Team KPI Tracker
AI-generatedSummary
Aggregates WaliChat and optionally Bullhorn KPIs for a selected period, grouping results by configured source labels. If an OpenAI API key is provided, classifies inbound replies to count positive replies.
Inputs
- Period — Select Today (the last 24 hours), Current Month, Last Month, or Custom. Custom uses Start Date and End Date.
- Start Date and End Date — Set the bounds for a Custom period; the operation reads these values only when Period is Custom.
- Enable Bullhorn — When enabled, fetches Bullhorn job submissions and related job pipeline counts; requires Bullhorn credentials. Defaults to true.
- OpenAI API Key — When provided, sends text from inbound messages in replied WaliChat chats to OpenAI for positive/negative classification. Leave empty to skip positive-reply classification.
- OpenAI Model — Select the model used for reply classification. Defaults to gpt-4.1-nano.
- WaliChat Settings: Device IDs — Comma-separated device IDs to query. If empty, the operation detects and queries all devices.
- WaliChat Settings: Source Labels — Comma-separated WaliChat labels and Bullhorn candidate source names used for grouping. Defaults to Stepstone, Indeed, Meta Pool, Meta Kundenspezifisch, and Meta Ad.
- WaliChat Settings: Label Groups — Map multiple labels to one output group, one group per line in the form GroupName=Label1,Label2. Defaults to grouping the Meta labels under Meta.
Output shape
A single item containing period, walichat, and bullhorn objects.
The period object includes the end date, month name and number, year, and ISO start and end timestamps. walichat contains total and per-group firstMessages, replied, and positiveReplies counts. bullhorn contains total and per-group submissionsWeitergeleitet, jobsWeitergeleitet3Plus, and jobsWeitergeleitet5Plus counts. Groups come from Label Groups and any ungrouped Source Labels, with Other included. When Bullhorn is disabled, its counts are zero. Positive reply counts remain zero if no OpenAI key is supplied; classification or message-fetch failures count as non-positive.