The original hiring loop looked reasonable on paper: phone screen, take-home, two technical panels, lunch, and a founder call. Six touches, four weeks, everyone calibrated. On paper. In practice, it was slow enough that the best candidates had signed somewhere else by the time we got to the offer stage.
So we opened the whole thing up. What follows is the teardown — what we measured, what we cut, and the thing we almost got wrong twice.
The numbers that mattered
We used to measure time-to-hire from the first phone screen to signed offer. The number that actually drove behavior was a different one: time between final panel and offer sent. That gap averaged 6.2 days and was the single biggest predictor of whether a candidate would accept.
A strong candidate who has waited six days for an offer has already talked to three recruiters, reopened a rejected LinkedIn DM, and mentally added "slow to decide" to your company's profile.
Internal retro, week 3
The fix wasn't about interviewing faster. It was about getting the decision committee in the same room on the same day.
The scorecard
We threw out the Slack-thread debrief. In its place: a four-field scorecard, filled in within 30 minutes of the interview ending.
- Signal observed. A specific thing the candidate did or said — not a trait rating.
- What it suggests. A hypothesis about how they'd operate on the team, explicitly stated.
- Where you'd want more evidence. What you'd ask in a second round, or a follow-up.
- Recommendation. Strong yes / yes / no / strong no. No "maybe".
The "no maybe" rule was the one that stuck. It pushed interviewers to make the case for or against, which meant the committee had material to discuss instead of averaging vibes.
A scorecard is only as fast as your slowest panelist. If one person habitually fills it in the next morning, your loop is as fast as "next morning". Calendar it as a 30-minute block after every interview, treat it like a meeting.
Scheduling rules
The second intervention was operational. Three simple rules, written on a card, stuck to the interview coordinator's monitor:
- Panel interviews happen in the first 48 hours after a candidate clears the phone screen — or the screen gets redone later.
- The committee meets within 24 hours of the last panel interview. If everyone can't make it, a quorum of three is enough.
- If we can't send an offer by Friday, we send a status email on Thursday. Silence is the enemy.
"Silence is the enemy" became the single most-repeated phrase in the team. It ended up saving three finalists in the first month alone.
— Week 4 retroWhat most teams skip
The thing we almost missed: the reference call is part of the hiring loop, not a formality at the end. For most roles we were running references after the offer was drafted. That meant two things — first, we couldn't use the reference to strengthen the offer, and second, we had a standing commitment to ignore anything we learned.
We moved references to between the last panel and the committee meeting. The committee then had actual data, not just interview impressions. Three times in six months that call surfaced something that changed a decision — twice toward a yes, once toward a no.
Takeaways
- Optimize for the gap between final panel and offer — not total hiring days.
- Force interviewers to write evidence, hypothesis, follow-up, and a binary recommendation.
- Calendar the debrief, not the interview. The interview fills itself; the debrief slides unless protected.
- Move references earlier. Use them as committee input, not ceremony.
- When you can't send an offer on time, send a status note. Silence loses finalists faster than any other single thing.
Want the scorecard template?
We'll send the Airtable template, a sample debrief agenda, and a copy of the 3-rule scheduling card.