Recovering a Project That's Gone Badly Wrong

The dashboard still shows green. The status update still says "on track, minor delays." Meanwhile three people on the team already know, in a way nobody has said out loud yet, that the current plan isn't going to get the project where it needs to go.

Why does generic project-recovery advice fail here?

"Have a tough conversation and reset expectations" names the right general direction without saying what actually needs to happen in the room. It skips the hardest part: getting an honest read on the real state of the project when everyone involved has some incentive, professional, emotional, political, to report things as better than they are.

Generic AI prompting tends to default to generic project-management language, "realign stakeholders," "recalibrate the timeline," that sounds like a plan without addressing why the status reports were optimistic in the first place. Ask an assistant to "help me get this project back on track" without more context, and it will produce a template recovery plan built on the assumption that the status information feeding it is already accurate.

Few status reports are deliberately dishonest; most are shaded slightly optimistic by people who genuinely believe the gap is still closeable. The deeper problem is that most project failures aren't sudden; they're the compounding result of small optimistic reports that each seemed reasonable at the time. By the point the failure is undeniable, the gap between the plan and reality is usually much larger than anyone individually realized.

Each individual status report along the way was rarely a lie; it was usually a reasonable-sounding account that left out the specific worry nobody wanted to be the one to raise. The accumulation of those small omissions is what produces the sudden-feeling crisis at the end.

What does the documented crisis-leadership record say to do?

Gene Kranz: demand an honest status before deciding anything

Kranz's documented standard at NASA mission control, "tough and competent," meant facing the real state of a crisis with complete honesty before deciding on a response, and demanding the same honesty from every specialist in the room regardless of how uncomfortable the real numbers were. The plan can only be as good as the information it's built on.

Applied: Before drafting a recovery plan, run an explicit status round where each person states the real state of their piece of the work, not the optimistic version, with no immediate pushback or judgment allowed during the round. The goal is an accurate picture first; the plan comes second, built on what's actually true rather than what was hoped. A plan built on an optimistic picture tends to fail again in a new, slightly different way, since it was never addressing the actual constraint.

Kranz's own account of the Apollo 13 recovery underlines why this sequencing matters: the eventual solution only became visible once the team had an accurate, shared picture of exactly what resources and constraints they were actually working with, not the picture anyone had wished for when the mission began. No amount of ingenuity in the response stage could have compensated for an inaccurate picture at the start of it.

Ernest Shackleton: focus the group on the next actionable step, not on what's already failed

Shackleton's documented leadership during the Endurance expedition, after the ship itself was lost, focused the crew's energy entirely on the next concrete, survivable step rather than on dwelling on the expedition's failed original goal. The situation was genuinely dire; the response was relentlessly forward-focused rather than backward-looking.

Applied: Once the honest status is on the table, resist the pull to spend the meeting re-litigating how the project got here. Name the single most important next action given the real situation, and move the group's energy toward that, saving a fuller retrospective for after the immediate crisis has passed.

This sequencing matters because dwelling on blame or past decisions while still inside the crisis consumes energy and attention the recovery itself needs. Shackleton's crew survived in part because the group's focus stayed relentlessly on what was actionable today, not on what should have gone differently in hindsight. The discipline wasn't denial; it was a deliberate choice about where to spend finite attention during the crisis itself.

Amy Edmondson: make the post-mortem blameless so the real causes surface

Edmondson's The Fearless Organization (2019) documents psychological safety as the precondition for a team surfacing the real, often uncomfortable causes of a failure rather than a sanitized version designed to protect individuals from blame. A blameless post-mortem, conducted once the immediate crisis has passed, is what actually prevents the same failure from recurring.

Applied: After the recovery, run a structured review focused explicitly on the process and decisions, not on which individual made which call. Ask what information was available at each decision point and why the choice made sense at the time, rather than asking who's responsible for the outcome. Framing the questions this way usually surfaces more useful detail than a blame-oriented review ever does, because people aren't spending their energy protecting themselves while answering.

Where does each framework fit best?

What can you actually try this week?

Why it works: specific, fact-based questions are harder to answer with comfortable vagueness than an open "how's it going," which tends to invite an optimistic summary by default.

Why it works: isolating the next action from the history of how things went wrong keeps the immediate response focused on what's actually changeable right now.

Why it works: structuring the questions around information and process, rather than people, makes it easier for participants to be honest about what actually happened.

What won't a crisis framework fix?

These approaches improve how a team gets an honest picture, stays forward-focused under pressure, and learns from what happened afterward. None of them can recover a project where the underlying resources genuinely don't exist, or where the goal itself was never realistic regardless of execution. Sometimes the honest outcome of a clear-eyed status check is that the project needs to be rescoped or ended, not rescued, and recognizing that early is itself a form of successful recovery, even though it doesn't feel like one in the moment.

Gene Kranz and Ernest Shackleton both sit in the Explorer & Adventurer category, and Amy Edmondson sits in the People & Culture category. For diagnosing after the fact whether a decision along the way was actually sound or just lucky, What Actually Happened is a $79 tool built for that specific judgment.

All the copy-paste prompts from this guide, and the rest of the Situations series, live at /prompts too, free to copy without reading the guide first.