Eight steps for managing technical debt
Record technical debt by its effect on future work, prioritize evidence, plan small repairs and verify that changes reduce friction.

Technical debt describes design or implementation choices that make future changes harder or more expensive. A deliberate shortcut can meet an immediate need, but the follow-up work should remain visible.
The Software Engineering Institute’s technical-debt research discusses explicit management of these trade-offs. For a small business, start with a record of the friction and the change needed to reduce it.
1. Describe a specific problem
Write down the affected system and example: a change requires editing the same rule in three places, or a release depends on an undocumented manual step. Keep these items in one list (a technical debt register) so they can be ranked and tracked.
Avoid classifying every old tool or feature request as debt. A bug, an unsupported component and a missing feature each need a different decision, so record them separately.
2. Record the continuing cost
Gather examples of repeated work, delays, incidents or difficult handovers. Separate measured effort from an estimate and record where the estimate came from.
A plausible future risk can still matter, but it should be labelled as such. Don’t turn an inconvenience into an invented savings figure.
3. Rank with the work the business needs
Consider dependencies, consequences of failure and planned changes. A repair that removes a barrier to the next project may deserve priority over a cosmetic cleanup.
Discuss the trade-off with the people using and maintaining the system. Record why an item is scheduled, deferred or accepted temporarily.
4. Assign bounded repair work
Give each selected item an owner, scope and acceptance check. Set aside time based on your own backlog and priorities. No standard percentage fits every team.
Include any supplier participation, outage window or records migration. If the work is too large to review, identify a smaller useful step.
5. Document what the next person needs
Keep a short description of the system, dependencies, configuration ownership and recovery process. Record why a temporary arrangement exists and when it should be revisited.
Have someone other than the author follow a handover or recovery procedure. Their questions can reveal missing assumptions. Keep passwords and keys in your password manager or secrets vault, never in the handover document.
6. Preserve behaviour while improving structure
For application changes, identify the business behaviour that must remain correct. Use relevant tests and representative examples before changing the internal design.
Make the repair small enough to verify and reverse. Refactoring shouldn’t become an unplanned change to what customers or staff can do.
7. Manage dependencies and security together
Record components and support conditions. Test updates against the workflows and integrations that rely on them. The newest version isn’t automatically compatible with the rest of the system.
The Cyber Centre’s patching guidance supports a planned process for security updates. Don’t postpone an urgent exposure simply because broader cleanup isn’t ready.
8. Verify that the repair helped
Repeat the original example and compare effort, failure points or handover quality. Close the item with evidence of the improvement and any remaining limitation.
Update the register when new compromises are made. Give temporary decisions an owner and review trigger so the cost remains part of ordinary planning.


