The accumulated cost of past shortcuts in software and systems: design compromises that made earlier work faster and make all future work slower, riskier or more expensive.
Technical debt is invisible on a balance sheet and decisive in an estimate. Almost every "why does a small change take four months?" conversation is a conversation about it, whether or not anyone uses the term.
Last reviewed 3 July 2026 · Free and ungated
Operators who have modernised estates and bought companies carrying hidden liabilities can pressure-test what the plan is standing on.
Get independent perspectivesTechnical debt is the future cost created when systems are built with shortcuts: hard-coded logic, skipped tests, outdated components, integrations only one engineer understands. Like financial debt, some of it is deliberate and sensible, taken on to hit a date that genuinely mattered. The interest is paid in slower changes, fragile releases and rising run costs, whether or not the borrowing was ever recorded.
Debt surfaces in three rooms. Estimation: it is the gap between what a change should cost and what your systems make it cost, so plans priced from clean-system benchmarks will overrun. Platform decisions: replacement business cases understate the debt on the legacy side, or assume the new platform will not accrue its own. And diligence: an acquisition target's technology can carry liabilities no data room lists, which is what technical due diligence exists to find.
The term fails in opposite directions. Treated as a permanent excuse, it becomes a veto: every unwelcome request is "impossible because of tech debt", unquantified and unfalsifiable, and leadership eventually stops listening. Treated as an invisible liability, it never reaches a risk register until it prices itself into an outage, a security incident or a nine-month integration that was scoped at six weeks. The workable posture matches how financial debt is run: name it, size it roughly, decide which of it to service and which to live with, and make that a portfolio decision rather than an engineering complaint.