Fixed Regulatory Deadline Meets a 380-Item Backlog
Northbridge Mutual, a mid-sized property and casualty insurer, must retire its 17-year-old claims adjudication platform before a state regulator's 1 July 2026 deadline because the vendor will stop supporting the underlying database. The programme has 14 months, a budget of $6.2M, and three business units (Auto, Home, Commercial) whose claims processes overlap by roughly 60% but diverge on regulatory reporting. The programme has adopted Scrum, but the product owner has a 380-item backlog written as solution requirements ('build a rules engine screen'), and the steering committee wants a 300-page requirements specification signed off before any development. Two teams have started sprinting; a third is blocked waiting for enterprise architecture to approve an integration pattern. The regulator has said it will accept a phased rollout if statutory reports are produced accurately from day one.
1. The steering committee insists on a full requirements specification before development. Using agile analysis concepts, how should the lead analyst respond, and what analysis artefacts would you produce instead to satisfy the committee's need for confidence?
Read the worked answer
Start by separating the committee's real need - predictability and audit evidence - from the artefact they have requested. A 300-page specification written before development will be stale within weeks and will not give the regulator or the committee reliable evidence. Explain the cone of uncertainty and the cost of rework.
Instead, produce:
- A one-page vision and outcome statement tied to the statutory reporting obligation.
- A regulatory traceability matrix linking each statutory report to its source data, business rule, and acceptance test; this is the audit artefact, not prose.
- A release roadmap with fixed regulatory milestones and variable scope inside each release.
- An initial story map of the claims value stream, with epics and examples, and acceptance criteria elaborated just in time.
- A decision log and a lightweight business case showing cost of delay.
Reasoning: the committee is managing uncertainty and accountability. Give them transparent, testable evidence and a forecast range (for example, 70% confidence) rather than a false-precision document. Agree a definition of ready for compliance-critical stories so nothing enters a sprint without traceability.
2. How would you decide which business unit to sequence first, and what analysis would you do before the first release?
Read the worked answer
Assess each unit on value, risk, learning, and dependency, not on revenue alone. A simple weighted matrix could compare regulatory exposure, claims volume, process fit to the standard core, reuse of shared rules, and the integration-pattern blocker. Auto may have the highest volume and the most shared rules, making it a good learning slice; Commercial may be lower volume and reveal exceptions. Use cost of delay and weighted shortest job first to compare.
Before the first release, do enough analysis to reduce the largest uncertainties:
- Value-stream map the end-to-end claims process for the chosen unit.
- Model the statutory reports and data lineage.
- Identify non-functional requirements such as audit, retention, and response time as constraints.
- Map the integration pattern so the blocked team can proceed.
- Define the minimum releasable slice that produces accurate statutory reports.
- Run a pre-mortem with operations and compliance.
Reasoning: sequencing is a portfolio decision. Choose the slice with the best ratio of learning and risk reduction per unit of cost, and ensure the regulator's day-one reporting constraint is designed in, not retrofitted.
3. A business stakeholder demands that all 380 backlog items be estimated and committed for the whole 14 months. What do you do?
Read the worked answer
Treat this as a forecasting and trust problem, not a documentation problem. Explain that committing 380 items at this distance would require false precision and would remove the ability to respond to regulatory clarification or learning. Offer a forecast range and a decision cadence instead.
Practical response:
- Use story mapping to group the 380 items into releases aligned to regulatory milestones.
- Estimate at the epic level for the first two releases, not every story.
- Commit only to the next one or two sprints at story level; publish a rolling three-sprint forecast.
- Provide a probabilistic forecast (for example, 70% confidence by month 12) and a burn-up chart.
- Set explicit decision gates at months 3, 6, and 9 where scope can be rebalanced.
- Track work in progress and cycle time to show actual capacity.
Reasoning: the stakeholder wants certainty and control. Give transparency and early warning instead of a false commitment. A fixed 14-month commitment would encourage batch analysis, late discovery, and hidden risk - exactly the waste agile analysis seeks to avoid.