Suitability score

Upwork's advice before spending Connects is to read the post closely, check the client's history and feedback, and look at the activity on the job, since many proposals or interviews "may reduce your chances". A suitability score is that triage, done on every job a campaign catches.

How it is computed. The model reads the job, the client facts and the agency's knowledge profile, and rates three dimensions from 1 to 5. Upsight then computes the weighted score itself rather than asking the model for the sum, because models asked to do the arithmetic drift from their own formula. The default weights are 0.35 for business fit, 0.40 for budget maturity and 0.25 for strategic value, and a budget governor caps the score at 2.5 when budget maturity is 1 and at 3.0 when it is 2. Weights and caps live on the scoring rubric, which the agency can edit, and a set of weights that does not add up to 1 is rejected. The campaign's floor then decides: a job below it is skipped with its score and reason recorded, and anything on the agency's list of projects to avoid is unsuitable whatever it scores.

What it does not predict. In our own sends, reply rate was flat across score bands, and the client's hiring history explained replies better than the score did. The score keeps clearly wrong jobs out; it is not a forecast of a reply. That is why hard client facts such as total spend and average rate paid are set as campaign filters rather than left to the model's reading.

How Upsight uses it. Every evaluated job keeps its three ratings, its weighted score and the model's reasoning, and the rubric a campaign uses is a named entry in the knowledge base, so a change can be compared against reply rate before it is kept.