Run your job search like a project

Six weeks into your search, a recruiter calls about a role you can't quite place. She is friendly, ready to schedule a screen, and using the company's internal name for the position. While you make agreeable noises, you are scrolling your sent mail with the other hand, trying to work out which of three similar postings this is, what the description actually emphasized, and which version of your résumé she has open in front of her. You applied. That much is confirmed. Everything else is gone.

Nothing failed there except a belief you held six weeks earlier: that you would remember.

A real search is a multi-month project with dozens of open threads. Applications in six different states, two recruiters mid-conversation, a take-home due Thursday, a follow-up you owe from last week, and a contact who said "ping me in a month" twenty-nine days ago. Memory is not a project tool. Nobody runs a months-long project with parallel workstreams out of their head, and the fact that this particular project is unpaid and emotionally loaded makes recall worse.

So keep a ledger. Per application, the minimum that earns its keep: current status; the dates things happened (applied, heard back, each conversation); which version of which document went where; the people you've talked to and who, if anyone, introduced you; and the next action with a date on it. Per week, one short note on what changed: what went out and what came back, plus anything you decided to do differently. That weekly note is the habit people skip, and it is the one that turns a pile of applications into something you can steer.

Our model evaluations at jobhunt.coach need the same discipline. Raw outputs live in artifact directories outside version control, where we retain them for replay and inspection. The decision documents record which model we chose and why. The standing instruction is to update the role's table when the decision is made. Months later, a pile of outputs may still exist, but it will not explain the choice as clearly as a note written at the time. Your search has the same shape: keep the materials, and also record what you decided to do with them.

A weekly review only helps if the records answer a concrete question: what did I actually do, and what came of it? Keep the source of each opportunity beside the application, and date the changes you make to your documents or targeting. Then you can look back without relying on a vague impression that one week felt busier than another. The record will not prove why an employer responded, but it can stop you from changing everything at once and forgetting what you tried.

I've written elsewhere about keeping records while you debug a quiet funnel; this is the standing version of that advice, a ledger you keep because the project runs for months either way, not a log you start once something breaks.

The discipline pays off hardest at interview time. You will be asked, with real specificity, about a job description you read six weeks ago, by a person who reread it yesterday. The candidate who can open a record and see the posting they applied against, the résumé that went out, and their own note from the first call walks in prepared. The one reconstructing from memory ends up talking about the job in generalities, and interviewers notice.

jobhunt.coach keeps this ledger as a side effect of doing the work: each application carries its status, its timeline of events and notes, and the documents that were generated and sent, so when the recruiter calls, the answer is a click away rather than a dig through sent mail. But I want to be plain about this: a spreadsheet works too. Plenty of well-run searches live in one, and a spreadsheet you actually update beats any tool you don't. Is maintaining it tedious? A little. Ten minutes at the end of the week covers it, and I'd argue it is the cheapest ten minutes in the whole search.

Before today ends, write down every thread you currently have open, with its state and its next action. Start the ledger before the phone rings.