Content language · English
Stale & Cold Pipeline Rules: Stop Running a Museum
A job Kanban without stale and cold timers is not a pipeline. It is a museum of hope: rows age in silence while you congratulate yourself for “keeping options open.” Options that never get a next action are inventory, not leverage.
This note is about ops language for rows after they pass (or fail) an Application Quality Gate. Gate protects the entry to Applied. Stale/cold rules protect the life cycle of every stage after that. For the Gate itself, see Application Quality Gate. For the Sunday decision ritual that consumes these flags, see Weekly Command Review.
Locked stages (same vocabulary, every SKU)
Sourced → Applied → Screen → Loop → Offer → Closed
Onsite and video are formats under Loop, not extra stages. Autorejects are not Screens. “Recruiter viewed my profile” is not Applied. Stale/cold rules only work if stage words mean one thing.
Define the two words in writing
Stale = the row’s next_action date is past, and you have not bumped it, acted, or Closed the row. Stale is about your discipline.
Cold = silence from the company (or contact) past a stage-specific timer after your last logged touch. Cold is about their signal—or the absence of one.
Most boards conflate the two. Then people feel guilty closing a cold Applied because they “might still hear back,” while fifty stale Sourced rows rot with no Gate work scheduled. Separate the flags.
Default timers (tune to your market; write yours)
| Stage | Stale (your next_action) | Cold (their silence) |
|---|---|---|
| Sourced | 3–5 days without Gate work or Cut | N/A (not submitted) |
| Applied | 7 days without follow-up plan bump | 10–14 days no Screen / recruiter reply |
| Screen | 3 days without debrief + next ask | 7–10 days no Loop schedule signal |
| Loop | 2 days without thank-you / debrief log | 7–14 days no decision or next panel |
| Offer | Same day: floors + compare started | Deadline is the timer—honor it |
These are operator defaults, not industry law. Fast markets run colder clocks. Slow enterprise loops run warmer. The point is a written number you can disagree with next Sunday—not vibes.
What to do when a row trips a flag
1. Bump next_action (rare, intentional)
Only if you have a concrete reason: waiting on a named contact reply by a date, or a known holiday blackout. Log the reason in one line. Infinite bumps are how museums grow.
2. Act once, then re-arm the timer
One polite follow-up. One warm-path nudge. One Gate rewrite if still Sourced. Then set a new next_action. Do not “check LinkedIn for vibes” as an action.
3. Escalate warm path or Close
If cold after Applied and you have a contact_id, ask once with Gate-clean materials. If no warm path and silence past cold, Close with reason: cold-silence, comp-mismatch, wrong-level, withdrew. Closed is a stage, not a moral failure.
Weekly Review consumes the flags
Sunday is not for re-reading every JD. Pull:
- Count of stale rows by stage
- Count of cold Applied / Screen / Loop
- Top Close reasons this week
- One channel to scale, one to change, one to stop (see funnel metrics)
If stale Sourced keeps growing, you do not have a sourcing problem—you have a Gate throughput problem. Fix Gate capacity before you add more Sourced. If cold Applied dominates and Screen÷Applied by channel is already trash, stop feeding that channel hours.
Anti-patterns
- Eternal Sourced. Saving roles forever “for later.” Later needs a Gate slot or a Cut.
- Ghost Applied. Marked Applied, no confirmation #, no next_action, no follow-up bank entry.
- Shame Close avoidance. Leaving cold Loops open so the board still “looks alive.” Alive is wrong; honest Closed frees attention.
- Timer theater. Changing cold from 14 to 45 days every time silence hurts. Write the number; review it weekly; don’t move it mid-week to feel better.
One worked example (fiction, not a promise)
Imagine twelve Applied rows. Four are cold past fourteen days with no Screen. Two are stale because you never wrote a follow-up date. Sunday Review: Close the four cold with cold-silence, bump one stale with a named warm ask by Wednesday, Close the other stale as priority-cut. Your Applied count drops. Your Screen÷Applied usually rises because junk left the denominator. That is the point—not a prettier board for screenshots.
How this fits the Command Loop
Command Loop = Gate before Applied + channel rates + weekly scale/change/stop. Stale/cold rules are the plumbing that keeps stages honest between Sundays. Without them, Weekly Review debates feelings about a board that never aged out. With them, Review debates decisions about a board that already flagged what needs Close, bump, or escalate.
Honest limit: cleaning a board does not create Screens. It creates a true denominator and frees hours for Gate-quality Applies and evidence work. No hire / interview guarantee attaches to better hygiene.
Minimal field set (CSV-friendly)
Whether you use Sheets or Markdown, each live row needs at least: stage, stage_entered_date, next_action, last_touch_date, channel, quality_gate_pass, close_reason (when Closed). Optional but useful: stale_flag, cold_flag as formulas against your timers.
Product: Core is a $19 one-time digital download on jobhuntkit.top. Steal the stage vocab and timers today. SAMPLE rows in the kits are fictional samples—not your results.