A single job ad can attract dozens of applications, and yet the decision to invite someone to an interview usually comes down to a handful of facts that a CV does not contain. When can the candidate start, what kind of contract are they looking for, do they hold the required licence, will they make it to a shift that begins at six in the morning. A voicebot can ask those questions on the recruiter's behalf, before anyone picks up the phone. Below we walk through how such a pre-screening process is put together – the flow of calls and text messages – and what you need to prepare to run it.
Why a CV is not enough to decide on an invitation
A CV is the candidate's marketing document, not an answer to the recruiter's questions. It describes the past, while recruitment is settled in the present and the near future. That is why the first phone call looks the same in most processes: a dozen questions about things that could have been established in two minutes, and some of those calls end within the first minute, because the candidate is looking for work in a different city, on a different contract, or has already accepted another offer.
The facts most often missing from a CV are:
- availability – when the candidate can start and how long their notice period is,
- expectations and type of contract – employment, freelance or own business, shift work, weekend work,
- licences and documents – a driving licence of a given category, medical clearance, a certificate, a food handling permit,
- language – a CV says “B2 level”, and a two-minute call tells you how it really is,
- willingness to commute – distance, own car, work at a specific site.
These are closed questions, or very nearly closed. They need no intuition, no experience and no judgement of another human being. They only need someone to call and ask them – and that is the most expensive and most repetitive part of a recruiter's day.
What a pre-screening call run by a voicebot looks like
The flow is simple and deliberately short. An AI voicebot calls the candidate from the company number, introduces itself, says which vacancy it is calling about and makes clear that it is an automated system. Then it asks a fixed set of questions – usually four to eight – and records the answers. The call lasts two or three minutes. At the end the bot says what happens next: that the answers go to the recruiter and that the person running the process will be in touch within a stated timeframe.
The recruiter does not receive a recording to listen through, but a ready summary: question, answer, and a quote where the candidate said something relevant outside the script. The summary can go into an ATS, a spreadsheet, the recruitment team's inbox, or all of those at once. Reviewing the shortlist stops being guesswork: you can see who is available immediately, who holds the required licence and who was ruled out on the very first condition.
The effect is that a human only calls the people worth talking to. The rest of the process stays exactly as it was – the interview, the meeting, the decision. The only thing that changes is who makes the first twenty phone calls.
The candidate did not pick up – this is where SMS starts
A missed call is the normal state of affairs in recruitment. People are at work, on a tram, on a building site, or they simply do not answer unknown numbers. Automatic redialling on its own achieves little, because the candidate still has no idea who was calling.
So immediately after a missed attempt the system sends a message that explains the situation and gives the candidate a choice. Sample wording:
“Hello, this is XYZ. We called to ask a few questions about your CV. Please call us back or let us know what time we can reach you.”
That single message turns a missed call into an open case. The candidate knows who called and why, so they respond far more often. The SMS layer is built on the same SMS API that handles confirmations and reminders in other processes – messages go out under the company name, and candidates' replies come back into the system.
When the candidate calls back
If the candidate calls back, they do not reach a void or a touch-tone menu. The system recognises the incoming number, matches it to the application in a specific recruitment process, and the bot carries the conversation on: it introduces itself, reminds the candidate what the company was calling about, and asks the same set of questions. If the previous call broke off halfway through, it picks up where it stopped instead of starting over.
Booking by SMS and a punctuality no human can keep
The most common reply to such a message is neither “yes” nor “no” – it is a time. “After 5 pm, please.” “I am free tomorrow morning.” “Not before Saturday.” For a recruiter that is one more thing to remember and one more calendar entry that will slip anyway, because the previous interview ran long.
The bot does not have that problem. If the candidate writes “after 5 pm, please”, the phone rings at 5:01. Not earlier, so it does not catch them still at work, and not three hours later, when the whole thing has long since slipped their mind. It sounds like a detail, but in high-volume recruitment this is exactly the stage where a candidate is easiest to lose: not because the offer does not interest them, but because nobody called back when they said they would.
The same logic works in the other direction. The bot can propose a specific time window, confirm it by SMS and send a reminder fifteen minutes before the call. All of this traffic – the outbound call, the missed attempt, the SMS, the reply, the second call – happens on one company number handled by a virtual PBX, so throughout the process the candidate sees the same number and knows who is calling.
Where the line runs: the bot collects, the human decides
This is the most important part of the whole deployment and it is worth stating plainly. A pre-screening voicebot collects answers. It does not assess the candidate, does not score them, does not reject applications and does not decide who gets invited to an interview. The recruiter receives an orderly set of facts and makes the call themselves – exactly as they did when they were dialling those numbers in person.
There are two reasons for drawing the line there. The first is practical: an automated system will not catch the context that often decides a hiring conversation, and a single answer rarely settles whether someone fits. The second is formal and ethical: recruitment decisions concern people and should involve a human being, and the candidate has a right to know that there is a machine on the other end. That is why the bot introduces itself at the start, says outright that it is an automated system, and lets the candidate end the call or ask to speak to a recruiter. Someone who does not want to talk to a machine goes onto the list for personal contact, not into the bin.
What you need to prepare before launch
Contrary to appearances, most of the work is not on the technology side but on the content side. Before such a process can start, you need:
- a list of questions for one vacancy – short, specific, with clearly defined answers you expect to hear,
- a call script – the introduction, the notice that this is an automated call, the order of questions and a closing line announcing the next step,
- the SMS wording – after a missed call, after a time has been agreed and as a reminder,
- consent to be contacted and an information notice – make sure the candidate knows they will be contacted by phone and that the first call may be handled automatically,
- somewhere for the answers to land – an ATS, a spreadsheet or the team inbox that receives the summaries,
- timing rules – the hours when calling is allowed, the number of attempts and the moment a case is handed over to a person.
Bear in mind that consent and information duties around candidate data and automated calls differ from country to country, so the wording is worth agreeing with whoever handles data protection at your company. On the technical side, the deployment consists of the number the bot calls from, the call script, the SMS layer and an integration with the place where results are stored. At ACTIO this is built on AI solutions in communications connected to telephony, so the company does not have to build its own voice infrastructure or buy an SMS gateway separately.
Who this makes sense for
Voicebot pre-screening pays off where applications are numerous and the qualifying questions repeat in every single call. In practice that means production, warehouse, logistics, hospitality and retail hiring, staffing agencies running a dozen processes at once, and companies that scale headcount seasonally. The more repetitive the question set and the more candidates per ad, the bigger the difference.
It makes far less sense in expert and managerial recruitment, where there are only a few applications and the first phone call is really a conversation about the role, the team and motivation. There an automated system saves no time, and it may damage the impression the company makes on the candidate.
Summary
A voicebot in recruitment does not replace the recruiter. It removes the first, most mechanical layer of the work: getting through to the candidate, asking the few questions the CV did not answer, and arranging contact at a time that suits them. It does this punctually, forgets nobody and does not give up after the third missed attempt. The invitation decision, the assessment of the candidate and the interview itself stay with the human – and that is how it should stay. If you want to see what such a process could look like in your company, get in touch and describe the one vacancy you recruit for most often.