For employers
How to run an asynchronous interview that actually predicts fit
A step-by-step playbook for employers: how many questions to ask, how to write them, how to score answers fairly, and how to keep the candidate experience good.
An asynchronous interview is only as good as its design. Ask vague questions and you get vague answers that tell you nothing. Ask too many and good candidates abandon it. Score on gut feel and you have rebuilt the bias you were trying to escape. Here is how to set one up so it earns its place in your process.
1. Decide what you are actually screening for
Before you write a single question, name the two or three things this stage needs to prove. For a support role that might be written clarity, empathy, and judgment under pressure. For a sales role it might be discovery instinct and how someone handles an objection.
You are not trying to make a final decision here. You are trying to decide who is worth a live conversation. Screen for the signals that separate “worth a call” from “not yet,” and leave everything else for later rounds.
2. Keep it to three to five questions
Completion rate falls fast as an interview gets longer. Three to five questions, at sixty to ninety seconds each, respects the candidate’s time and still gives you enough to compare. If a question does not map to one of the things you named in step one, cut it.
3. Write questions that force a real answer
The best async questions cannot be answered with a rehearsed script. Ask for a specific example, a decision and its reasoning, or a short response to a realistic scenario.
- Weak: “Are you a good communicator?”
- Strong: “Tell me about a time you had to explain something technical to someone who did not have the background. What did you do?”
Scenario prompts work especially well because they show thinking, not just history. See worked examples by role for prompts you can adapt.
4. Score against a written rubric, not vibes
Decide what a good answer looks like before you watch any of them. A simple rubric, one to four points on each of two or three traits, keeps reviewers honest and makes answers comparable across a large pool. It also gives you a defensible record of why one candidate advanced and another did not.
Have more than one person score the borderline cases when you can. Disagreement is useful information.
5. Protect the candidate experience
A recorded interview can feel cold if you let it. A few things make it feel respectful instead:
- Explain why you use this format and how long it will take, up front.
- Offer a practice question so the real one is not the first time they see the tool.
- Allow at least one re-record.
- Give a generous deadline, several days, not hours.
- Tell people when to expect to hear back, and then do.
6. Keep a human in the decision
Software can transcribe answers, organize them, and surface the parts that match your criteria. That saves real time. The decision about who advances should still belong to a person reading against the rubric. Tools that score and rank are useful for triage. They are not a reason to stop watching the answers that matter.
7. Review fast, while it counts
The whole point of moving the first screen to an asynchronous format is speed. Watch answers at 1.5x, skim the transcript, score, and move the shortlist into live interviews within a day or two. A fast, respectful screen is a competitive advantage. A recorded interview that sits unreviewed for two weeks is worse than a phone call.
When you are ready to pick a tool to run all this, the software comparison walks through the main options.