Dev.to WebDev πŸ›  Dev πŸ‘ 0 πŸ“– 4 min read

Building WUUUH: A Rejection Log Tells a Truer Story Than Memory Does. What Worked, What Broke

A job search after a layoff produces two records. One is written down: applications sent, replies received, interviews held, stages reached. The other is the record memory keeps, and memory is not a neutral archivist. Re

A job search after a layoff produces two records. One is written down: applications sent, replies received, interviews held, stages reached. The other is the record memory keeps, and memory is not a neutral archivist. Rejections arrive as events with dates and wording, so memory files them prominently. Progress arrives quietly, a reply here and a second round there, and memory files it as background. After a few months, the search a person remembers can look considerably worse than the search they actually ran.

Building a peer community for the weeks after a layoff made that gap impossible to ignore. People arrived convinced their search was failing, then described, in the same conversation, a steady rate of interviews. The problem was not the search. The problem was the record they were reading it from. So the part of the build this post is about is deliberately unglamorous: a rejection log.

What was built

WUUUH at https://wuuuh.com is a peer community for people rebuilding after job loss, and the log sits underneath the community as shared infrastructure. Each entry holds only four things: the employer, the role, the stage reached, and any reason actually given, recorded in the employer's words rather than in a paraphrase softened or sharpened by mood.

Four fields was a deliberate choice, and it was the second attempt. The first version had fields for everything and died the same death every overbuilt tracker dies: people filled it in faithfully for a week and then stopped, because each entry had become homework. The four-field version survives because it takes under a minute and asks nothing a person does not already know.

The log is shared in the sense that matters: members compare patterns, not scores. One person's cluster of rejections at application stage next to another person's cluster after final interviews reads very differently from either one alone. The first pattern usually points at targeting or presentation. The second usually points at shortlists and timing rather than at any fixable fault. Neither conclusion is available from a single rejection, and both are available from a log kept for a month.

What broke

The first break was the scoreboard effect. The early log led with totals: applications sent, replies received, interviews held. In week six of a search, a totals page is a cruel object. Counts of applications sent started to stand in for progress even when nothing was moving, and a low week read as a failed search. The fix was to lead with a different question: what is the next single action on the oldest open application? One action, named, finishable today. The totals are still there, one tap away, but they stopped being the headline.

The second break was the reason field. Left free-form, people wrote what the rejection felt like rather than what was actually said. A polite "we went with another candidate" became "not good enough" by the time it was written down, and the log started amplifying the very distortion it was built to correct. The fix was small and strict: the reason field holds the employer's words or stays blank. A blank is honest. An invented reason is not a record.

The third break was timing. A log updated at the end of the week was a log updated from memory, which defeated the purpose. Entries now go in when the reply arrives, while the wording is still on screen. It is a smaller discipline than it sounds, because the entry is four fields, and it is the discipline the whole feature depends on.

What the numbers look like

The honest measure here is not applications per week. It is whether the log changes a decision. The pattern that shows up most often is the protected decision: a search producing interviews at a steady rate, read from the log, does not get torn up and restarted in a low week. Instead one thing changes at a time, for a stated reason, and the log shows whether the change helped. That is the difference between adjusting a search and restarting it repeatedly.

The second measure is the closed thread. A short, specific follow-up referencing the posting and the date applied draws answers from recruiters who had gone quiet, including honest ones: the role was filled, the team was paused, the shortlist had closed. Each of those answers ends a thread that would otherwise have stayed open in someone's head for weeks. A log that records the answer is a log that lets the search move on.

What stayed

Keep the record at the moment the news arrives, record the employer's words or nothing at all, and lead with the next single action instead of the totals. A search read from a log is a search that can be steered, and a search that can be steered is one a person can keep going through a discouraging month.

The lesson that survived every revision is simple: memory keeps the worst of a search and loses the rest. Write down the four fields, and the search you actually ran is the one you get to judge.

πŸ“° Read the original article on Dev.to WebDev

Originally published by Dev.to WebDev. Aggregated on AIWithGhost for educational purposes β€” full credit and traffic to the original publisher.