CAUSE OF DEATH: Unrealistic expectations sustained over approximately 18 months.

TIME OF DEATH: Somewhere between your third "final round" interview and the moment your preferred candidate accepted a role elsewhere.

NEXT OF KIN: Your Q3 delivery roadmap. Also grieving.

The Unicorn was, by all accounts, a remarkable creature.

Full-stack AI and cloud engineering. Deep Azure and AWS. Production ML experience. Strong stakeholder communication. Team lead potential. Available immediately. Open to $130k. Based in Sydney.

We tried to warn you. So here we are.

CONTRIBUTING FACTORS

1. The Job Description Written by a Committee

It started as a cloud engineer. Then the data team added "familiarity with our ML pipeline." The CTO added "some people leadership would be ideal." HR standardised it against a 2019 salary band. Someone from the business added "comfortable with ambiguity." Legal asked for "strong attention to detail." The office manager suggested "team player."

By the time it went live, you'd written a job ad for a person who doesn't exist. And if they do exist, they're not looking at Seek - they're being headhunted by a hyperscaler for twice the money, with four weeks annual leave and a Macbook Pro bought before the last two interest rate cycles.

2. The Six-Round Interview Process

Round one: recruiter screen. Round two: hiring manager. Round three: technical assessment. Four hours. Unpaid. Due Monday. Round four: panel with three people who weren't in the brief and one who hadn't read it. Round five: "culture add" conversation with someone from a completely different team who asked where you see yourself in five years. Round six: two weeks of silence, then a calendar invite marked "feedback."

Your best candidate withdrew after round three. You found out via a LinkedIn notification that they'd started somewhere else. They left a Glassdoor review. You don't want to read it.

3. The Salary Band Set In 2019

I'm going to give this one the respect it deserves and say nothing further.

4. The "We'll Know It When We See It" Brief

Nobody could define what success looked like at 90 days. So the search ran for four months. The goalposts moved twice - once because a new stakeholder joined the panel, once for reasons nobody could articulate. Three strong candidates were passed on. The reasons given would have also eliminated the person you eventually hired, but nobody wrote them down, so we'll never know.

The role was eventually backfilled six months later. Hi again.

CORONER'S FINDINGS

The Unicorn didn't die because the market is impossible.

It died because every single thing you did to reduce hiring risk - the extra interview round, the broadened scope, the delayed decision, the internal alignment meeting that needed its own alignment meeting - those aren't caution. They're attrition.

They filter out candidates with options. Which is, with exquisite precision, the exact candidates you wanted. The process designed to protect you from a bad hire is the reason you have no hire.

WHAT TO DO NEXT

Split the role. A focused cloud engineer and a focused ML engineer will outperform the mythical hybrid - and in eighteen months, you'll still have both of them.

Write the brief for the first 90 days. Not the next five years, not your theoretical future state architecture. The actual problem, the one keeping you up right now.

Three rounds maximum. If you can't make a decision in three conversations, you don't have an information problem, you have an alignment problem. Fix that first.

Pay the market rate. The 2019 band isn't a constraint. It's a choice. Own it or change it.

The right person exists., they're just not going to survive six rounds to meet you.

When you're ready to run a search that doesn't eat its own candidates - drop me a message.

Jason Pretorius - Recovering Unicorn Hunter.