| name | feedback-analyst |
| description | Analyze application outcomes to improve future applications. Triggers automatically every 10 applications or when asked to analyze results. |
Feedback Analyst -- Learn From Outcomes
When To Run
- Automatic: Every 10 applications (check count with
get_applications)
- Manual: When asked to "analyze", "how am I doing", or "feedback"
Workflow
Step 1: Pull Application Data
Use get_feedback_analysis to get the analysis from the applicant database. This includes callback rates, score correlations, and timing data.
Step 2: Identify Patterns
Look for:
- Callback rate by resume version: Which set of highlights gets more responses?
- Cover letter score correlation: Do higher humanization scores lead to more callbacks?
- Research depth: Do applications with more company research bullets succeed more?
- Source performance: Which job boards produce better callback rates?
- Time patterns: How long between application and first response, by company size?
Step 3: Generate Recommendations
One actionable recommendation. Not five. One.
Examples:
- "Your callback rate jumped from 8% to 22% when you started including deployment metrics. Keep leading with infrastructure numbers."
- "Applications from SearXNG-sourced LinkedIn posts have a 0% callback rate -- those listings may be stale. Try filtering to jobs posted within 7 days."
- "Cover letters scoring 75+ on humanization get callbacks 3x more often than those scoring 60-74. Aim higher before submitting."
Step 4: Report
"Application Analysis ([X] applications, [Y] days):
Callback rate: [X]%
Best resume version: [version] ([X]% callback rate)
Best source: [source] ([X]% callback rate)
Recommendation: [one specific, actionable suggestion]"
Rules
- Need at least 10 applications for meaningful analysis
- Never make up statistics -- only report what the data shows
- One recommendation, not a laundry list
