The question: for the next appeal, who should we ask first? Ask — "Which donors are most likely to give to the Spring Appeal campaign?" — and Blossom answers with a ranked list of donors and the evidence behind each ranking: how much of their giving has answered campaigns before, how recently they last gave, how often they gave in the last twelve months, and their average gift. Donors who already gave to the campaign you name drop out of its list — they have already responded.
The ranking is labelled, on the answer itself, a model estimate: a score composed from recorded giving by fixed, published weights. It ranks donors for a person planning an appeal; it does not predict that anyone will give, and no accuracy is claimed for it. The Donations Manager (§5.14) carries the same reading in its panel. In the Analytics Lab (Chapter 2), the question Group donors by how they give groups your donors into segments by their giving, and donations can be forecast like any other series — experiments learn from the same governed gift records these reports read.