Asynchronous interview completion rates (and how to raise them)

The completion rate of an asynchronous interview is the share of invited candidates who record and submit their answers. There is no clean published benchmark. The few real numbers recruiters share online tend to land between 50% and 70%. The rate is not fixed. It moves with four things you control: deadline, retakes, invite copy, and number of questions.

What counts as a good completion rate

There is no industry benchmark you can quote with a straight face. Nobody has published a clean, large dataset on one-way interview completion, and the vendors who could rarely do.

What exists is scattered and worth more than a made-up average. On Reddit, a recruiter at a Fortune 100 company described their setup plainly: “we have a 50% take rate.” That is a company with a strong employer brand and a lot of applicants, and half the people invited still did not record.

A different recruiter, hiring for roles that pulled 100-plus applications in a week, took a sharper approach. They sent the interview link to only about 40% of applicants, the ones who cleared a first filter, and reported that “around 60 to 70%” of those invited completed the video. Their read was that the step “helps weed out the half-serious candidates.”

So two real data points, from two very different funnels:

• Invite everyone, get roughly half.
• Invite a pre-filtered slice, get roughly two-thirds of that slice.

Use those as rough goalposts. If you are inviting your whole pool and clearing 60%, you are doing well. If you are well under 50%, the design is probably costing you good candidates, not just filtering weak ones.

The two ways to read a low number

Before you try to raise the rate, decide what a drop-out means to you, because the answer changes what you should do.

One view: every non-completion is a candidate you lost to friction. A great hire saw the invite, felt put off, and walked. Under this view, a low rate is a leak to plug.

The other view: the people who do not finish are mostly the people who were not that interested, and the format did you a favor. This is the “friction as a filter” argument, and the second recruiter above clearly held it.

Both are partly true, and the mix depends on the role. For a hard-to-fill senior position, treat every drop-out as a loss and make the interview as frictionless as you can. For a high-volume role with hundreds of applicants and thin resume signal, a moderate filter is doing real work, and chasing 100% completion would just bury your reviewers. Read friction as a filter for the full version of this trade-off, and is a one-way interview a red flag for how candidates experience the same step from the other side.

The rest of this page assumes you want the rate higher than it is now. Here are the levers, roughly in order of impact.

Lever 1: how many questions you ask

This is the biggest one and the easiest to get wrong. Completion falls as the interview gets longer, and the fall is steep.

Three to five questions is the range most teams land on. It is enough to judge communication and role-specific thinking without turning the interview into an evening’s work. Past five or six questions, the interview reads as unpaid homework, and the people most able to walk, your strongest candidates with other options, are the first to do it.

If a question does not map to something you are actually screening for, cut it. Every question you add costs you completions, so each one has to earn its place. See how long should a one-way video interview be for the time math behind this.

Lever 2: retakes

A one-way interview with no re-record is a high-stakes test. One stumble, one dog barking, one lost train of thought, and the candidate cannot fix it. Some people respond to that pressure by abandoning the interview rather than submitting something they think is bad.

Allowing at least one retake lowers the stakes of every recording. It tells the candidate the tool is not trying to catch them on a bad take, which both lifts completion and tends to produce answers people are willing to stand behind. It does not make answers less honest. Most candidates use a retake to recover from a genuine fumble, not to script a fake.

What to avoid is the opposite extreme of unlimited retakes with no time pressure, which lets a few candidates over-polish for an hour. One to three retakes, with a recording time limit per question, is the balance most teams want. How many retakes on a one-way interview covers what candidates expect here, and the time limit guide covers the recording side.

Lever 3: the invitation

The invite is the first thing the candidate sees, and a bad one loses people before they ever open the tool.

A few things move completion at this step:

• Say why. A line explaining why you use this format and that a person will review the answers makes the step feel less like being processed by a machine.
• Set expectations. State the number of questions and the rough time it takes. “Five questions, about fifteen minutes” lets someone decide to do it now instead of putting it off and forgetting.
• Name the deadline clearly, and make it generous (see the next lever).
• Keep it short. A wall of text reads as friction. So does a link buried three paragraphs down.

You can lift this straight from the invitation email template, which is built to do exactly this.

Lever 4: the deadline

A common failure mode here is simple procrastination. Candidates postpone, then run out of time and never finish. The interview was not hard. It just sat on a to-do list until the window closed.

The fix is counterintuitive. A deadline that is too tight, “complete within 24 hours,” reads as disrespectful and pushes busy people to give up rather than scramble. A deadline that is too loose removes the nudge to act at all. A window of several days, with a single reminder partway through, tends to work best. It is enough time for someone with a job and a life to find a quiet half hour, and the reminder catches the ones who meant to and forgot.

A quick diagnostic

If your completion rate is lower than you want, walk these in order:

1. Count your questions. More than five or six is the most likely culprit. Cut to the ones that map to a real screening goal.
2. Check retakes. If you allow none, turn on at least one. That alone often moves the number.
3. Reread your invite as a candidate. Is it short, clear about time, and does it explain why? If not, rewrite it.
4. Look at your deadline. Tighter than two days is probably costing you. Set several days and add one reminder.
5. Only then, decide if the rest is a filter. Once the design is clean, the candidates still not finishing are closer to genuine non-interest, and that is information, not a leak.

The goal is not the highest possible number. It is a rate where the people dropping out are the ones you would have screened out anyway, and the people you want are getting through without friction. Fix the four levers and the rate that remains tells you something real.

When you have the completion side handled, how to run an asynchronous interview covers the rest of the design: writing questions that force a real answer, scoring against a rubric, and keeping a human in the decision.