Watching the Interview Recording Behind a Shared Scorecard

A hiring manager at a logistics company told one of our pilot recruiters that she reads only a minority of shared scorecards start to finish, and watches the recording behind even fewer of those, because in her words, “if the transcript already tells the story, watching the video is just reading it slower.” She’s right about most candidates. She’s also, without quite framing it this way, describing exactly the calls where a recording is worth the four extra minutes: the ones where the transcript doesn’t tell the whole story on its own.

What’s actually on the page

A shared scorecard carries the verdict, the scored dimensions with the reasoning behind each one, any recruiter notes and flags, the speech-quality read, and the full transcript, all covered in detail in how sharing a scorecard actually works. What that post doesn’t go deep on, because it’s a genuinely separate question from how the link itself gets shared and revoked, is the recording sitting alongside all of it: a short-lived video link to the interview as it actually happened, not a summary or a transcript of it.

This kind of “the mechanism does one job well, and a related but separate job lives in its own place” pattern shows up elsewhere in the product too. A JD’s lifecycle mechanics work the same way: archiving a job description and renaming one mid-campaign are two distinct, narrowly scoped actions rather than one catch-all “manage this JD” button, for the same reason a scorecard’s sharing mechanics and its recording-viewing mechanics are described separately here rather than as one undifferentiated feature.

That link is the part this post is about. Text and video answer different questions, and knowing which one you actually need saves a hiring manager real time across a week of scorecard reviews.

What a transcript can’t show you

A transcript is a faithful record of what was said. It is not a faithful record of how it was said, and for a genuinely borderline candidate, how often carries more signal than what.

Consider two answers to the same question that read identically on the page: “I led the migration, though there were some challenges with the legacy system integration.” Spoken with steady pacing and a confident close, that’s a candidate who handled a hard project and is comfortable describing it plainly. Spoken with a long pause before “challenges,” a half-laugh, and a trailing-off voice on “integration,” that’s a candidate who might be papering over something they’re not fully confident they actually did. The words are the same. The two-evaluator scoring already captures this distinction in the score itself, pace, filler words, and CEFR fluency are measured from the audio, not guessed at from the transcript, but the score is a number. If you want to hear the actual hesitation that number is describing, rather than trust it on faith, the recording is the only place to hear it.

When reading is genuinely enough

Most scorecards don’t need this. A clear Strong Go with confident, specific answers reads as confident on the page too, and watching four minutes of video to confirm what the transcript already tells you plainly is time a busy hiring manager doesn’t need to spend. The same goes for a clear reject: if the phase-0 knockout questions ended the call in the first two minutes because a hard requirement wasn’t met, there’s very little tone or delivery nuance left to recover by watching, the transcript says what happened and why.

Watching earns its place specifically on the On Hold band, the candidates whose scores sit in the middle precisely because something about the call didn’t resolve cleanly one way or the other. That’s where a hiring manager’s four minutes are best spent, not spread evenly across every scorecard they open in a week.

A concrete workflow: what to actually watch, not the whole call

Nobody needs to rewatch a 15-minute interview start to finish to get the value here. The scored dimensions on the page point at specific moments, the answer that drove a low score on a particular skill, or the section where the speech-quality read flagged unusual hesitation. Skip to that part of the recording rather than starting from zero. A hiring manager checking whether a middling “communication” score feels fair only needs the two or three minutes the score is actually describing, not the candidate’s full answer about their previous role, which likely scored fine and doesn’t need re-litigating.

This is also the right moment to compare what you hear against what the reading-your-scorecard guide tells you the numbers mean in general. A score without the recording behind it is a claim you’re taking on trust; a score you’ve spot-checked against thirty seconds of the actual audio is one you’ve verified for yourself, which is a meaningfully different level of confidence going into a final call.

Why the link is short-lived, on purpose

Unlike the scorecard page itself, which stays reachable until someone actively regenerates the share link, the video link specifically expires. That’s a deliberate asymmetry, not an inconsistency. A transcript is text; a recording is a candidate’s actual face and voice, and explainers on voice as a special category of personal data make the point clearly: a voice recording is biometric, tied permanently to one person in a way a transcript of the same words is not, and can’t be reset the way a password can if it’s ever handled carelessly. Systems handling that kind of data should minimize how long it stays casually reachable, not treat it the same as any other file.

A forwarded link that works forever is a link that outlives the reason it was created. Someone opens a scorecard to make one hiring decision; the transcript and scores stay useful as a permanent record of that decision, but the raw recording doesn’t need to stay one tap away six months later, forwarded into a group chat nobody remembers exists. Limiting the video link’s life while keeping the scorecard itself durable and revocable, the mechanics covered in the share-scorecard post, reflects that difference honestly rather than treating every part of the page as equally low-stakes to leave lying around.

We won’t publish the exact expiry window here, on purpose, for the same reason we don’t publish the share token’s format: the specific number isn’t the useful part to make public, and it isn’t what determines whether the design is sound.

Calibrating a rubric by ear, not just by score

There’s a second use case for the recording that has nothing to do with a single hiring decision: calibration. A recruiter who’s new to a role, or a hiring manager seeing HireQwik scores for the first time on their own req, often has no intuition yet for what a middling score versus a clearly strong one actually sounds like in practice. Reading the number alone doesn’t build that intuition; hearing two or three On Hold candidates back to back does, because the ear picks up the pattern faster than a spreadsheet of scores does.

We’ve seen TA leads use this deliberately at the start of a new campus drive: pull three or four On Hold recordings from the first day’s interviews, listen to them together with whoever’s reviewing that week, and use them to agree on where the team’s actual bar sits before the volume ramps up. That conversation is much harder to have over a shared spreadsheet of numbers than over three minutes of actual audio everyone in the room just heard. It’s a training exercise more than a screening one, but it makes every scorecard review after it faster, because the team now shares a real reference point for what the rubric is actually measuring, not just an abstract description of it.

Watching on a phone, between meetings

The same mobile-first design that makes the scorecard page usable one-thumbed applies to the recording too, which matters more for video than it does for text. A transcript is comfortable to skim on a small screen in a spare two minutes. A video wants a moment of actual attention, which is harder to find standing in a hallway between a client call and a status meeting, the exact situation the shared-scorecard mechanism was built for in the first place.

The practical fix isn’t a smaller player or a faster codec, it’s the same one described above: skip straight to the moment the score is describing rather than opening the recording cold from the start. A hiring manager with ninety seconds between meetings can watch the specific thirty seconds that produced a borderline communication score. The same hiring manager does not have ninety seconds for a full fifteen-minute replay, and expecting them to find that time is how the recording ends up unwatched, and the decision ends up made on the score alone, which was the original problem a shared, evidence-backed link was supposed to solve.

When the recording matters for a reason that isn’t the score at all

There’s a category of case the score itself doesn’t capture, and it’s worth naming directly: sometimes what a hiring manager needs to verify isn’t whether the number is fair, it’s what actually happened on the call. A candidate disputes a rejection and says the connection dropped mid-answer. A recruiter wants to confirm a knockout question was asked clearly, in plain language, and not buried in jargon a nervous candidate might have misheard. A hiring manager wants to hear, in the candidate’s own voice, how they described a specific project before extending an offer that hinges on that experience being real. None of these are about auditing the score. They’re about wanting the actual record of an actual conversation, the same instinct that makes a recruiter ask for a call recording rather than a summary when something feels off, whether or not HireQwik was involved.

The transcript technically answers all three questions too, in the sense that the words are there. But “what actually happened” is a question about the encounter, not just its content, and a candidate who says “the call dropped” is making a claim about the audio, not about what was said before it dropped. In cases like these, the recording isn’t a richer version of the transcript. It’s a different kind of evidence entirely, closer to a phone log than a meeting summary, and it’s the only thing in the product that answers those questions directly rather than by inference.

What this doesn’t replace

Watching a recording is a spot-check, not a re-interview. It’s not meant to replace the two-evaluator score with a hiring manager’s own gut read of thirty seconds of video; it’s meant to help that hiring manager trust or question the score with actual evidence in hand, rather than deciding based on a number alone or on a full independent replay of a 15-minute call they don’t have time for. If a hiring manager finds themselves routinely rewatching entire interviews rather than spot-checking specific moments the score points to, that’s usually a sign the JD’s rubric needs a closer look, not that every future candidate needs the same full-length treatment.

It’s also not a substitute for good judgment about when a decision genuinely calls for more than a scorecard at all. A borderline call on a senior, high-stakes hire is sometimes worth a live conversation with the candidate directly, recording included, rather than a decision made entirely on the strength of an asynchronous review. The recording lowers the bar for getting real evidence in front of a second opinion quickly; it doesn’t remove the value of talking to a promising candidate yourself when the role warrants it.

The take

A transcript answers “what did the candidate say.” A recording answers “how did it actually sound,” and that second question only matters on the calls where the first one didn’t already settle things. Reserve the four minutes for the On Hold band, skip to the moment the score is describing rather than replaying the whole call, and treat the recording as evidence for a decision you’re already leaning toward, not a fresh interview from scratch. Talk to us to see what a shared scorecard, recording included, looks like on your own candidates.