> ## Documentation Index
> Fetch the complete documentation index at: https://support.metaview.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Screening Overview

> Hear from every candidate with short AI-led screening interviews, scored against your criteria.

Screening lets you run short AI-led screening interviews with your applicants. You build a screen for each job, invite candidates to take the interview from their browser, and every completed interview is scored against your criteria with transparent reasoning. This means you can hear from every candidate, not just the shortlist, and make better-informed decisions about who to move forward.

## How Screening Works

To create your first screen, open Metaview, click **Screening** in the sidebar, and click `New screen`.

<Steps>
  <Step title="Select a role and stages">
    Pick the role you want to screen for from your connected ATS, give the screen a name, and choose the pipeline stage to import candidates from. If your ATS isn't connected yet, the setup flow walks you through it. Candidates in that stage appear in Metaview, but nobody is contacted until you enroll them. You also choose how decisions flow back to your ATS: the stage progressed candidates move to, the stage rejected candidates move to, and the rejection reason and email template used when candidates are rejected.
  </Step>

  <Step title="Share context">
    Attach a job description or other role documents, and add any extra context about what you're looking for. This step is optional but recommended, as it produces a much better first draft of your interview.
  </Step>

  <Step title="Refine the interview plan">
    Metaview drafts an interview plan for you: a series of steps, each with a goal and one or more questions. Edit anything directly, or use `Agent chat` to describe changes in plain language and review the agent's suggestions. Each question has a few settings:

    * **Format**: `Conversational` questions are open spoken questions, `Multiple choice` questions present options on screen, and `Scenario: Conversation` questions play out a short realistic exchange.
    * **Follow-up**: an optional probe to draw out specifics. Leave it blank to let the interviewer decide.
    * **Rubric**: an example of what a strong answer looks like. This is what the candidate's answer is reviewed against, so the more specific you are here, the more useful the scoring.
  </Step>

  <Step title="Set up the candidate experience">
    In the `Setup` card you can choose the interview language and voice, upload your logo, customize the welcome message, record or upload a short intro video, and make an active camera mandatory for the interview.
  </Step>

  <Step title="Preview and test">
    Click `Preview as candidate` to take the interview exactly as a candidate would, or use the preview pane to test a single question or step. Previews are free and never consume credits.
  </Step>

  <Step title="Publish and enroll candidates">
    Click `Publish` to make the interview live, then enroll candidates from the `Candidates` tab. You can enroll them individually, in bulk, or turn on `Auto-enroll` to automatically enroll every current and future candidate from your ATS stage. Enrolled candidates receive an invitation email with a personal link, plus reminders if you've set a deadline. See [Screening Setup](/screening/setup) for the full invitation and deadline options.
  </Step>

  <Step title="Review results">
    Each completed interview is assessed against your rubrics and rated `Great`, `Good`, `Okay`, `Poor`, or `Inconclusive`, with a written summary and per-question notes. Open a candidate to watch the recording, read the synced transcript, and see exactly which answers the scoring is based on. When you're ready, `Progress` or `Reject` candidates and Metaview moves them to the right stage in your ATS. You always make the final call — Metaview never auto-rejects anyone unless you've explicitly turned on rejection at the deadline, and that's triggered by a missed deadline, never by a rating.
  </Step>
</Steps>

## Ratings and Scorecards

Every completed interview gets a scorecard. It includes an overall rating and summary, and for each question: how well the answer matched your rubric, factual notes on what the candidate said, what remains unknown, and links to the exact moments in the transcript. Ratings are based on your own rubrics rather than a generic standard, and they're intended to support your review, not replace it.

Ratings are produced from the transcript of the conversation — there's no video analysis, and no audio analysis beyond turning speech into text — so the strongest rubrics describe the substance of a good answer rather than its delivery. Recordings are there for your playback, not the AI's. For more on writing rubrics and using Screening responsibly, see [Screening Best Practices](/account-management/privacy-and-security/ai-best-practices-screening).

If you edit and republish your interview plan, scorecards are regenerated so results stay comparable across candidates.

## Detecting AI Use and Misrepresentation

Screening automatically checks each interview for signs that answers were read from an AI tool, and signs that the person interviewed may not match the profile they applied with. Each signal is shown as low, medium, or high risk, with reasoning and links to the relevant moments in the interview.

<Warning>
  These signals are indicators, not conclusions. They never change a candidate's rating, they're never grounds for a rejection on their own, and you can override a signal with your own assessment and justification after reviewing the interview.
</Warning>

## What Candidates Experience

Candidates receive an email invitation with a personal link and take the interview from their browser, on desktop or mobile, whenever suits them before the deadline. There's nothing to schedule. If they get disconnected, they can rejoin within a few minutes and pick up where they left off. For the full candidate-side walkthrough, see [the candidate guide](/screening/candidate-guide) — it's written for candidates, so you can also link it in your own communications.

## Screening Credits

Screening credits are used to pay for interviews: one credit is consumed each time a completed candidate interview is evaluated. Building, publishing, and previewing screens is free. Every workspace includes 25 free credits, and you can buy more from **Screening** → `Credits`. See [Screening Credits](/screening/credits) for the full details.

***

## Troubleshooting

<AccordionGroup>
  <Accordion title="My candidate count doesn't match my ATS">
    A screen imports candidates from one specific stage of one role, so comparing against a different stage or the whole pipeline won't line up. New applications and recent stage changes in your ATS can also take up to \~30 minutes to appear in Metaview. If the numbers still look off after accounting for this, contact support with the role name and workspace.
  </Accordion>

  <Accordion title="A candidate needs another attempt">
    Click the `…` menu next to the candidate and select `Reset attempt`. This removes their previous interview so they can take it again using their invite link, and re-opens their invitation and reminder emails.
  </Accordion>

  <Accordion title="I can't enroll a candidate">
    The most common causes are:

    * **Unconfirmed email** — candidates need a confirmed email address before they can be enrolled. Click `Confirm email` on the candidate's row to pick from the addresses in your ATS or enter one.
    * **No credits remaining** — enrolling candidates requires available screening credits. Add credits from **Screening** → `Credits`.
    * **Unpublished screen** — the interview must be published before candidates can be enrolled.
    * **Opted out** — candidates who have opted out of AI screening can't be enrolled unless you opt them back in.
  </Accordion>

  <Accordion title="A result shows as Inconclusive">
    This means there wasn't enough in the interview to assess against your rubrics — usually because the interview was cut short or the candidate didn't engage with the questions. Open the `Interview` tab on the candidate to see the recording and transcript, and use `Reset attempt` if you'd like to give them another go.
  </Accordion>
</AccordionGroup>