A stalled pilot rarely dies. It lingers — still technically running, still on a slide somewhere, quietly consuming a fraction of someone's week while producing nothing anyone can point to. Eighteen months later it's mentioned in past tense and nobody can say exactly when it ended.
We've now seen enough of these to notice that they fail in a small number of recognizable ways, and that each one is visible early if you know the signal. Here are the five.
1. Pilot purgatory
The pilot works. Everyone agrees it works. And it never scales, because no one defined what "working" would have to mean to justify the next investment.
The signal is early and specific: at kickoff, ask what number would have to move, by how much, for this to proceed to production. If the room can't answer in one sentence, you're already in purgatory — you just haven't arrived yet. A pilot without a pre-agreed success threshold cannot succeed, because success was never defined. It can only produce interesting results, and interesting results don't unlock budget.
The fix
Before the pilot starts, write down: the metric, the current baseline, the target, the measurement window, and who decides. One paragraph. Get it agreed by whoever controls the budget for phase two.
2. The orphaned champion
One person drove the initiative. They were genuinely good at it. Then they were promoted, reorganized, or left — and the effort quietly stopped, because it had one point of failure and that point was a human being.
The signal: ask who would keep this running if the champion were unavailable for a month. If the honest answer is nobody, the initiative isn't a program, it's a personal project with a budget code.
The fix
Two named owners minimum, documented decisions, and the outcome written into a performance objective rather than living in someone's enthusiasm. Enthusiasm is not a governance model.
3. The tool nobody adopted
It was bought, deployed, announced, and then used by eleven percent of the intended users. Usually because it made someone's job harder in a way nobody asked about beforehand, or because it required a behavior change that no one was accountable for driving.
The signal: whether anyone who actually does the work was in the room during selection. If the tool was chosen by a committee of people who won't use it, adoption is already in trouble.
The fix
Involve two or three end users in evaluation with real influence, not as a courtesy. Then track adoption weekly for the first quarter and treat a dip as an incident rather than a communications problem.
4. The data reckoning
Halfway through, someone discovers the data is inconsistent, incomplete, or defined three different ways across three systems. The project quietly becomes a data project, which nobody scoped or budgeted, and which is far less exciting than what was approved.
This is the most common one. It's also the most predictable, which makes it the most frustrating. In survey after survey, data quality and availability lead the list of reported obstacles — roughly half of businesses name it as their primary barrier.
The fix
Assess data readiness before committing to an approach, not after. It's a week of unglamorous work that routinely saves a quarter, and it occasionally reveals that the fastest path to value is fixing the data and doing nothing else.
5. The impressive irrelevance
The pilot chosen was the one that demoed best to leadership rather than the one that would move the most money. Customer-facing applications photograph well. Internal process work does not. The money is usually in the second category.
The signal: if the initiative was selected in a meeting where someone said "imagine showing this to the board," it may be the right project, but nobody has checked.
The fix
Score candidate initiatives on value and effort before selection, using actual cost data. Then deliberately fund one unglamorous high-value project alongside the visible one.
What the successful ones share
Every pilot we've seen convert into a durable result had four things in place before it started. A named owner with authority. A measured baseline. A written success threshold. And an honest assessment of whether the underlying data could support it.
None of that is technical. All of it is the work that happens before anyone opens a vendor demo — which is precisely why it gets skipped, and precisely why the failure rate looks the way it does.
Have a pilot that stalled?
That's one of the more common reasons people call us, and it's useful information rather than a mark against you. The Modern Readiness Assessment usually starts by identifying which of these five patterns you hit.
Book a free call