Short answer: For first-round screening in Canada, an AI interview usually gives you the fastest time-to-screen and the most consistent questions, a one-way video interview gives you consistency without the conversation, and a phone screen gives you a live human connection at the highest recruiter cost per candidate. Whichever you choose, structure (same job-related questions, same rubric) matters more than the format, and the AI-assisted options bring Ontario disclosure and Quebec automated-decision duties with them.
#The three formats, defined
- Phone screen: a live call, usually 15 to 30 minutes, between a recruiter and a candidate. It's synchronous: both people have to be free at the same time.
- One-way video interview (also called asynchronous or on-demand video): the candidate reads or hears pre-set questions and records answers on camera, usually against a timer. People watch the videos later. Some products also run AI analysis on the videos.
- AI interview: an AI interviewer asks structured questions aloud and listens to the answers, and the platform turns the answers into a transcript and an evaluation for a person to review. Like one-way video, candidates can take it at any time.
With one in five Canadian businesses (opens in a new tab) (19.2%) now using AI, according to Statistics Canada's second-quarter 2026 data, many hiring teams are weighing exactly this choice.
#Side-by-side comparison
| Phone screen | One-way video | AI interview | |
|---|---|---|---|
| Scheduling | Needed for every candidate | None: candidate chooses the time | None: candidate chooses the time |
| Recruiter time per candidate | Full call length plus scheduling and notes | Watching each recording | Reviewing a scorecard, and the recording where needed |
| Consistency of questions | Varies by recruiter and day unless scripted | Identical questions | Identical questions (order can be randomized) |
| Consistency of evaluation | Depends on notes and memory | Depends on reviewers and any rubric | Same rubric for every candidate, with quoted evidence |
| Candidate experience | Human, but hard to schedule | Often disliked: talking to a blank screen | Conversational, but must be clearly disclosed |
| Accessibility | Hard for Deaf or hard-of-hearing candidates unless relay is offered | Camera and timers can be barriers | Needs captions and alternatives (no video, typed answers, a person) |
| Ontario AI disclosure | Not triggered | Triggered if AI screens or assesses the videos | Triggered |
| Quebec s.12.1 | Not applicable | Applies only if a decision is exclusively automated | Applies only if a decision is exclusively automated |
| Audit trail | Recruiter notes | Recordings | Recordings, transcript and scorecard |
We haven't put numbers in this table on purpose. There's no reliable, independent Canadian dataset comparing these formats head to head, and figures from vendors (including us) are based on their own customers. The sections below show how to work out your own.
#Time-to-screen: where the days go
Time-to-screen is the gap between "candidate applies" and "we have enough evidence to decide on a next step". The interview itself is rarely the slow part. The waiting is.
A phone screen needs two calendars to line up. Each round of "does Tuesday at 2 work?" can add a day, and every no-show restarts the cycle. For a recruiter with 40 applicants for a contact-centre role, the calls themselves might fit in two days, but finding the slots can take a week or more.
Asynchronous formats remove that step. The candidate can interview the evening they're invited, which matters for people working shifts. The remaining delay moves to review: someone still has to watch videos or read scorecards. That's why the review format matters as much as the interview format. Watching 40 eight-minute videos is about five hours of viewing; reading 40 scorecards with quoted evidence, and opening the recording only where something needs a closer look, is usually much faster.
Our high-volume contact-centre hiring playbook shows how this plays out across a full hiring sprint.
#Cost per screen: how to calculate it
The honest way to compare cost is to price your own recruiter time. Job Bank lists a median wage of $33.33 an hour for recruiting officers in Canada (opens in a new tab) (2023–2024 data, last updated November 2025).
Use this formula for each format:
- Recruiter minutes per candidate = scheduling + interview or viewing time + notes and follow-up.
- Labour cost per screen = (minutes ÷ 60) × hourly wage × (1 + benefits and overhead loading).
- Add tool cost per screen: the subscription divided by screens used, or the per-interview price.
- Adjust for no-shows: divide by the share of scheduled screens that actually happen.
As an illustration only: a phone screen that takes 30 recruiter minutes, at the Job Bank median, is about $16.67 in wages per candidate before benefits, overhead and no-shows. If an AI interview cuts recruiter time to the 5 or 10 minutes it takes to review a scorecard, the labour cost falls proportionally, and you then add the per-interview price. Plug in your own minutes; they vary widely by role.
#Consistency and fairness
The strongest evidence in hiring science favours structure. As summarized in SIOP's newsletter, Sackett and colleagues' 2022 meta-analysis found that structured interviews had the highest mean operational validity (r = .42) (opens in a new tab) of the common selection methods studied, and the authors' follow-up concluded that structured interviews "emerged as the predictor with the highest mean validity" (opens in a new tab).
What does that mean for each format?
- Phone screens drift. Different recruiters ask different follow-ups, and notes vary. A script and a scoring sheet fix most of this, but in practice many phone screens are informal conversations.
- One-way video keeps the questions identical, but evaluation depends on whoever watches, and on how tired they are by video 30.
- AI interviews keep questions and the rubric identical. The risk moves to the AI's evaluation, which you should test: compare its scores against experienced human raters, and check results across groups. Our guide to benchmarking and bias-auditing an AI interviewer walks through how.
Whatever the format, avoid judging on appearance, accent, facial expressions or "energy". Those signals aren't job-related and can disadvantage people with disabilities and people from different cultures.
#Candidate experience
Candidates tend to judge screening formats on three things: whether they were treated as a person, whether they knew what was going on, and whether they heard back.
Phone screens score well on the first, and poorly when scheduling drags. One-way video often scores poorly on all three: a countdown, a webcam and silence. AI interviews can go either way. Clear disclosure, a conversational interviewer, live captions, control over when an answer is finished and a visible route to a human make the difference. We cover this in detail in what candidates want from AI interviews.
#Compliance in Canada (as of November 2026)
This is general information, not legal advice.
- Ontario job postings. Since 1 January 2026, employers with 25 or more employees must disclose in publicly advertised postings if they use AI to screen, assess or select applicants (opens in a new tab). They must also tell interviewed applicants within 45 days of the last interview whether a decision has been made. The disclosure applies to AI interviews and to AI-analyzed one-way video; a phone screen alone doesn't trigger it.
- Quebec. Section 12.1 of Quebec's private-sector privacy law applies when a decision is based exclusively on automated processing (opens in a new tab): you must inform the person and, on request, explain the factors and let them submit observations to someone who can review the decision. A genuine human decision keeps you outside the "exclusively automated" case.
- Accessibility. In Ontario, employers must tell individually selected applicants that accommodations are available on request (opens in a new tab) for the materials or processes used in assessment and selection. That applies to every format.
- US candidates. If you hire into Illinois roles and AI analyzes recorded video interviews, the Artificial Intelligence Video Interview Act (opens in a new tab) requires notice, an explanation of how the AI works, consent, limited sharing and deletion within 30 days of a request. Our guide to US AI hiring laws for Canadian employers covers New York City, Colorado and California too.
#Which format should you use?
- Fewer than about ten candidates per role, or senior roles: a structured phone or video call with a person is usually worth the time.
- Steady or high volume with similar roles: an asynchronous format saves scheduling. Prefer one that gives reviewers structured evidence rather than a stack of videos to watch.
- Candidates who work shifts or across time zones: asynchronous formats are kinder, as long as a human alternative is available.
- Any format: write job-related questions, score with a rubric, disclose AI use where required, offer accommodations, and keep a person making the decision.
#Where MSZ Hire fits
MSZ Hire runs structured AI voice interviews in Chrome or Edge on a computer. The AI interviewer asks every candidate the same job-related questions aloud, in a randomized order, and the hiring team gets an evidence-based scorecard with quoted answers. The AI is advisory and a person makes every decision. Pricing is a monthly plan with a set number of interviews (unused ones roll over, up to twice your monthly allowance) plus top-ups when you need more, so the per-screen tool cost in the formula above is easy to plug in. See how it works, pricing and how MSZ Hire compares with other AI and video interview platforms, or start free with five free interviews.
Written by the MSZ Hire Editorial Team, MSZ Systems Inc., Toronto. Last reviewed 17 November 2026. General information, not legal advice.