Why Candidates Drop Out of Video Interviews, and How to Fix Completion Rates

Candidates abandon video screening interviews for four repeatable reasons: the invitation does not say what the interview involves or how long it takes, the process demands effort before the employer has shown any commitment, the technology fails on a phone, and the deadline is too short. Completion improves most from three fixes that cost nothing: state the time commitment and question count in the invitation, cut the interview to the length you promised, and let candidates start on a phone without installing anything.

Video screening is sold on throughput, and the throughput is real right up until you look at how many invitations turned into finished interviews. Teams that send 100 invitations and get 45 completions have not screened 100 candidates. They have screened whoever was patient enough, and patience is not one of the things they were hiring for.

Worse, drop-off is not random. The candidates most likely to abandon are the ones with the most options and the least free time, which is the opposite of the selection you wanted. Fixing completion is usually a bigger lever on shortlist quality than anything you do to the questions.

Why do candidates drop out of video interviews?

When you read what candidates say about this, the same four complaints come up.

They do not know what they are agreeing to. An invitation that says "complete your video interview" without saying how many questions, how long it takes, whether they can re-record, or who watches it is asking for a blank commitment. People defer blank commitments, and deferral is where most abandonment actually happens. Nobody decides to quit, they decide to do it later.

The effort feels one-sided. A candidate who has not spoken to a human is being asked to perform on camera for a company that has invested nothing in them yet. That reads as being processed, and it is the specific complaint behind most negative sentiment about one-way video. The imbalance is worse when the role is junior or the market is candidate-friendly.

The technology gets in the way. A large share of candidates start on a phone, often on mobile data. Anything requiring a desktop browser, a download, a plugin, or a stable connection for ten unbroken minutes will lose a meaningful group, and it will lose them disproportionately by income and by location.

The deadline is wrong. A 48 hour window sent on a Friday afternoon reaches people who work full time and have families. Shift workers and caregivers get filtered by scheduling rather than ability, which is both an unfair outcome and a legally uncomfortable one if it shows up in your selection rates.

What is a good video interview completion rate?

Treat published benchmarks with suspicion. Vendors quote wildly different numbers, rarely define whether the denominator is invitations sent, invitations opened or applicants, and have an obvious interest in the answer. A rate calculated on opened invitations will always flatter a rate calculated on applicants.

The useful benchmark is your own, measured consistently. Define completion as finished interviews divided by invitations sent, track it per role, and watch the trend rather than the absolute number. Then segment it, because the segments are where the real information is:

• By device, since a large gap between mobile and desktop is a product problem you can fix this week.
• By time from application to invitation, because interest decays fast and a five day lag drops completion sharply.
• By day and time the invitation was sent.
• By source, since referred candidates and cold applicants behave very differently.
• By demographic category where you collect it, because a completion gap becomes a selection gap.

That last one matters more than teams expect. If completion differs materially by group, the screening stage is filtering on availability and technology access rather than on suitability, and that flows straight into the outcome numbers a bias audit examines. Our guide to writing a screening rubric that survives a bias audit covers what those checks look for.

Seven fixes, roughly in order of payoff

1. Say exactly what is involved, in the invitation. Number of questions, realistic total time, whether re-recording is allowed, who reviews it, and when they will hear back. This one change usually moves completion more than everything else combined, because it converts an open-ended obligation into a task somebody can schedule.

2. Keep it short and keep the promise. Five to seven questions for a first-round screen is plenty. If the invitation says 15 minutes, it should take 15 minutes for a normal person who thinks before answering, not for someone who answers instantly. Overrunning the stated time is a broken promise at the worst possible moment.

3. Make a phone work with nothing installed. Browser-based, no download, resumable if the connection drops. Test it yourself on mobile data rather than office wifi, on both an iPhone and an Android device, before you blame candidates for not finishing.

4. Give a real window and let them start immediately. Five to seven days, with the option to begin the moment they open the invitation. A short deadline does not create urgency, it creates a scheduling filter that selects for free time.

5. Give something back before you ask. A short, honest video from the hiring manager about the role and team makes the exchange feel two-sided, and it costs one recording that every applicant sees. Teams without video capacity often produce these by turning the job description into a short video rather than filming anything. Salary range in the invitation works too, and works even better.

6. Send it fast. Interest peaks the day someone applies. An invitation that arrives six days later competes with every other process they started since, and the completion difference between same-day and week-later invitations is large.

7. Send one reminder, not four. A single reminder around 48 hours before the deadline recovers a real share of deferrals. A drip sequence reads as pressure from a company that has still not spoken to them.

Does a conversational AI interview complete better than a one-way recording?

The formats fail differently, and it is worth being precise rather than promotional about it. A one-way recording asks the candidate to perform to a camera with no response, which is the part people describe as uncomfortable, and it lets them re-record, which is the part people like. A conversational interview, by voice or video, feels closer to a normal phone call and asks follow-up questions, so it is a more familiar interaction, but it happens in real time and cannot be rehearsed.

Where a conversational format has a structural advantage is the effort imbalance. An AI interviewer agent responds to what the candidate says, which reads as a conversation rather than a submission, and voice removes the camera anxiety entirely for people who do not want to be filmed. Where it loses is with candidates who specifically wanted the control that re-recording gives them, and with anyone who objects to AI in the process at all, which our piece on whether candidates accept AI interviews looks at directly.

Neither format saves a bad invitation. A vague message with a 48 hour deadline for a role with no posted salary will underperform whatever technology sits behind the link. Our AI video interview page compares how the formats work in practice, and one-way video interview software covers the asynchronous approach on its own terms, including where it genuinely wins.

Give people a way out that is not dropping out

Some candidates cannot or will not do a video interview: a person with a disability the format disadvantages, someone in a shared living space with no private corner, someone who simply objects. Every one of them currently shows up in your data as abandonment, which means you are losing candidates and misreading why.

Offer an alternative in the invitation itself, whether that is a voice-only interview or a call with a recruiter, and make requesting it a one-click action rather than an email to a generic address. In Illinois, notice, an explanation of how the AI evaluates and candidate consent are required for AI-analyzed video interviews, and accommodation obligations apply everywhere. Beyond the legal floor, an accommodation route converts people who would have quietly disappeared. The state-by-state duties are set out in our guide to AI interview laws, and the employer-side obligations on our AI hiring compliance page.

Frequently asked questions

How long should a video screening interview be?

Ten to twenty minutes, covering five to seven questions. Beyond that, completion falls and answer quality does not improve, because a first-round screen is confirming availability, motivation and a few core competencies rather than making a hiring decision. If you need 40 minutes, you are running an interview, not a screen, and it should be scheduled as one.

When is the best time to send a video interview invitation?

As close to the application as possible, which matters far more than the hour of day. Where you do have a choice, weekday mornings outperform Friday afternoons, mostly because a Friday invitation with a 48 hour deadline effectively lands on a weekend. Same-day invitations beat carefully timed late ones every time.

Should we tell candidates an AI conducts the interview?

Yes. Illinois, New York City and Maryland require disclosure in specific forms, and elsewhere it is simply the practical default: candidates work it out within a question or two, and an undisclosed AI interview turns into a reputational problem long before it turns into a legal one. Disclosure up front does not reduce completion the way teams fear, whereas discovering it mid-interview does.

Does requiring video reduce the diversity of our shortlist?

It can, through access rather than intent. Requirements for a stable connection, a quiet private space, a desktop browser or a fast turnaround all filter on circumstances that correlate with income, housing and caregiving responsibilities. Measuring completion by segment tells you whether it is happening to you, and offering a voice-only or scheduled-call alternative is the cheapest way to prevent it.

Is a low completion rate always a problem?

Not always, but assuming it is self-selection is usually wrong. Some drop-off is candidates correctly deciding the role is not for them, which is useful. The test is whether the people abandoning look different from the people finishing. If they do not, the process is filtering on interest. If they do, it is filtering on circumstance, and you are losing candidates you wanted.