Business licences measurement: how to judge licensing services in 2027

Measurement should help a licensing team decide what to change, where a case is exposed and whether applicants can complete a task safely. It must not turn application volume into a success score or suggest a support service caused an authority decision.

This guide focuses on England and labels wider UK evidence. It uses sources checked on 5 September 2026. The 2027 date is a planning horizon, so service definitions, forms and official pages need to be confirmed again before measurement is used publicly.

What to take away

Write the decision before choosing a metric

State what the team will do differently if the number rises, falls or remains unchanged. A measure of failed authority routing can support content improvement. A count of conditions without owners can trigger operational review. Page views without a decision question are activity data, not a strategy.

Government Digital Service guidance on using data to improve a service links measures to user needs, baselines and improvement decisions. It is UK service-practice guidance rather than licensing law. Use it to shape the measurement plan while sourcing regulatory claims from the relevant authority.

For each metric, record its owner, population, period, event definition, source system, exclusions and action threshold. Define missing and unknown states so they are not silently counted as positive outcomes.

Worked metric card example

This illustrative card uses a wrong-authority routing problem. It is a design example, not a published benchmark.

Model the licensing journey accurately

The journey may include activity described, possible route found, authority confirmed, evidence started, submitted, acknowledged, decided, conditions implemented and renewal scheduled. Each event needs evidence. A click is not an application; a submitted form is not a grant; a grant does not prove that operating conditions are working.

The Department for Business and Trade's Find a licence service searches official records by activity and place across the UK and warns that its catalogue may not include every licence. A completed search can be measured as discovery activity. It cannot be counted as a confirmed requirement or successful application without additional evidence.

Operator, establishment and activity details are used within the Food Standards Agency's food-business registration route to direct the case appropriately. Service teams can examine completion and assistance at those task stages where they lawfully hold suitable data. They should not infer acceptance or other permission outcomes from form progress.

Two England examples show why event definitions differ. A premises licence under the Licensing Act 2003 is decided by the local authority licensing committee and can carry conditions. A scrap metal dealer licence under the Scrap Metal Dealers Act 2013 is issued by the local authority for a site or mobile collector.

The Environment Agency public register of environmental permits is another authority data source with separate permit types and statuses. Each regime has its own application, decision and condition events, so one journey model cannot be copied across them.

Choose measures across four layers

User measures show whether people can identify the right route, understand the information requested and recover their case. Operational measures cover ownership, evidence quality, handover and condition control. Outcome measures may include completed authority decisions or resolved changes, with clear boundaries. Guardrail measures watch for privacy problems, unsupported advice or faster processing that damages record quality.

Use a balanced set rather than one headline rate. A team could reduce completion time by skipping checks, so speed needs an accuracy or rework companion. A high digital completion rate may hide applicants who could not use the standard channel.

Avoid making approval rate the main service target. Authorities decide cases on their facts, and the applicant population may change. If approval is reported, describe the case definition, exclusions and inability to attribute the decision solely to the support service.

Create a reliable case denominator

Decide whether the denominator is people, organisations, premises, applications, permission rows or support interactions. One company may have several establishments and activities. One person may return many times. Mixing units can create a plausible but meaningless percentage.

The Department for Environment, Food and Rural Affairs' animal activity licensing process describes application, inspection, decision, renewal and enforcement for that English regime. Within that activity, stage measures can follow the documented lifecycle. Do not apply its denominator or sequence to unrelated permissions.

Maintain a data dictionary with identifiers, event rules and source fields. Version it when the service changes. Historical charts should mark the break rather than comparing unlike definitions as one trend.

Design a dashboard for action

Begin with decisions and exceptions. Show unresolved authority questions, cases awaiting internal evidence, conditions without owners, approaching renewal reviews and missing source dates. Provide drill-down to the controlled record for authorised users instead of copying sensitive documents into the dashboard.

Sample dashboard card layout:

Each card needs a title, definition, period, denominator, data owner, last refresh and known limitation. Use counts alongside rates where small denominators could mislead. Separate England from other nations when the underlying route differs.

Status colour should reflect consequence, not only lateness. One unresolved permission blocking the principal site may matter more than many minor administrative tasks. Show the owner and next decision beside the exception.

Measure quality and rework

Create defect categories such as wrong authority, incorrect activity, inconsistent applicant details, missing attachment, unsupported date or status error. Record where the defect entered the case and which quality gate found it. This helps the team improve instructions or training instead of merely correcting files.

Sampling can monitor repetitive low-risk fields, but declarations and other high-consequence facts may require full checking. Publish the sampling method and expand review if repeated defects appear. A falling defect rate is meaningful only if scope and detection remain comparable.

Track recoverability by asking a deputy to find the authority, submitted version, decision, conditions and next date. A successful handover is evidence of operational resilience, not a cosmetic documentation score.

Attribute contribution cautiously

Different methods answer different questions. Before-and-after analysis can show change over time but may be affected by season, case mix or a new authority route. Cohort comparison can separate users exposed to a service, but selection may differ. Process tracing follows evidence from intervention to immediate effect and is often more honest for small services.

Government Digital Service guidance on planning user research can support interviews and observed tasks that explain why a metric moved. Qualitative evidence can identify plausible mechanisms; it does not supply a national prevalence estimate from a small sample.

Do not credit a content page, tool or adviser with every grant or all revenue after launch. State the contribution measured, alternative explanations and evidence missing.

Handle commercial claims separately

A licensing provider may want to advertise speed, savings, coverage or success. The Advertising Standards Authority's CAP Code section on misleading advertising covers substantiation, material information and comparisons in non-broadcast marketing. The measurement method must match the public wording.

Define start and finish points for turnaround, the counterfactual for savings and the population for success. Exclude authority time if the claim concerns the supplier's internal service, or label it accurately. Preserve negative cases and do not change exclusions after seeing the result.

Editorial dashboards and advertising evidence should not be conflated. A useful internal indicator may be too provisional or narrow for a public comparative claim.

Benchmark like with like

The Office for National Statistics publishes business activity, size and location tables with stated definitions and regional breakdowns. These data can describe business populations; they do not count universal licence applications or support demand.

For service benchmarks, match activity, geography, case stage, channel, time period and denominator. A multi-site animal-licensing portfolio should not be ranked against first-time food registrations using one blended completion rate. Local authority processes may also vary, so identify the council through the GOV.UK local authority search and keep area context visible.

Use ranges and distributions where averages hide case mix. Report sample size, missing data and definition changes. If evidence is too thin for a league table, publish descriptive findings instead.

Illustrative only: if team A records 60 wrong-authority routing cases in 400 first-time enquiries, that is 15 per cent. If team B records 20 in 400, that is 5 per cent. The distribution spans 5 per cent to 15 per cent. That spread matters more than one blended average.

Build the reporting cycle

Review operational exceptions frequently enough to protect active cases. Analyse service trends on a stable period and conduct deeper case review when a metric crosses its threshold. Assign every action and record whether it changed the process.

At each cycle, confirm that the measure remains connected to a decision. Retire metrics that no longer influence work. Add new ones only with definitions, data controls and a plan for interpretation.

Reopen source pages before publishing current claims. If an official service changes its fields or scope, annotate the measurement break and reassess historical comparisons.

Audit the story told by the numbers

Ask whether the title matches the unit, whether the denominator excludes unknowns, whether the geography is explicit and whether a correlation has been described as causation. Check that charts do not suppress negative cases or shifts in scope. A colleague should be able to reproduce the result from the documented rules.

The immediate next step is to select one decision, such as reducing wrong-authority handoffs, and write a metric card. Define the event, denominator, evidence, threshold and safeguard. Test it on a small set of real records before building a dashboard. This guide remains on editorial hold pending human fact-checking and review.

Before you act

Common questions

What should be defined before selecting a metric?

State what the team will do differently if the number rises, falls or remains unchanged. A measure of failed authority routing can support content improvement, while a count of conditions without owners can trigger operational review. Page views without a decision question are activity data, not a strategy.

How should the licensing journey be modelled for measurement?

The journey may include activity described, possible route found, authority confirmed, evidence started, submitted, acknowledged, decided, conditions implemented and renewal scheduled. Each event needs evidence.