
When Everyone Passes, the Review Failed
Forty-seven reviews. Forty-three 'meets,' four 'exceeds,' zero 'below.' The cycle was a calibration exercise that had stopped calibrating.

Forty-seven reviews. Forty-three 'meets,' four 'exceeds,' zero 'below.' The cycle was a calibration exercise that had stopped calibrating.

They named the category, ranked first for the term, and the term drew eighty searches a month. The category existed in the deck and nowhere else.

Six hundred thousand a quarter, four channels, clean CAC on the dashboard. The dashboard was wrong — it took a Series B investor and a holdout test to prove it.

Year one he took five customer calls a week. By year three he was down to two a month, and the roadmap had drifted without anyone deciding to let it.

Six months of engineering. Three demos to leadership. Logged into twice in production. The tool was a project nobody had asked for.

She didn't leave for a raise. She left because she'd spent a year and a half working around one person, and the math finally stopped working for her.

Pipeline was light, so marketing introduced a new ICP. The new ICP didn't exist. The deck had a name for it anyway.

By July the invoice was 120 days old. The customer was still polite. The receivable was already gone.

The signal shows up in spring. The pivot happens the next winter, after months of burn and the best people leaving. Most of the delay is identity, not data.

Thirty-one action items over six retros. Four done. The retro was a ritual that mistook complaint for improvement.

He ran them every quarter for two years, then stopped because he was busy. Three resignation letters later, he saw what the skip-level had been buying him.

The board said go global. The product still didn't work for half the US customers. London opened in March. By December, it was a tax.

Same price per share. New terms in the legal language. The founder found the down round in a footnote eight months later.

It came in at $40M when the last round said $30M, and he laughed it off. Eighteen months later, the company took a $12M down round.

A standup that never changes anyone's plan has stopped coordinating and started reporting to the manager. Kill the format; ask one question instead.

He shipped twice as much as anyone. He documented none of it. Two years in, the company couldn't onboard anyone without his calendar.

A $40,000 booth scanned 2,000 badges and closed one deal. The revenue at a conference shows up at the dinner, not the ten-by-ten carpet.

One logo was 42% of revenue. The slide said 'enterprise customers,' plural. The investor asked the number anyway, fifteen minutes in.

Their logo opened doors. Their partner showed up to one board meeting in two years. The check was real. The relationship was vapor.

When the wiki is empty, Slack becomes the wiki — and Slack is the worst filing system ever shipped. The five-line decision log is what actually survives.

Sixty days of check-ins and three signed documents produced an outcome the team called on day one. The plan wasn't improvement — it was a record for the file.

Signups up forty percent, conversion down sixty: the trial worked perfectly for people who would never pay. A trial is a selector, not a funnel widget.

Invoices went out on the first. Money landed somewhere between day forty and day ninety. Nobody could explain the gap because nobody was watching it.

The role sat open four months. The pipeline was empty. You hired the one candidate who said yes. They left in nine weeks.