Productivity service standards that begin with user-critical moments

Business productivity software service standards turn vague expectations into decisions that users and operators can understand. They cover the service the organisation runs, not only the availability figure a supplier may publish.

This guide is for an England-based organisation setting local commitments. It does not replace a supplier agreement, legal duty or security requirement. Obtain specialist approval for those boundaries.

What to take away

Define standards around user-critical moments where delay or failure causes harm, not every click.
Keep supplier uptime commitments separate from internal operating standards and user-facing messages.
Choose a small balanced set of standards covering intake, ownership, access, quality, support, disruption and change.
Make measurement reproducible by documenting numerator, denominator, inclusion rules, time zone and paused states.
Review standards as the service changes and record definition changes with effective dates.

Begin with user-critical moments

List the points at which delay or failure causes harm: submitting priority work, assigning an owner, revoking a leaver's access, escalating a defect, recovering a necessary record and communicating during disruption. Avoid setting a target for every click. Standards are most useful where someone must decide or intervene.

Fields for each critical moment

Event that starts and stops the clock
Coverage hours and excluded periods
Population or priority class in scope
Source of the measurement
Named owner and escalation point
Target, tolerance and review date
Known limitations and approved exceptions

For each moment, capture:

the event that starts and stops the clock;
coverage hours and excluded periods;
the population or priority class in scope;
the source of the measurement;
the named owner and escalation point;
the target, tolerance and review date;
known limitations and approved exceptions.

Without those fields, two teams can report different results against the same phrase.

Separate supplier and customer commitments

A contractual uptime measure may exclude scheduled maintenance or depend on a supplier's definition. Your internal service can still fail while the platform is available because intake is unmonitored, an integration has stopped or nobody owns a queue.

Three rows to keep separate

Vendor commitment

Definition: Contractual uptime
Exclusions: Scheduled maintenance
Owner: Supplier
Source: Contract record
Interpretation: Commercial counsel

Internal standard

Definition: Operating standard
Exclusions: Intake, integration, queue
Owner: Internal team
Source: Operating record
Interpretation: Accountable owner

User-facing message

Definition: Plain promise
Exclusions: Not stated
Owner: Service owner
Source: Published standard
Interpretation: Service owner

Keep the vendor commitment, internal operating standard and user-facing message in separate rows. Link the contract or official service record and preserve its effective date. Commercial counsel should interpret credits, exclusions, notice and remedies rather than relying on a dashboard label.

Choose a balanced set

Start with a small group of standards:

Balanced set of standards

Intakeacknowledge complete request, identify gaps
Ownershipmaximum age without named owner
Accesstime to remove or change access
Qualityclosed items passing completion check
Supporttime to triage reproducible issue
Disruptionupdate frequency and restoration priority
Changenotice, testing, approval before release

Intake:time to acknowledge a complete request and identify missing information.
Ownership:maximum age of accepted work without a named accountable person.
Access:time to remove or change access after an approved trigger.
Quality:proportion or count of closed items that pass the defined completion check.
Support:time to triage a reproducible issue by severity.
Disruption:frequency of updates and priority for restoration or fallback.
Change:notice, testing and approval required before a material configuration release.

Do not insert universal numbers. Establish current performance from trustworthy records, assess user harm and resource capacity, then let the accountable owner approve a defensible target.

Make measurement reproducible

Describe the numerator and denominator, inclusion rules, time zone, paused states and treatment of reopened work. Preserve raw events where proportionate so an analyst can investigate anomalies. Averages alone can conceal an old tail of unresolved items; add a percentile, ageing band or oldest-item review when it answers the operational question.

Government guidance on performance metrics recommends starting with the service question, documenting data sources and interpreting figures with research. Its required KPIs apply to specified government services. Private teams can use the method without presenting a government measure as an industry benchmark.

Cover reliability and response

GOV.UK's standard for operating a reliable service asks public teams to plan for downtime, monitor performance and respond to problems. Use these prompts to check whether the local standard covers detection, decision, communication and recovery, not just supplier availability.

Link severity to user impact and data risk. A cosmetic defect and inaccessible priority workflow need different treatment even if both create one ticket. Keep a fallback for essential work and say who can activate it.

NCSC incident-management material emphasises maintained roles, plans and links to continuity and recovery. Security specialists should determine incident thresholds. Service standards should make escalation quick, not ask an operator to diagnose legal consequences before raising concern.

Review the standard as the service changes

Examine misses and apparent successes. A perfect acknowledgement measure may mean an automatic message fired, not that anyone took ownership. Ask users whether the commitment addresses their real difficulty and check for groups whose experience is hidden by an aggregate.

Revise a definition through controlled change. Record the old and new rule, effective date and reason; do not splice unlike periods into one trend. Trigger an early review after material supplier, integration, staffing, data or workflow changes.

Publish the concise standards people need and keep the technical measurement notes with the operating record. The goal is a clear promise, visible ownership and an evidence trail when reality differs.

Before you act

List user-critical moments where delay or failure causes harm.
Capture start and stop events, coverage hours and scope for each moment.
Separate vendor commitments from internal standards and user messages.
Document numerator, denominator and inclusion rules for each measure.
Link severity to user impact and data risk.
Record old and new rules with effective dates when revising definitions.

Common questions

What fields should be captured for each user-critical moment?

Capture the event that starts and stops the clock, coverage hours and excluded periods, population or priority class in scope, measurement source, named owner and escalation point, target, tolerance and review date, and known limitations and approved exceptions.

Why should supplier and customer commitments be kept separate?

How can measurement be made reproducible?

Describe the numerator and denominator, inclusion rules, time zone, paused states and treatment of reopened work. Preserve raw events where proportionate so an analyst can investigate anomalies. Add a percentile, ageing band or oldest-item review when it answers the operational question.