Running a Remote-First Hiring Operation

The loop redesigns that distinguish remote-first from remote-tolerant

Most companies that “hire remote” are running in-person loops on Zoom. That's why the remote candidates churn at twice the rate of in-office ones. The four loop changes that close that gap:

• Async-default first round. Replace the 30- minute recruiter call with a 10-minute Loom intro and a written candidate-prompt. Saves an hour on both sides and gives the candidate a re-watchable artifact.
• Work sample over whiteboard. 60-90 minutes, on the candidate's own machine, on their schedule. Reviewed asynchronously by two engineers. Paid.
• Onsite as a half-day, not a full day. Three sessions, well-spaced, with a 30-minute buffer between each. A full day of Zoom is a usability failure.
• Optional in-person culture day after the offer. If they want to fly out and meet the team, you pay; if they don't, you don't make it a condition. Cuts “offer accepted but later declined” in half.

Work samples that don't feel like unpaid labor

• Scope ≤ 90 minutes. If it can't be done in 90 minutes, you're auditioning, not assessing.
• Synthetic data, never real production prompts. A take-home that builds a feature you'll ship is theft.
• Pay $150-200 per assessment. Yes, even for junior candidates. Yes, even when they don't get the job. The economics work and the candidate sentiment is a separate marketing channel.
• Same prompt across the cohort. Calibration is impossible if each candidate gets a different problem.

Calibration changes for distributed interviewers

In-person calibration sessions are post-mortems. Distributed teams need both: pre-mortems before a loop (alignment on the rubric for this role, expectations on rigor) and post-mortems after (where did our scores diverge, why). Both work over Zoom. Skipping the pre-mortem is the single most common failure mode.

Geo-compensation policies that don't backfire

Three policies most companies pick from. There's no right answer, but there are wrong implementations:

1. One global band. Same offer to a Senior Engineer in Lagos, Lisbon, and Los Angeles. Expensive, simple, retention-positive.
2. Tiered by cost-of-labor zone. Three or four zones, transparent multipliers, published. Less expensive, more complex, common compromise.
3. Anchored to a benchmark city. Offer pinned to, say, Seattle market rate. Acceptable in good times, brittle when the benchmark moves.

The wrong implementation, in any policy: secrecy. If you can't publish your geo policy, your geo policy is wrong.