The very first call a candidate has with your company is usually a 20-minute conversation with a recruiter. They check the basics: start date, location, salary range, why the candidate applied. The call is superficial by design, and that's fine. It's a triage step, not a signal round. Teams that worry about the initial screening call are usually worrying about the wrong meeting.
The round that actually matters, and that is easy to get wrong, is the one after it. The first technical screen. Thirty to forty-five minutes with the hiring manager or a senior engineer, often the first time the candidate talks to anyone who can actually evaluate their craft. On paper it's a filter. In practice, it's trying to be three filters at once, and that's why many teams find theirs inconsistent.
Three jobs, one meeting
A first technical screen is usually trying to do all of the following:
- Decide whether this candidate can probably do the job at all. Technical fit.
- Decide whether they'd plausibly work out on this team specifically. Team fit.
- Decide whether they actually want the role, or are just interviewing widely. Interest level.
Each of these is a legitimate thing to filter on. None of them is easy to do in 30 minutes. All three being smashed into the same meeting is why first technical screens quietly produce unreliable decisions.
The technical fit question wants a specific, time-boxed exercise. A question the candidate couldn't bluff through. The team fit question wants an actual conversation about values, working style, and context. The interest level question wants the candidate to do most of the talking while the interviewer listens for where they lean forward. Three very different shapes of meeting.
What actually happens
What you get, in most processes, is a compromise. The interviewer opens with small talk, asks two or three technical questions that don't really test anything specific, spends a chunk of time describing the company, and then rushes the candidate's questions at the end. Decisions come out of a gut feel about "whether it's a yes."
Gut feel isn't nothing. Experienced hiring managers are often right. But the format is producing reads that are hard to calibrate between interviewers, hard to defend in a debrief, and hard to improve over time. When a first technical screen goes wrong, it's usually impossible to tell which of the three jobs it failed at.
Pick one, drop the other two
The fix isn't adding more rounds. Four touchpoints before an offer is already a lot of calendar for the candidate, and five is usually where good people start politely dropping out. The fix is picking the one job this round can actually do well, and being honest that the other two will get read later in rounds you were going to run anyway.
Pick technical fit. It's the job that most benefits from a dedicated, structured filter, because it's the one the rest of the process can't easily rescue if you skip it here. The cheapest version is an async exercise: a short prompt the hiring manager reviews before any live call, so that by the time you talk to the candidate, you already know they can probably do the job and the conversation can be about something else.
The default shape of an async exercise is a take-home, and take-homes in their usual form aren't quite right for this job. They drift long, they measure output more than reasoning, and they assume the candidate isn't using AI when almost every candidate is. A tighter version is a short structured prompt the candidate works through in their own words, with the AI collaborator they'd use on the job in the room, and a transcript the interviewer can read before the follow-up. That's the shape CriticCode is built around.
The other two jobs get handled without adding rounds:
Team fit shows up in the take-home follow-up interview, or in whatever the next substantive conversation is. That's where you watch how the candidate reasons, how they disagree, how they respond to being pushed. A 30-minute "cultural screen" on top of that is usually redundant.
Interest level shows up across every touchpoint. Whether they ask about the team. Whether they follow up between stages. Whether they push back when you describe the role. You don't need a dedicated round for this. You need interviewers who notice and a hiring manager who reads the pattern.
That's the whole change. The process doesn't grow. You've just stopped the first technical round from pretending to be three rounds in one.
What about candidate time?
The argument against replacing the first technical screen with an async exercise is usually "we'll lose candidates if we ask for 20 minutes of work before a real conversation." Sometimes true, especially for senior candidates who have options. A few responses that help:
- Be explicit that the exercise replaces a live screen rather than adding to it. Strong candidates tend to prefer this. Nobody who is currently employed wants another 30-minute call on the calendar if they can avoid it.
- Scope the exercise to a single sitting. How long it ends up taking depends on the prompt you write, but if a candidate needs a weekend to finish it, the prompt is doing too much. You don't need an hour of their work to learn something useful from it.
- Respond to every submission within two business days. Candidates accept async screens when the team is obviously taking them seriously. They drop out when the submission goes into a void.
One practical test
If you can't say which of the three jobs (technical fit, team fit, interest) your first technical screen is doing, it's probably doing all three badly. Pick technical fit. Do it async. Trust the rest of the process to pick up the rest. You don't need more rounds; you need fewer rounds pretending.