"Run a technical debt registry like a risk register"

Most organisations believe they track technical debt because their issue tracker has a tech-debt label. Look inside that label and you will find the same thing everywhere: hundreds of tickets, no owners, no costs, sorted by age, read by nobody. A backlog is where debt goes to be forgotten.

A registry is a different instrument. Companies already know how to run one — every risk department does. A risk register works because each entry is forced through the same questions: what is it, what does it cost us, who owns it, and what did we decide to do about it. Technical debt deserves exactly that treatment, for exactly the same reason: it is a liability the organisation is carrying whether or not anyone writes it down.

What a registry entry contains

Six fields do the work. Everything else is decoration.

Description — in system terms, not code terms. "The scheduler and the transport layer share state, so they cannot be released independently" is a registry entry. "Refactor the scheduler" is a wish. The test: someone outside the team should understand what is constrained and why it matters.

Cost. The estimated cost to close the item — an honest range, not wishful thinking. This number is allowed to be large; if it's small — why not fix it immediately?

Interest. What the item costs while it stands: where delivery is slower, which defect class keeps recurring, what every integration pays again. This is the field organisations skip, and it is the only one that turns the registry into a management tool. Interest is paid every single day, in the currency that is actually being tracked — team velocity, time to resolution.

Trigger conditions. The future event that changes the decision: "if we add a third product variant, this becomes blocking", "if this module needs certification, the shortcut fails audit". Triggers are what make accepted debt safe to accept — they define when accept expires.

Owner. A name. Not a team, not a guild. Unowned debt is unmanaged debt.

Decision. The standing verdict, dated: pay down (scheduled against the budget), accept (with triggers), or watch (interest unclear — measure it). The decision field is the entire point of the registry. Every item has one, which means nobody can later claim the debt was invisible. It was on the ledger, and the organisation chose.

Ok, there may be a 7th field — Reason — why the actual shortcut has been taken. It's helpful for tracking the psychology of the organisation, but it does require a mature approach to managing technical debt, or reverse-engineering of decision making.

Cadence: portfolio review, not confession

A registry that is written once is a backlog with more data. The registry that is being worked on a cadence lives — once a quarter, align the registry review with roadmap planning. The review is portfolio management: has any interest estimate changed, has any trigger fired, does the pay-down budget go to the highest-interest items, does the work down the road trigger any registry items. Thirty minutes with the right five people, not an all-hands lament.

Anti-patterns worth naming

The amnesty sprint. A one-off purge applied to a recurring liability; covered in the previous post.

No owner. See above. This kills more registries than any tooling choice.

Treating all debt as bad. Some debt is deliberate leverage — taken to ship, with a recorded trade-off and a trigger. The registry is not a wall of shame; it is the record that separates leverage from loss.

Start with ten items

Do not inventory the whole system — a complete registry is a stalled registry. Take the ten items engineers complain about most, force each through the six fields, and hold one review. The first review usually retires two items nobody still cared about, prices three that turn out to be expensive, and produces the first genuinely informed decision the organisation has made. That is more debt management than most companies have ever done.

The tooling does not matter — a spreadsheet with six columns and a review log is enough. The cadence is what does the work.

The registry is one of three moves — visible, priced, decided — from my webinar on managing technical debt: watch it here.