Ask any SMB executive when they last killed a project — really killed it, publicly, with the reasoning documented — and watch how long it takes them to answer.
Most can’t remember. Projects here don’t get killed; they get quietly de-prioritized, moved to next quarter’s roadmap, and then next quarter’s, until the person championing them leaves and the whole thing evaporates without anyone acknowledging it happened.
That pattern is expensive under normal conditions. Under AI conditions, it becomes ruinous.
Here’s why. AI adoption produces a lot of promising-looking prototypes. Vendors run beautiful demos on your data. Enthusiastic employees build working proofs-of-concept in a weekend. The whole environment is optimized to generate initiatives — most of which will never earn their keep in production, but each of which will attract a small constituency of people who think it’s about to work.
If your organization can’t say “this didn’t work, we’re moving on,” AI will multiply your existing pattern of accumulating unfinished initiatives. You’ll end up with fifteen half-implemented AI experiments, none producing measurable value, all draining engineering time and creating security surface area you don’t want.
The muscle to stop matters more than the muscle to start. It’s also harder to build.
The best small businesses I work with have a simple discipline: every quarter, they take one initiative that isn’t working and kill it publicly, with the reasoning written down. Not because that specific one deserved to die — sometimes it did, sometimes it was borderline — but because the practice of killing things keeps the option alive when they need it. Muscles atrophy when they aren’t used.
One client I worked with turned this into a ritual: every quarterly meeting ended with the question “what are we going to stop doing?” Half the time the answer was nothing. But the question stayed on the agenda, and when a real underperformer needed to be shut down, no one froze.
The cost of killing a project that had promise is small. The cost of nurturing five zombie initiatives for three years is enormous. The math almost always favors the cut.
You can’t build this muscle by talking about it. You build it by using it, on something small, before you need it.
Diagnostic question 2 tests exactly this. Score honestly.