Stopping Rules for a Job Search

Most of the hardest decisions in a job search are decisions to stop. Stop waiting on that application. Stop putting hours into a channel that hasn’t produced a conversation. Stop running the approach you’ve been running since April. Stop for the night.

They’re hard because you always make them at the worst possible moment: tired, at the end of a bad week, with a rejection in your inbox. Judgement made in that state is either far too harsh — this whole thing is pointless — or far too soft, which looks like waiting eleven weeks on a role that closed in May.

A stopping rule is the fix. You write the condition down when you’re calm, and then you obey it when you’re not. That’s the entire mechanism, and it’s the closest thing this site has to a trick.

Four levels, and they’re different rules

People collapse these into one anxious question — should I give up? — when they’re four separate decisions operating on completely different timescales.

Level 1: a single application. When do you stop expecting a reply and mark it Ghosted?

Level 2: a live process. When do you stop treating a stalled process as real?

Level 3: a channel or a segment. When do you stop putting hours into a job board, a recruiter, a company type, a role title?

Level 4: the approach. When do you stop running this version of the search and design a different one?

There’s also a daily one — when do you stop for the evening — which is really a cadence question and lives in a weekly cadence for a job search.

Level 1: the ghost cutoff

The easiest rule to write and the one that clears the most clutter.

Pick a number of days from application. Past it, with no human reply and your follow-ups sent, the row moves to Ghosted and you stop thinking about it. Thirty days is a defensible default in a lot of fields; if you’ve been tracking for a few weeks you can do better and use your own median time-to-first-reply, roughly doubled.

Two things make this worth doing. It keeps your response-rate denominator honest, and it converts an open loop into a closed one. Open loops are what make a search feel heavier than it is — nineteen rows you’re vaguely still hoping about is a much worse psychological load than four live processes and fifteen closed ones, even though it’s the same information.

Marking something Ghosted isn’t a prediction. Occasionally one of them comes back to life eight weeks later. Fine — reopen the row when it happens. The rule is about where your attention goes today.

Level 2: the stalled process

Harder, because a process that reached a human feels like it deserves indefinite patience.

A workable rule: two follow-ups past their own stated timeline, then it’s dormant. Not dead — dormant. Move it out of your live-processes view, stop holding calendar space for it, and stop counting it when you decide whether you have enough in flight. Follow-up timing and wording is its own topic; see the follow-up that is not annoying.

The reason this rule matters isn’t politeness. It’s that a phantom live process makes you under-apply. You feel covered, so you don’t source, and six weeks later the top of your pipeline is empty and the thing you were waiting on never existed.

Level 3: retiring a channel

This is where a tracker earns its keep, because the rule has to be numeric or you’ll never apply it.

Write it like this: if a channel has produced no conversation after N applications through it, it moves to low priority for a month.

You pick N based on how expensive the channel is. A board where an application takes five minutes deserves a large N. A channel that costs an hour per application — bespoke outreach, an involved portal, a long form — deserves a small one, because you’re spending real hours to find out.

Two cautions. First, small numbers lie: three applications through a channel with no reply tells you almost nothing, and if you retire channels at N=3 you’ll retire all of them. Second, don’t confuse a channel with a segment. “No replies from job boards” and “no replies from senior roles at large enterprises” are different findings with different responses, and your source column only answers the first one. If you want the second, you need a column for it.

Retiring is also reversible and should have an explicit re-entry condition: revisit in four weeks, or sooner if something material changes.

Level 4: changing the approach

The big one, and the one to make on a schedule rather than in response to a bad day.

Set a review point in advance — a date, or a volume threshold, whichever comes first. Something like: at the end of July, or after forty applications, I look at the whole thing and decide whether the shape is right.

Having the date written down does two useful things. Before it, you can stop relitigating: the approach is decided, the question is closed, execute. After it, you’re allowed a genuine rethink without it feeling like defeat, because you scheduled it.

At that review, the questions worth asking are structural rather than motivational:

• Where in the funnel does it stop? Applications with no replies, replies with no screens, and screens with no second rounds are three different problems.
• Which channels produced every conversation I’ve had? Usually a small number of them produced almost all.
• What am I applying to that I wouldn’t accept? Remove it; it’s costing hours and distorting your numbers.
• What’s one variable I could change for the next four weeks?

One variable. Changing five at once means the next review can’t tell you anything.

Write them down where you’ll see them

A stopping rule that lives in your head isn’t a stopping rule, it’s an intention. Put a short block at the top of your tracker:

STOPPING RULES — set 23 June, review 31 July

Ghost cutoff:      28 days after applying, follow-ups sent
Stalled process:   2 follow-ups past their stated date -> dormant
Channel review:    no conversation after 12 applications -> low priority 4 weeks
Approach review:   31 July, or at 40 applications, whichever first
Daily stop:        21:30, mid-sentence if necessary

Six lines. Re-read them in your weekly review block and only change them at a review, never in the moment you want to break one.

What stopping rules are not

They’re not a plan for quitting. Almost nobody’s real problem is that they searched too long; the real problems are waiting on dead applications, pouring hours into a channel that never worked, and running an approach for four months because changing it felt like admitting the first version failed.

Stopping rules address all three, and they do it by moving the decision out of the moment where you’re least equipped to make it. That’s also why they feel like relief rather than pressure. When something goes quiet, you don’t have to decide anything — you already did, weeks ago, when you had the capacity for it.