Software is the only major asset class that companies routinely pretend is finished. The launch party happens, the team disbands, and the system begins its real life — meeting traffic patterns nobody predicted, dependencies that age monthly, and a threat landscape that never sleeps. Maintenance is not the boring epilogue to development. It is the longest and most consequential phase of a system’s life.
The industry’s own numbers make the case: the majority of software’s lifetime cost is incurred after launch, and the majority of breaches exploit vulnerabilities for which patches already existed. Neglect is not a savings. It is a loan, at compound interest, from a lender who chooses the repayment date.
What rot actually looks like
Nothing dramatic happens at first. A dependency falls a major version behind, then three. A framework reaches end-of-life and stops receiving security patches. Performance degrades a few percent a quarter as data grows. Then something forces the issue — a vulnerability, an incompatibility, a feature that cannot be built on the old foundation — and the accumulated debt comes due at crisis prices.
The tragedy is that every step was cheap at the time. Updated monthly, dependencies are minor chores; updated after five years, they are a migration project with a war room.
Maintenance as practice, not promise
Real maintenance is scheduled, not reactive: dependency updates on a cadence, security patches on an SLA, monitoring reviewed by a human, backups restored — actually restored — on a rehearsal calendar, and small refactors that keep the codebase legible as the team changes. It also includes honest capacity: someone on the hook when the alert fires, with the context to act.
This is what an SLA-backed maintenance engagement buys: not insurance paperwork, but the ongoing engineering attention that keeps a system in the cheap-to-fix zone permanently.
The asset view
Maintained software appreciates in usefulness: it stays secure, stays fast, and — critically — stays buildable-upon, so next year’s feature lands in weeks instead of waiting on a rewrite. Unmaintained software converges on the same fate at different speeds: the big-bang replacement, the most expensive project in software.
The most underrated engineering discipline is the one that makes rewrites unnecessary. That is the whole pitch, and it is enough.
