A pilot the customer cannot stop after is not a pilot, whatever it is called in the proposal.
Why it matters when the plan changes
Buying decisions about organisational work are made on very little evidence, because the output is hard to evaluate before it exists. A pilot converts that into a small real piece of work with a defined question, which tells the buyer more than any reference call and tells the supplier whether they can actually help. It functions as the investigation a reasonable party undertakes before committing. Where it works, it becomes the first step of a longer arrangement, with terms for what follows agreed only once the pilot has answered its question.
The tension is between bounding and usefulness. A pilot small enough to be low risk may be too small to answer a question anyone cares about, producing a clean result nobody acts on. The useful version is bounded in scope and consequential in subject, which is harder to scope and harder to sell. Work already paid for in the pilot should be credited rather than billed twice once the relationship moves into a fuller agreement.
In practice
A company pilots on one division with a specific question about an operating-model change, a defined deliverable, a fixed fee and a review point. The finding changes the sequencing of the change. Whether or not the relationship continues, the pilot paid for itself on its own terms, which is what a pilot is supposed to do.
Evidence
A framework agreement sets terms for subsequent work, which is what a successful pilot typically leads into.
Framework agreement, Wikipedia (2026)Due diligence is the investigation a reasonable party undertakes before committing, which is what a buyer is doing in a pilot.
Due diligence, Wikipedia (2026)
What it cannot tell you
A pilot tests whether a supplier and method work on one bounded question; it does not tell you whether the same approach holds at scale, across a different unit, or under a different sponsor. A clean pilot result is not a guarantee of an ongoing relationship, and a small question answers only itself.
Questions
An agreed question, defined deliverables, a fixed fee and a review point at which the customer can stop without penalty. This mirrors what Due diligence, as defined in the Wikipedia entry updated in 2026, calls the investigation a reasonable party undertakes before committing. Without a stop point, the pilot is already a commitment.
Small enough that stopping is a real option and large enough that the question matters. A pilot on something inconsequential produces a clean result nobody acts on, which tells the buyer nothing about whether the work would help on a decision they care about.
Where it covers the same deliverable, yes. A pilot that leads into a framework agreement, as described in the Wikipedia entry updated in 2026, sets terms for the work that follows, so charging twice for the same deliverable makes little sense and crediting the overlap is straightforward if scope was defined clearly.
Whether the question was answered, whether the answer changed anything, and what the customer would want next if anything. A review that only assesses satisfaction with the deliverable misses whether the work was useful, which is the thing being tested.
Choosing a safe subject. A pilot on a low-stakes question is easy to agree and produces a finding nobody needs to act on, so the buyer learns that the work is competent and not that it is useful. The subject has to matter for the test to mean anything.