Repeatability
High
The task is structurally identical each time: ingest tabular data, segment into cohorts, compute retention and time-to-placement metrics, and surface patterns. This is a repeatable analytical pipeline with no unique judgment required per run.
Ambiguity Tolerance
Medium
The five input variables and two output metrics are clearly defined, but 'cohort combinations' and 'refine sourcing strategy' leave some interpretation open — the agent needs to decide how to bin company size, education level, and job title level into meaningful groups. A brief spec from the user resolves this quickly.
Data & Tool Availability
High
The user explicitly states they have all the required data fields for all 85 placements. As long as the dataset is provided in a structured format (CSV, spreadsheet, or database), the agent has everything it needs to execute without external API calls or permissions.
Error Cost
Low
The output is an internal analytical report used to inform strategy, not to execute irreversible actions. Errors in grouping or calculation are visible and correctable before any sourcing decisions are made, and the dataset is small enough for a human to spot-check.
Human Judgment Required
Low
Segmenting, aggregating, and ranking cohorts by quantitative metrics is mechanical work. The human judgment comes after — interpreting which patterns are actionable given market context — but the analysis itself does not require intuition or relationship knowledge.