Privacy policy

What we record

Everything that happens in a session is written to a recording of that session:

• Your microphone audio, for the whole session. This is the recording of your voice, not a summary of it.
• The interviewer’s audio, the same way.
• Text transcripts of both sides of the conversation, including anything you say about yourself — your name, for instance, which the interviewer asks for at the start.
• The name you confirm at the start of the interview, kept as a field on the recording rather than only as a line in the transcript. The interview does not start until you have confirmed it, and the name is whatever you agree to or type — it does not have to be your legal name, or your full one.
• The full contents of the code editor, at every change. This is worth stating plainly, because “the session is recorded” does not imply it: we do not keep only your final answer. Every intermediate state of the editor is recorded as you type it, along with anything the interviewer writes into it.
• Any feedback you write on the screen at the end of the interview. It is stored alongside the recording of the session it was about, so a note you leave is kept the same way, in the same place, as everything else here.
• Session metadata: when the session started and ended, how long it ran, which coding question you were given, which programming language you chose to write in, which AI model conducted the interview, and how many times the connection dropped and reconnected.
• Which browser you used, and how big its window was. Your browser sends a line of text identifying itself to every site it visits; we keep that line, along with the size of the window you took the interview in. We use it to work out whether the interview is failing on particular kinds of device — on a phone, for instance, where the code editor is hard to use. It is stored on the one recording it came from, and it is not used to recognise you, to link one session to another, or to build a profile of you.

We do not ask you for an account, an email address, or any other identifying detail, and we do not record one. A recording is identified only by the moment it started, the question you were asked, and a random string. We do not record your IP address or your location, and nothing we keep is meant to follow you from one session to the next.

Advertising

We advertise SignalLoop, and we report how far an interview got to Google Ads — so that we can tell which ads bring people who go on to actually use it. Six steps are reported: starting an interview, confirming your name, choosing a language, writing your first line of code, writing five, and reaching the end of the session. A report is a step’s name and the time it happened, and nothing else.

This happens in your browser, through an advertising tag from Google. The tag is loaded from Google on the home page and on the interview page, and it stores cookies on your device so that a conversion can be matched to the ad you clicked. It is the only third-party script we load and the only thing that stores anything in your browser. It runs on those two pages and nowhere else: this policy page and the rest of the site load no tag and set no cookies.

In particular, none of the following is reported to Google Ads or to any other advertising service: your name, anything you said, anything you typed, any audio, any transcript, and the identifier of your recording. There is nothing in what we send that could be used to join an advertising click back to your interview.

What we do not collect

Apart from the advertising tag described above, SignalLoop stores nothing in your browser: no other cookies, and nothing kept in local storage between visits. We load no analytics of any kind, and no third-party script other than that one. We have no accounts and no logins, so there is nothing here that recognises you as the same person from one visit to the next.

The microphone check before the interview records nothing. It turns your microphone on to measure how loud it is and how quiet your room gets, and those measurements never leave your browser — no audio is kept, and nothing is uploaded. You have to pass it to start an interview, so it runs for everyone; it is over before the session opens, and the recording described above begins when you press Start.

Where the recording goes

Recordings are written to private storage that we control, not to the machine that served your interview. They are not published anywhere, and there is no public link to a recording.

Who else processes it

Google. The interviewer is Google’s Gemini live model, so for the duration of the interview your microphone audio is streamed to Google’s Gemini API, which is what produces the interviewer’s speech and the transcripts. Google is processing your voice in real time while the session is running, under their own terms for that API. We name them rather than referring vaguely to third parties, because they are the only one.

The storage described above is operated by the same company, acting as our storage provider, and the advertising described above is theirs as well — three roles, one company. Apart from those, we do not sell, share, or otherwise hand your recording to anyone. Nothing of your recording goes to the advertising side of it.

Who can listen to it

We do review recordings, to improve the interviewer. There is no review portal, no shared link, and no way for another candidate or a member of the public to reach your session. In practice, the people who can listen to your interview are the people who operate SignalLoop, working directly on the systems that hold it.

How long we keep it

A month at most. Recordings older than that are deleted.

Deletion requests

We do not currently offer deletion of a recorded session on request, and we would rather tell you that than promise something we could not carry out. A recording carries the name you confirmed at the start, but nothing that establishes it is yours: no email address, no account, no verification of any kind, and you are never shown the recording’s identifier. Acting on a name alone would mean deleting somebody else’s session on the word of whoever asked. What we can tell you is that it goes within a month either way, as above.

The practical control you have is the one before the session starts. Everything you say is recorded, so please keep sensitive or personal information out of the conversation, and do not type anything into the editor you would not want kept.

Your copy of the interview

When the interview ends, you can save a copy of your code and the transcript as a single text file. That file is assembled in your browser from what is already on the screen and written straight to your own machine. Nothing is uploaded when you press it, and no copy of that file reaches us — it is yours, and it is the only copy you get, so it is worth taking.