# Questions — Criterio Talent > What the interview is like, how it gets built, how every conclusion is backed, and how you get started. The questions talent teams ask, answered. URL: https://criteriotalent.com/en/faq/ --- **Questions** # What teams ask before they start. Four blocks: the interview, the process, the results, and getting started. Short answers, and none we cannot back by showing you. **The interview** ## What is the interview like for the person taking it? It is a conversation. They open their own link whenever it suits them, talk by voice and video with an interviewer that listens, and answer in their own words instead of filling in a form or recording themselves against questions nobody picks up on. Nothing has to be scheduled with anyone, so the first conversation stops depending on two calendars lining up. ## Does it follow up, or run a fixed questionnaire? It follows up. When an answer stays on the surface — a situation with no action of their own, an outcome with no numbers — it comes back to it and asks for what is missing. And it knows when to stop pushing: if a second follow-up does not get there, it closes the point and moves on. Pressing three times on the same thing does not produce evidence; it produces an uncomfortable person. ## How long does an interview run? As long as your process needs. It is declared in the script, from five minutes up, and the time per question is shared out so that each answer gets somewhere. The maximum for your implementation is agreed before you start, and raising it later is configuration. Duration is set per interview, not per account. The same company can run a short opening round and a deep one for finalists. ## Does it work for senior roles, or only for volume? Both. What changes between them is depth, not the product: an operational role usually resolves with a few items and a short conversation, while a management or director search asks for more time, finer competencies and a more demanding rubric. The mechanics are identical either way: script, conversation, evidence, report. **The process** ## How much work is it to build an interview? You paste the job post and an assistant builds the whole definition: competencies, items and levels. From there you adjust it by talking to it — “drop this competency”, “also ask about supplier management” — and it applies the change. When a change destroys something, it proposes it and waits for confirmation before touching anything. ## Does the same interview work for every candidate for a role? Yes, and that is exactly what makes the results comparable. The definition stays versioned: every candidate for a role answers the same one, and an interview from March still says which version it ran with. ## Can one role have several rounds? Yes. A role carries ordered rounds, each with its own name and its own interviewer profile, and the same person walks through them without turning into three different candidates: they remain a single application. The name and position of each round are frozen into the interview once it has run, so renaming or reordering it later does not rewrite what already happened. The questionnaire comes from the role: today all of its rounds run the script published for it. ## How does someone move from one round to the next? A person does it from the platform, or the system the company already runs does it through the API. At that moment the next round’s interview is issued with its individual link, which you get to the person wherever you already talk to your candidates. The step is recorded with who took it and when. **The results** ## What is in the report? The level assigned to each competency, why it was assigned, and the fragment of the conversation that supports it. Below that, the full transcript and the recording, in case anyone wants to go and listen. It also carries what is worth picking up in the next conversation: whatever was left half-answered, written as a working list for whoever interviews afterwards. ## How do I know a quote in the report is real? Because what appears in quotation marks is not what the model wrote: it is the literal slice of the transcript. The model only points at where to look, and what gets published is cut from the turn itself, word for word. The consequence is the part that matters: an invented quote cannot reach the report however much the model insists, because there is nothing to cut it from. ## How do you compare two candidates? Side by side, requirement by requirement, with each one’s evidence in view. Since they all answered the same definition, the comparison is between answers rather than between the impressions of whoever interviewed them. The role board summarises the whole process and exports carrying the company’s identity. ## Who decides? Your team, with the reading already done. Advancing or rejecting is the action of an identified person, and it is recorded with their name, the time and their reason. That is the point of everything above: the team’s judgement gets spent on deciding, not on reconstructing what was asked in each conversation. **Getting started and trust** ## How do I start? With a demonstration on a real role of yours. We paste the job post, look at the script that comes out of it, walk through an interview, and open the report with its evidence. That session is where the scope gets written down: how many roles, how many candidates, and which integrations are in. ## Does it connect to the system we already use? Yes. There is an integration API with five operations — sync the role, request the interview, check on it, retrieve the report, and read the interviewer catalogue — plus eight signed events that report every change so nobody has to keep asking. The connector to a specific system is defined and implemented per project: which fields travel, in which direction, and what happens when the other side does not answer. ## What is the candidate told before it starts? That they will be talking to a system, what is recorded, what it is used for and how long it is kept. The interview does not begin until they expressly accept. It is read in full on one screen, before the first second of recording. ## Where does the information live, and who can see it? Each client operates in its own environment, with permissions by area and by role, and a log of who opened a report or downloaded a recording. The retention period is agreed and enforces itself. The detail is on the Security page, and the statement of measures for your instance is delivered in writing before signing. ## How much does it cost? It depends on the volume of roles and candidates, on which integrations are in scope, and on the service level contracted. It is closed in writing before anything starts, and there are no charges that show up later. The demo produces a concrete figure for your case, because by then it is clear how many roles you will move and at what depth. Your instance can also carry its own caps on spend, on simultaneous interviews and on starts per hour. When one is exceeded, the interview is refused at that moment and nothing is charged: the cap is a cap, not an alert that arrives late. **Next step** ## See it with a role of yours on the table. Thirty minutes: an interview is defined from your job post, walked through the way the candidate sees it, and a report is read with its evidence.