failed confirmation level; breach at legal or contractual deadline; stop affected sends; recover from approved suppression snapshot
Separate four useful measures
Use the worked card above for suppression propagation. The ICO's preference guidance document, Direct marketing guidance: respect people's preferences, supports the underlying suppression control; it does not publish an operating target for this workflow.
Four measures compared
Suppression propagation
- Starts
- Objection recorded
- Stops
- All destinations confirm
- Population
- In-scope changes
- Evidence
- Confirmation query
Release quality
- Starts
- Defect found
- Stops
- Severity classified
- Population
- Signed content criteria
- Evidence
- Review record
Complaint clock
- Starts
- Defined intake event
- Stops
- Triage complete
- Population
- Intake messages
- Evidence
- Intake log
Recovery
- Starts
- Service state lost
- Stops
- Restoration evidence
- Population
- Affected systems
- Evidence
- Baseline config
A release-quality measure counts only defects found against the signed content, population, configuration and acceptance criteria. Classify severity before review. A late wording preference should not be mixed with an unauthorised recipient or broken opt-out route.
A complaint-handling clock begins at a defined intake event, not when a team happens to read the message. State office-hour treatment, duplicate handling, pauses awaiting evidence and the event that completes triage. Legal deadlines, contractual responses and internal service aims remain distinct.
A recovery measure identifies the service state to restore, the approved configuration baseline, required suppression data and acceptable evidence of restoration. NCSC SaaS guidance keeps customer configuration and recovery work in scope. A supplier uptime statement cannot prove that the buyer can reconstruct its audience safely.
Set thresholds without making up a norm
Start from the harm that a missed control could cause, any current legal or contractual requirement, tested staffing and system capacity, and the maximum exposure accepted by accountable owners. Preserve the source and date for each input. Run a rehearsal before approving the threshold.
A rehearsal is run by the named measure owner, on the last full period of real or representative events, with two operators independently calculating the same result from stored evidence. The pass condition is identical results, or a documented reason for any difference that the owner accepts.
The rehearsal tests whether the measure can be calculated consistently, but it does not set the threshold number. The owner sets the number from the risk inputs above; no industry default exists.
The government Service Standard applies to government services, and the Technology Code of Practice has its own public-sector scope. Their emphasis on ownership, accessibility, security, performance and reliable operation is useful as a method prompt. Neither supplies a private email team's target.
Connect measurement to action
Define a warning that prompts investigation, a breach that requires escalation and a stop condition that suspends further release. Name the recovery evidence and the person who can restart. If the data source fails, report the measure unavailable rather than treating missing events as successful ones.
Review a standard when workflow scope, supplier terms, audience volume, law, system design or evidence quality changes. Keep prior definitions so trend breaks remain visible.
Before you act
- Complete one card for suppression propagation and test whether two people calculate the same result from the stored evidence.
- Log release-quality defects against the signed content, configuration and acceptance criteria before review.
- Test the complaint intake clock with one duplicate and one out-of-hours example.
- Preserve the source and date for each threshold input.
- Run a rehearsal before approving any threshold.
Common questions
What should a suppression-propagation measure cover?
The worked card above is the single source for the definition, population, unit, clock, exclusions, evidence query and escalation. The ICO preference guidance supports the control but sets no operating target.
How should a buyer set thresholds without inventing a norm?
A threshold is a decision, not a published benchmark. Set it after a rehearsal that two operators can repeat from stored evidence, then record the source and date for each input.
When should a standard be reviewed?
Review when any of the change conditions above applies. Record the review date and keep the superseded definition so trend breaks stay visible. Name the person who can restart release and the recovery evidence before approval.
