Video interviewing stopped being a question some years ago. It is standard, applicants expect it, and for a first round it saves everyone a commute.
What applicants object to is narrower. Being asked to record answers into a camera. No person on the other end. No idea who watches it. No reply afterwards. Most complaints about "video interviews" are complaints about that one format.
It matters more in a tight market. The Ministry of Statistics and Programme Implementation says "Unemployment Rate (UR) in usual status (ps+ss) was 3.2% for both male and female of age 15 years and above during July 2023 to June 2024." At that level, the people you want already have jobs and options.
What is the difference between live and one-way video?
| Live video | One-way async | |
|---|---|---|
| Candidate experience | A conversation. Broadly accepted | Recording into a void. Highest drop-off |
| Best used for | Any round where you assess a person | High-volume first-pass screening only |
| Scheduling cost | A slot both sides hold | None, which is the whole appeal |
| Risk | Low | Strong applicants with options do not finish |
The trade is real. Async saves your team hours. It costs you some of the best applicants, because people with 3 other processes running spend the evening on the one that offered a conversation.
Why do applicants abandon a one-way interview?
- No human on the other end. They cannot ask what a prompt means, so a badly worded question becomes a bad answer they cannot fix.
- Unclear stakes. Nobody said who watches it, how it is scored, or whether it replaces a real interview.
- Technology friction. Camera permissions, a browser that fails, a retake limit nobody explained.
- Silence afterwards. They recorded themselves answering questions and heard nothing. This is the one they tell other applicants about.
Does bandwidth rule out video in India?
Not usually, but setting does. A live round assumes a quiet room and a stable connection. Many strong applicants have neither on a weekday. Someone currently employed is often taking your call from a stairwell or a parked car. India's connectivity picture is tracked by the Telecom Regulatory Authority of India, and coverage is no longer the binding constraint. Privacy is.
Flexibility is not a nicety here. The Ministry of Statistics and Programme Implementation says "LFPR for female of age 15 years above in usual status (ps+ss) has increased from 37.0% during July 2022 to June 2023 to 41.7% during July 2023 to June 2024." A growing share of your pipeline is balancing interviews against caregiving. Two adjustments remove most of this. Offer an audio-only fallback and do not ask anyone to justify it. Offer slots outside working hours. A candidate joining on audio from a corridor is not less serious. They are a candidate with a current job.
What should the invitation tell them?
Everything they need to decide whether to spend 20 minutes. In the invitation, not discovered halfway through.
- How many questions, and the total time expected
- Whether retakes are allowed, and how many
- Who watches the recording, and what happens to it afterwards
- Which round comes next, so it is a step and not a substitute for being met
- A date by which they will hear back, either way
Keep it to 4 or 5 questions. Twelve prompts is not a screen. It is unpaid work, and the completion rate shows it.
How do you write prompts that work without a human?
Most async screens fail on question design, not technology. A prompt written for a conversation breaks when nobody can ask what you meant.
| Weak prompt | Why it fails | Better |
|---|---|---|
| Tell us about yourself | No scope. Produces a CV read aloud | Describe the payroll cycle you ran last: headcount, states, what you owned |
| How would you handle a difficult stakeholder? | Hypothetical. Rewards fluency, not experience | Describe a time someone rejected your numbers. What did you do next? |
| Why do you want this role? | Invites rehearsed flattery | What part of your current role would you not want to repeat here? |
Three rules hold. Ask about something done, not something hypothetical. Put the time limit in the prompt. Make each question stand alone, so a misread question two does not wreck question three.
Who should score the recordings?
One person, for the whole requisition. An async screen with no scorecard is a vibe check with extra steps. Decide the 2 attributes it tests, write what a strong answer sounds like, and score against that.
Consistency beats parallelism at the screening stage, and the volume is usually small enough for one reviewer to hold the standard.
What should you measure?
| Metric | What it tells you |
|---|---|
| Invitations sent against recordings completed | Whether the format costs you applicants |
| Completion by seniority | Senior drop-off is far worse. Stop using async there |
| Time from invitation to completion | Several days means it sits in a to-do list, losing to live rounds elsewhere |
| Async-screened against live-screened hires at 6 months | Whether the screen predicts anything |
If completion runs below half, the format is filtering on tolerance for your process, not on ability. For context, the same PLFS release says "Worker Population Ratio (WPR) in usual status (ps+ss) for persons of age 15 years and above was 58.2% during July 2023 to June 2024", per the Ministry of Statistics and Programme Implementation. Most of your pipeline is employed, and employed people abandon slow processes first. Statutory rules for the roles you are filling sit with the Ministry of Labour and Employment.
Use async only where volume makes live screening impossible, keep it to one early round, and never for senior roles. Everywhere else a 15-minute live call screens better and costs nothing in goodwill. Either way, keep the scheduling and the record against the candidate rather than in a mailbox, which is how 247HRM applicant tracking handles interview scheduling and panellist review tracking.