Before an outside advisor starts, I would want one page with two separate paragraphs. The first says what decision the work is meant to inform and what the advisor will hand over, including the evidence it will rest on. The second says who makes the decision. Later, the page lets both sides tell whether a request sits inside the agreed work.
Take a hypothetical software company of about forty people with two promising customer segments: regional accounting firms and finance teams inside mid-sized manufacturers. The founder has to decide whether next quarter’s product and sales capacity goes mainly to one of them, and if so which. An advisor is brought in for four weeks.
The first paragraph might say that the founder will decide before the quarter ends whether to concentrate on one segment. The advisor will deliver a written comparison of the two segments built from twelve customer interviews, six per segment, drawn from a list the founder approves in week one, plus the company’s existing product usage data and the last twelve months of support tickets. The comparison will report retention and per-account support load, list the product gaps customers name, give the advisor’s reading of which segment the current product serves better, and state where the interviews and data disagree or run out.
The second paragraph is short. The founder decides the allocation. The advisor may recommend, and the comparison is finished once delivered as described, whether or not the founder follows it.
Suppose the comparison leans toward manufacturers with two caveats. Four of the six manufacturer interviewees were suggested by the sales lead, and support data for that segment covers only nine months because its first contracts closed late. The founder reads it and says the engagement was supposed to tell them which segment to pick. The advisor says it was supposed to compare two segments on an agreed evidence base and report the limits, and that it did.
Without a written scope, they could end up arguing about who remembers the kickoff call correctly. Here both people can read the same sentences: the deliverable was a comparison of product fit with stated limits, and the choice was reserved for the founder. The founder may still think the caveats leave the comparison too thin to act on. That is a fair position, and it can now be argued on the evidence, because what was promised is on the page.
Then the founder asks for eight more manufacturer interviews, chosen at random, and adds that logistics companies came up at a board meeting and should be compared too. Both requests may be sensible. The page shows each as new work: the extra interviews change the research base the comparison was built on, and logistics is a segment the decision paragraph never named. The advisor can quote them as a second piece of work, and the founder can weigh whether more evidence is worth having if it arrives after the allocation is due.
If the original work carried a fixed fee, in this example it covered the twelve interviews and the two-segment comparison as written, along with an honest account of what they could not settle. The founder’s view that the caveats are too large does not by itself make the work unfinished.
The page records what the founder asked for at the one moment when neither side had a result to defend. Whether twelve interviews were enough to allocate next quarter’s capacity, and whether choosing between these segments was the right question, remain the founder’s judgment to make.