Content language · English
Funnel Metrics for Job Search: Four Numbers That Matter
Most seekers track the wrong thing: how many applications they “sent.” Volume feels like progress. It is usually spray. Funnel metrics exist to answer one operator question—what should next week’s hours do?—scale a lane, change the channel mix, or stop. If a number cannot force that decision, drop it from the Sunday board.
This is the short primer. For the longer channel scoreboard and sample math, see Job Search Funnel Metrics That Actually Change Next Week’s Hours. The weekly ritual that consumes these numbers is in Weekly Job Search Review.
Lock the stage vocab first
Rates are garbage if stages are creative. Use one locked chain:
Sourced → Applied → Screen → Loop → Offer → Closed
- Applied means the packet left your hands and an Application Quality Gate passed—see Application Quality Gate. Easy Apply without Gate is not Applied; it is Sourced cosplay.
- Screen is a real conversation on the path (recruiter or HM). Autoreject emails are Closed, not Screens.
- Loop is the deeper panel set. Onsite or video is a format under Loop, not a new stage.
- Stale / Cold are timer states with next-action rules—not vibes. See Stale & Cold Pipeline Rules.
The four numbers
1. Screen / Applied (by channel)
Compute weekly, then split by channel tag: job_board, company_site, referral, recruiter_inbound, community, outbound_hm. A blended rate hides a dead board lane eating sixty percent of hours. Small n lies—treat n<10 as directional, not gospel. Trend across two Sundays beats one dramatic week.
2. Loop / Screen
If Screens happen and Loops do not, the leak is story, role fit, or Screen hygiene—not “ATS conspiracy.” Fix debriefs and evidence before buying another tracker. If Loop/Screen is healthy and Offers are rare, the problem moved downstream; stop optimizing the top of funnel for comfort.
3. Signal per hour
Signal events = Screens + Loops + meaningful replies that change a next action. Divide by hunt hours that week. Application count can rise while signal/hour falls—that is spray with a spreadsheet. Prefer the lane that buys signal cheaper, even if it “feels slower.”
4. Gate hygiene
Percent of new Applied with quality_gate_pass = Y. If this is low, every other rate is poisoned. Demote ungated rows back to Sourced; Gate or Close. Cap gated Applied to quality capacity (often a handful per week), not inbox anxiety. Spray-without-Gate is a failure mode with a name—fix it before you “optimize outreach copy.”
Sunday decision rule. Write one of: scale this channel’s hours, change mix or Evidence, or stop the lane for 14 days. Mood is not a decision. The worksheet shape lives in Weekly Command Review.
What not to optimize
- Raw Applied count as a KPI
- “Messages sent” without reply or stage change
- Color themes, emoji statuses, or Notion chrome
- National unemployment headlines as your personal forecast
Those metrics reward activity theater. Funnel metrics reward reallocation.
Minimum board to compute them
You need stage, channel, Gate flag, next_action_date, and a weekly hours total. CSV is enough. Fancy dashboards are optional. If you cannot recompute Screen/Applied from rows in five minutes, the board is decoration.
Sample size honesty: one Screen from three Applied is not a 33% “strategy win.” Keep hunting sample while you cut obvious dead lanes. When a channel has adequate sample and still yields nothing with clean Gate and deep Evidence—stop that channel. That is the system working.
Worked micro-example (fiction numbers)
Week of Nov 3 (Asia/Shanghai). Hunt hours: 12. Applied gated: 8 (job_board 5, referral 2, company_site 1). Ungated Easy Apply you almost counted: 11—demoted to Sourced before Sunday. Screens: 2 (both referral). Loops: 0. Meaningful replies that changed next_action: 1. Signal events = 2 + 0 + 1 = 3. Signal/hour = 3/12 = 0.25.
Screen/Applied blended = 2/8 = 25%, but by channel: referral 2/2, job_board 0/5, site 0/1. The decision is not “apply more.” It is scale referral hours, change job_board Evidence or targeting, or stop that board lane for 14 days if Gate was clean and sample is already loud. Writing that sentence in the review beats redrawing the Kanban colors.
How this ties to the Command Loop
Gate protects the Applied denominator. Rates diagnose the week. Weekly review forces scale / change / stop. Together they are the Command Loop—not a vision board. Upsell packs (metrics dashboards, interview sprint) deepen a stage; they do not replace Core vocabulary.
Keep the Sunday sheet ugly and honest. Four numbers beat twenty vanity charts.
Honest limits
Metrics improve diagnosis and hour allocation. They do not create interviews on demand, thaw a hiring freeze, or substitute for skill gaps. No kit can promise a hire. Job Hunt Command Kit Core is a $19 one-time digital download—this article is operator teaching, not a checkout pitch.