Frame a Service Problem from the User Journey and Its Data
6 min read
Lesson objectives
- The learner writes a testable public-service problem statement with a denominator spanning all channels and distinguishes valid completion from merely starting a transaction.
Please sign in to use all the tools
Once you sign in, the reading tools are yours: larger or smaller text, typeface, line spacing, focus mode, highlighting and bold, and notes saved to your account and gathered in My study.
A digital public-service policy begins with the outcome a person should obtain, rather than a choice of a new screen. 'Increase portal use' does not say whether an applicant received a correct permit decision in reasonable time. Visits may rise because a form repeatedly requests information or because applicants return to repair an earlier error. Define an observable unit: one eligible application initiated by a person or representative during a specified period, followed by a valid decision without avoidable rework. Do not confuse applications with people; one person may submit twice. Start the journey at the first attempt to obtain the service, including telephone or in-person contact, and end it when a comprehensible, reviewable valid decision reaches the applicant. This boundary makes abandonment before an online form visible. The public outcome is not minimum elapsed time at any cost. Decision accuracy, accessibility, a route to challenge and applicant burden remain constraints alongside speed. This is a design method, not a claim that every jurisdiction gives the same legal entitlement; the policy team must check the competent authority's rules before drafting a binding rule.
Map five service stages: understanding eligibility, starting an application, supplying evidence, receiving a decision, and obtaining the result or requesting review. At each stage count entrants, exits and returns, with reasons where known. Record the channel at both start and completion. A transfer from web to telephone may rescue an application rather than represent a failed digital service, yet it adds time and effort. Beware the denominator 'people who reached submit': it excludes those who could not understand the rule or use digital identity. Choose a reasonably comparable cohort: one permit type, period and eligibility rule. If the rule changed mid-period, split the cohort or disclose the confounding change. Write a compact metric dictionary: numerator, first-attempt valid decisions; denominator, all eligible applications started through any channel; time window, one month; source, a linked record that removes duplicate channel transfers. Calculate separate channel and relevant user-group rates. An overall mean can hide deterioration for a smaller group, while a subgroup gap alone does not establish its cause. User research and examination of the actual journey are needed before choosing a remedy.
Administrative records are traces of work, not a complete picture of everyone who needed a service. Someone who never learned that the service exists is absent from an application register. A caller who gives up before identification may also be absent if calls are not logged. Present two quantities when necessary: completion among recorded starters and a separate estimate of reach into the intended population, which needs another source. Never use one as the denominator of the other without a defined link. Before declaring a policy problem, test rival explanations such as an onerous evidence rule, unclear instructions and slow staff review. A high return rate may be caused by the evidence requirement rather than the interface. Sample return reasons, check whether staff code them consistently, and compare channels over the same period. State what is observed and what remains unknown: association is not a causal finding. Possible interventions differ. Simplifying a proof rule is a policy decision; rewriting instructions is a content change; expanding assisted service is an operational choice. They may be combined, but locating the failure tells a team which intervention to test first and what evidence would change the judgment.
Turn diagnosis into a policy statement that can be challenged: 'For the measured month's eligible permit cohort, only 73.5% of all starters receive a first-attempt valid decision; the web rate is 70%, compared with 80% by assisted telephone and 90% on paper. We must identify return reasons and case mix before changing an evidence rule or closing a channel.' This bounds the cohort, denominator, result and remaining ignorance. Add a testable hypothesis: if avoidable document requests explain web non-completion, removing them with the policy owner's authorization should raise valid completion without increasing ineligible approvals. Test it against a sample of return reasons and staff decisions, not click counts. Specify a safeguard measure for erroneous eligibility decisions, a measure of assisted-channel workload, and a recovery path for people who stop. In a funding proposal, name a real alternative: simplify the current instructions and form while retaining assisted channels, rather than assuming new automation is the only comparison. If unrecorded starts are found, revise the estimate and explain the effect; do not extend the observed percentage to the entire population.
Lesson summary
Define the problem with a denominator covering all eligible starts, calculate valid completion by channel, and test the cause of any gap before changing a rule or withdrawing assistance.