Decision Proposal Writer

We've shipped three releases this year that needed same-week hotfixes, and each one has cost roughly two engineer days to fix on top of the support load that comes with it. Right now engineers are testing their own work and each other's, which is slow and, more importantly, catches the wrong class of bug: the person who wrote the code is not well positioned to find the problems in it. Last quarter this pattern ate about 15% of engineering time in unplanned fixes. It's also changing behavior in a way that's hard to reverse quietly: the team has started padding estimates because they expect rework, which means our planning numbers are already absorbing the cost of a problem we haven't fixed. We looked at four options. The first is hiring a dedicated QA engineer at roughly 55k plus the usual overhead. The second is buying an automated testing tool, around 8k a year, and having engineers write more tests themselves. The third is contracting a testing agency per release, quoted at about 3k per release, or 12k a year. The fourth is doing nothing and accepting the current hotfix rate. The tool doesn't address the actual problem. Engineers would still be testing their own work, just with better instrumentation, and the core issue isn't a lack of tooling, it's that the wrong people are doing the testing. The agency option fixes the "wrong person" problem but introduces a new one: an outside agency has no context on our codebase, our past incidents, or the tricky parts of the system that tend to break, so they're realistically only going to catch shallow issues. Doing nothing means we keep absorbing the 15% cost as a fixed feature of how we work, plus whatever it's doing to morale and estimate accuracy, which is harder to put a number on but is clearly showing up in how the team plans its time. The recommendation is to hire. A dedicated QA engineer solves the actual root cause, which is that the people building the software are also the only ones checking it. Someone in the role full time will build the context that a per-release agency never gets the chance to build, and will catch the kind of bug that comes from familiarity with the system rather than a generic checklist. On cost alone, the hire pays for itself if it recovers even half of the 15% of engineering time we're currently losing to unplanned fixes, and that's a conservative bar given the scope of what's going wrong. The ask is approval to open a headcount for a QA engineer at roughly 55k plus standard overhead. Once approved, we'll move directly into writing the job description and opening the search. We'd like a decision by [insert date] so we can factor the hire into next quarter's planning and start closing the gap before another release needs a hotfix.