Use this register for technical conditions that create recurring cost, increase failure exposure, or constrain a valuable future option. Keep approved delivery tasks in the normal product backlog.
Portfolio view
| ID | Condition | Category | Exposure | Reach | Carrying cost | Treatment | Owner | Trigger | Review date | State |
|---|---|---|---|---|---|---|---|---|---|---|
| TD-001 | Example: incompatible request identifiers prevent reliable cross-service incident reconstruction | Operability | 16/25 | Platform | High | Reduce | Platform owner | Before the next product-area rollout | YYYY-MM-DD | Assessing |
Suggested states:
Observed— the condition has been captured but evidence is incomplete.Assessing— consequence, exposure, and treatment are being evaluated.Accepted— exposure is understood and has an explicit trigger and review date.Planned— a treatment has been approved and delivery work exists.Treating— repayment, reduction, containment, or retirement is in progress.Monitoring— treatment is complete; outcome signals are being observed.Closed— exposure has been removed or reduced to the accepted level.
Debt decision record
ID and short name
TD-___ — concise condition
Current condition
What is true today? Describe the constraint without committing to a preferred solution.
Consequence
What recurring cost, failure exposure, or blocked option does it create?
Evidence
List incidents, lead-time data, repeated manual work, support load, audit gaps, onboarding friction, or concrete examples. State the measurement boundary and identify estimates.
Scope and expected lifetime
- Critical user or business flows:
- Services or components:
- Teams and owners:
- Expected remaining lifetime of the affected capability:
Category
Select one primary category and any secondary categories:
- Delivery friction
- Reliability and performance
- Security and compliance
- Architecture and changeability
- Operability
- Knowledge and ownership
Exposure assessment
| Dimension | Score | Rationale |
|---|---|---|
| Impact | 1–5 | |
| Likelihood or frequency | 1–5 | |
Initial exposure (impact × likelihood) |
1–25 | |
| Reach | Local / product area / platform | |
| Carrying cost | Low / medium / high | |
| Option value | Low / medium / high | Which future decision is constrained? |
The numeric signal structures discussion; it does not replace judgment.
Decision owner
Who owns the next decision? This person does not need to implement all resulting work.
Treatment decision
Choose one:
Repay— remove the condition.Reduce— lower probability, reach, or carrying cost.Contain— limit blast radius or isolate the condition.Accept— keep it deliberately until a defined trigger.Retire— remove the affected capability.
Explain why this treatment is preferable to the alternatives.
Outcome hypothesis
What observable system behavior should change if the treatment works?
Possible signals:
- lead time in the affected area;
- repeated manual steps;
- incident investigation time or missing diagnostic context;
- change failures and rollbacks;
- active exceptions;
- ownership concentration;
- onboarding time;
- conformance to a platform contract;
- product work delayed by the constraint.
Trigger and review date
- Trigger that invalidates the current decision:
- Next review date:
- Evidence required at the next review:
Delivery links
Link epics, stories, ADRs, RFCs, incident records, dashboards, and rollout evidence only after the treatment has been selected.
Monthly review checklist
- Remove duplicates and conditions that no longer exist.
- Confirm evidence, owner, treatment, trigger, and review date for material entries.
- Escalate repeated local conditions into a product-area or platform decision.
- Revisit accepted debt whose trigger has fired.
- Move approved implementation into the normal delivery system.
- Check whether completed treatments changed the intended outcome signals.
Quarterly portfolio questions
- Where is exposure increasing despite local work?
- Which conditions create recurring cost across multiple teams?
- Which product bets depend on a technical option that is being lost?
- Which accepted risks have reached their trigger?
- Which capabilities should be retired instead of repaired?
- Are platform standards reducing repeated debt or merely centralizing it?
- Does investment match the highest exposure, or only the most visible requests?