Sit the interview before the interview.
Rehearse the round you’re about to face, out loud and against the clock, with an interviewer who follows up the way real ones do. Afterwards, every answer is scored against your own words.
You can do the job. Now prove it in forty-five minutes.
Nerves lower interview scores, and they say little about how people go on to do the job. What moves the score is practice out loud, with feedback.
Reading a list of likely questions doesn’t give you that. A list never asks, “And what did you do?”
See the research
From job posting to feedback in one sitting
Four steps, and the first one takes about a minute.
Tell it the job
Paste the posting or type the title, at the level you’re aiming for. Add your CV and the interview is built around your own record.
Choose the round
Screening, technical, behavioural or final, for fifteen minutes or a full hour. Set the pressure to comfortable, realistic or challenging.
Sit it
Out loud, with an interviewer who listens and presses where an answer thins out. The last minutes are yours for your own questions, as they would be in the real one.
See the score, then go again
Every answer is scored and quoted back to you, with the one change that would make the biggest difference. Then retake any question and compare the two.
Interviews are won on the follow-up
Most candidates rehearse a first answer. The real test is the second question: “How exactly?”, “What did you do?”, “And then?” Say “we”, and your interviewer will ask what you did. Mention a project on your CV and expect to be asked about it.
Questions that keep pace with hiring
Hiring has changed quickly. Google, Cisco and McKinsey have brought back in-person rounds for some candidates, and Canva has rebuilt its technical screen around AI-assisted coding.
Rosby’s research pipeline tracks shifts like these. It studies what employers ask for each role and level and refreshes the questions as the market moves, so you never practise for a test that no longer exists.
See the reporting“Tell me about an architecture decision you led that later had to be revised when new technical or operational evidence emerged.”
A score that shows its working
Each answer is scored on what you said, how clearly you owned it and how well you structured it.
Every point sits beside the words it came from, so you can see where the score came from and what to change.
“We moved the queue to Postgres because the team knew it. I wrote the rollback plan before we shipped, so when it backed up we were out in four minutes.”
- LandedYou said what you did, not what the team did.
- ChangeLead with the four minutes. It’s the number they’ll remember.
- What you said30 of 40
- What you owned25 of 30
- How clearly20 of 30
A sample, written for this page.
Go again, and hear the difference
Retake any question and Rosby sets your two answers side by side: how long each ran, how soon you said “I”, and the outcome you named.
“I think that was better” becomes something you can see.
| Side by side | First | Again |
|---|---|---|
| How long the answer ran | 212 words | 131 words |
| Words before you said “I” | 48 | 6 |
| The outcome you named | none | out in four minutes |
| Softening phrases | 5 | 1 |
A sample, written for this page.
Questions people ask
Which roles can I practise for?
The one you’re going for. Paste the posting and Rosby builds the interview around it, whether it’s your first role or your next step into leadership.
Is it like a real interview?
As close as practice gets. The round has the shape of a real one, the follow-ups come as they would, and you decide how much pressure to face.
Is my interview recorded?
No. Your voice is never kept, and your camera never leaves your browser. Rosby keeps a written transcript so the feedback can quote you.
What happens to my CV?
Rosby uses it only with your permission and stores it encrypted. It’s deleted when you discard the round or close your account.
How long does it take?
Anything from a fifteen-minute warm-up to a full hour. You choose before you start.
Where the evidence comes from
Every finding on this page comes from published research or on-the-record reporting. Each card links to the original.
Research
Nerves lower interview scores
A meta-analysis of 30 studies and 3,428 people found that the more anxious the candidate, the lower the interview score.
Powell, Stanley & Brown (2018), Canadian Journal of Behavioural Science 50(4)Nerves say little about the job
Studies that followed candidates into the job found that their interview nerves had almost no bearing on how well they performed.
Schneider, Powell & Bonaccio (2019), IJSA 27(4); Zhang, Powell & Bonaccio (2022), IJSA 30(2)Practice with feedback helps
People who practised with a virtual interviewer that gave feedback did better in a later interview, and candidates who were coached earned higher ratings.
Langer et al. (2016), IJSA; Tross & Maurer (2008), Journal of Occupational and Organizational Psychology
Reporting
In-person rounds are back
Google, Cisco and McKinsey have reinstated in-person interviews for some candidates, in part to counter cheating with AI.
Computerworld, 26 August 2025Technical screens, rebuilt around AI
Canva replaced its computer science fundamentals screen with an interview in which candidates are expected to code with AI assistance.
Canva Engineering Blog, 11 June 2025