Three Questions to Ask When You Inherit a Troubled Program
I've been brought into broken, late, and over-promised situations more times than I can count. Here's what I've learned — the hard way — about what actually works when you're the person expected to fix things.
I've frequently been brought into situations where something was broken, late, or over-promised. Problems existed well before anyone fully understood what delivering it — whatever it was — would actually require. I've made some of the mistakes that come with being the person expected to fix things, and I've learned the hard way what works.
What were you asked to promise before you understood the project?
Early in my career, when an Oracle database or HP-UX system went down, nobody wanted a philosophical discussion about the problem. They wanted it running. I developed an instinct for responding immediately, and that instinct served me well for years.
Unfortunately, I carried it into executive leadership. When someone handed me a troubled project with an existing deadline, I initially treated the date with the same urgency as an outage. It took time to recognize the fundamental difference. A failed production system is an observable condition. A promised delivery date is often somebody's expectation, built on assumptions you haven't had the chance to examine.
I spent years learning how to recover systems quickly. It took considerably longer to recognize that accepting someone else's untested assumptions wasn't recovery — it was how the next failure began.
Which discovery changed your initial assessment most?
The most valuable discoveries rarely came from the executive who asked me to fix the problem. They came from the people who had been living with it.
Engineers know which parts of the platform are fragile. Operations teams understand the workarounds keeping services running. Product teams know which requirements changed without corresponding changes to delivery expectations. I've seen this pattern across legacy technology, regulatory initiatives, and newer healthcare platforms. What appears to be a technical failure often turns out to be a series of decisions and compromises that nobody has examined together.
What I've learned: if you don't make time to understand the people doing the work, you'll eventually spend that time explaining why the work didn't get done.
There's a second problem. By the time you're called in, those people may have stopped believing anyone wants to hear the answer.
What would you handle differently in your first conversations?
I'd spend less time asking when the project will be finished and more time understanding how everyone arrived at the current expectations.
When I helped advise on vaccine scheduling capabilities at CVS, the work reinforced something I'd seen throughout my career: delivering software is only a small part of delivering a functioning service. A scheduling platform ultimately depends on the availability and coordination of resources well beyond the application itself.
The same applies to regulatory projects and ambitious startup products. Executives can promise a release without fully appreciating everything that must happen for that promise to become reality.
Knowing what I know now, I'd ask the people closest to the work what they feel — or can prove — that hasn't made it into the executive conversation.
The most uncomfortable part of recovering a troubled program, and the most overlooked, is often discovering that the deadline isn't the real problem. The real problem is that everyone has been working toward a different understanding of what was promised.