Reading the Report
Job search analytics from your own tracker: response rate and interviews by job source, form, location, title, fit score and resume, plus what to change.
/seekter-report is your job search analytics: how many applications got a reply, which job sources and resumes bring interviews, and what to change. It's read-only: it never applies, moves a record or edits your settings.
Before You Run It
Run /seekter-log first. The report reads your tracker, not your email, so replies that haven't been logged don't show up in it.
/seekter-report
Weekly is a good rhythm. It's saved as runs/report-<date>.md, and you get the short version in the chat.
What It Counts
For the last week and for all time:
- Applications sent, skips, and hand-offs still waiting on you.
- Replies: interviews, offers and rejections, as a share of what was sent: your response rate. Rejections count; a reply is a reply. The breakdowns below are stricter and only count applications that have been decided.
- Interviews: anything that reached
interviewingoroffer, even if it was later rejected. - How long companies take to answer.
- The most common skip reasons.
The Breakdowns
Each number is broken down by what you can change:
| By | What it tells you |
|---|---|
| Source and apply path | Which sources and form types bring replies |
| Form system | The application system each went through |
| Location track | Remote anywhere, ambiguous remote, or relocation. Many sends and no replies is a targeting signal. |
| Title and seniority | Which role families answer |
| Fit score | Whether your scores predict replies, which is the reason the score exists |
| Resume type | Which resume was sent, and how it did |
| Month | Whether the funnel is moving |
You can run any of them yourself. By fit score:
node scripts/seekter.mjs stats --by fit
By source, since a date:
node scripts/seekter.mjs stats --by source --since 2026-10-01
Reading the Numbers Fairly
- In-flight applications don't count against a rate. One still waiting has its own column, so a week of fresh sends doesn't drag a group down.
- Small groups show counts, not rates. Fewer than five decided applications prints
n<5instead of a percentage. - With only a handful of interviews, groups are compared on replies, and the number of interviews is stated rather than turned into a rate.
What to Change
The report ends with three to five plain observations, each with the number behind it, and at most two suggested changes, each tied to a file: a query in profile/settings.json, a rule in profile/profile.md. Nothing is changed until you say yes.