What Is an Applicant Tracking System? How ATS Software Works and How to Choose One

TL;DR

• An applicant tracking system (ATS) is a recruitment system of record, collecting applications, turning CVs into searchable records, moving candidates through the recruitment process, and generating reports on the hiring funnel.
• Nearly every large employer already runs one, so the real decision isn’t whether to buy an ATS but which type fits your hiring model: HCM suite, best-of-breed, small-business, public sector or agency.
• The big 2026 shift: ATSs now rank and filter candidates with AI, mostly from CV data. Before you switch it on, ask what the score is built from, who audited it and where a human steps in – under UK data-protection and equality law, that’s not optional.
• ATSs are only ever as good as the data fed into them. Adding structured interview evidence often improves hiring more than replacing the ATS.

The relationship between a recruiter and their applicant tracking system (ATS) is a lot like the relationship between a person and their phone. It’s a powerful tool that almost everyone has. Jobscan’s 2026 analysis found an ATS was powering the careers sites of 487 Fortune 500 companies.

But just as people grow dissatisfied with their phones, employers often become unhappy with what they’ve bought. In a 2025 webinar Aptitude Research ran with talent intelligence vendor Eightfold, only 28% of organisations are satisfied with their current ATS.

And the stakes just went up. ATSs no longer just store candidates – they rank them, with AI match scores, agents and even AI interviewers deciding who a hiring manager ever sees. Get that wrong and it’s not just bad hires: Workday is defending a court-certified age-discrimination collective action that now covers the HiredScore AI features in its recruiting platform.

Below, we’ll explain what ATSs do, the main applicant tracking system benefits, how AI is changing them and how to choose one, with UK rules and some CV myths explored along the way.

What is an applicant tracking system (ATS)?

An applicant tracking system (ATS) is multi-functional recruiting software that collects applications, turns CVs into structured records, moves candidates through the hiring stages, and reports on the hiring funnel. It’s used by recruiters, hiring managers, and wider hiring teams, bringing everyone onto one platform that acts as the system of record across the entire recruitment process.

What an ATS does: core functions

To qualify as an ATS, G2 requires software to (at a minimum): collect, sort and parse CVs, match candidates with roles, build talent pools, and handle communication. But beyond those essential features, most ATS recruiting software does a lot more, including:

• Posting jobs to your careers site and job boards
• Application capture
• A single database of searchable candidate records
• Pipeline stages and knockout questions
• Interview scheduling, video interviews and scorecards
• Collaboration and approvals with hiring managers
• Offers and the hand-off to onboarding
• Automated workflows for candidate emails and text messages
• Funnel reporting
• Integrations with your HR system, calendars, job boards and assessment tools

ATS vs HRIS vs recruitment CRM

By default, an ATS is a librarian, not a critic: it catalogues every applicant but has no opinion on who’s any good, recording judgements made elsewhere by recruiters, interviewers and assessments. Increasingly, though, vendors are teaching the librarian to review the books – using nothing but the blurb on the back cover. That blurb is the CV.

Who uses an applicant tracking system?

The classic ATS users are in-house talent acquisition (TA) teams and public-sector bodies. Recruitment agencies use them too, but agency ATSs are built around clients and placements owing to their unique positioning.

Adoption is near universal among big organisations but drops sharply across the wider economy. The Chartered Institute of Personnel and Development (CIPD) Resourcing and Talent Planning Report 2026 found that 32% of UK organisations use an ATS, up from 26% in 2024, with smaller firms dragging the average down: 53% of small and medium-sized enterprises (SMEs) used none of the recruitment technologies the CIPD asked about.

How an applicant tracking system works (from job ad to offer)

Most hires follow the same six steps, with the ATS handling the hiring process end to end.

1. Requisition and approval. A hiring manager requests a new role, and the ATS sends it for sign-off (whether by HR teams or finance).
2. Posting. Once approved, the ATS lets you post vacancies to your careers site and multiple job boards in one go.
3. Application capture and parsing. As applications come in, the ATS parses each CV into fields such as employer, job title, dates and skills.
4. Screening. Knockout questions filter job applicants based on requirements like right to work or a driving licence. Recruiters review what’s left over and, increasingly, AI ranks the pile with match scores.
5. Interviews. At the interview stage, the ATS books interviews, shares candidate details with the panel and helps collect feedback through scorecards.
6. Offer and hand-off. Offers, right-to-work checks and references are dealt with inside the ATS, with new employees’ records then passed to an HR system.

Applicant tracking systems for recruiters: what changes day to day

With an ATS, manual tracking is replaced by a central database: every application (successful or not) is stored with a full, searchable history. That means top talent who just missed out on a role can quickly be brought back into the fold once similar job openings come up, which is especially valuable when skills are in high demand.

Recruiters also save time on repetitive tasks like candidate communication and scheduling interviews, while reporting shows which recruitment efforts produce hires, where candidates drop out and how to reduce time to hire.

Of course, an ATS is only ever as useful as the candidate information fed into it. Typically that information is gathered from CVs.

How an ATS reads your CV (and the “75% of CVs rejected” myth)

It all starts with parsing. If a CV is a scanned PDF or an image, optical character recognition (OCR) pulls out the text first. The parser then sorts employers, job titles, dates and skills into searchable fields.

The simpler a CV’s layout, the easier it is for machines to read. Clean, single-column CVs are reliably processed, while image-based, over-designed versions can result in dropped fields.

Total rejection is rare, however, and certainly rarer than a figure often bandied around claiming applicant tracking system software rejects 75% of CVs before a human sees them. That figure traces back to around 2012 and Preptel, a CV-optimisation company with a vested interest, which closed the following year. For a more up-to-date (albeit small-sample) number, when CV builder Enhancv asked 25 US recruiters in late 2025 whether their ATS auto-rejects CVs for formatting or content, 23 said no.

A far bigger issue than layout is criteria. According to Harvard Business School and Accenture’s 2021 Hidden Workers report, 88% of employers using recruiting management systems agreed that those systems screen out highly skilled, qualified candidates because they don’t match the exact criteria in the job description.

Newer AI-powered ATSs can apply those criteria automatically, which risks filtering out qualified applicants at scale, so check exactly how yours works. Our guides to resume screening and removing bias by not using CV data will give you a clearer picture.

If you’re a candidate

• Use a single-column layout.
• Stick to standard headings such as Experience, Education and Skills.
• Favour text over images or graphics.
• Mirror the job description’s language where it truthfully describes your experience.

As self-reported records of the past, CVs are inherently compromised, even before they’re parsed to varying levels of success. So as AI takes on more of the ranking responsibility, the more important question becomes: what does an ATS’s AI build its match scores from?

What is an AI applicant tracking system?

An AI applicant tracking system is an ATS with machine learning, natural language processing or generative AI built into its workflow. Approaches vary widely. Take match scores. As of October 2026, Greenhouse Talent Matching sorts candidates into categories from Strong to Limited, Workday HiredScore grades them A–D, and the Workable Agent scores each candidate out of 100, with reasoning against each criterion.

How are AI match scores created?

When generating a match score, most ATS software simply compares a CV or profile with the job description. While that undoubtedly saves recruiters time when they screen candidates, it also means personal details like university, previous employers, and name are carried into the score.

This can lead to bias, and names are particularly problematic. In 2024, University of Washington researchers tested three open-source AI models of the kind used to match CVs to jobs. Across 554 CVs, 571 job descriptions and more than three million comparisons, they found that white-associated names were preferred in 85% of tests and Black-associated names in just 9%.

Bias doesn’t always flow the same way, however. A 2025 PNAS Nexus study scored around 361,000 synthetic CVs with five large language models and observed a different result. This time, the models generally favoured women over men, while most penalised Black men compared with otherwise identical white men.

The lesson for buyers is that bias changes with the model and set-up, so a vendor’s general assurances tell you very little. Ask them to test their AI on your own roles.

What good AI governance looks like

Responsible AI use keeps bias in check. Make sure:

• Blind screening is in place to reduce bias, with names and special-category data removed before scoring.
• Applications are scored in categories or with supplementary explanations, not a single magic number.
• Candidates are reviewed by a person before anyone is rejected. Our guide to human-in-the-loop AI in hiring covers how.
• The AI feature has a named, published, independent bias audit, with a date and whether results are role-level or pooled. Standards vary wildly between vendors, from monthly third-party audits to none published, so always ask before you buy.

Types of applicant tracking systems

There’s such a wealth of options when it comes to applicant tracking software that making a choice can be overwhelming. Thankfully, ATSs tend to fit in a few broad categories, each aimed at a different kind of employer.

A note on cost: check how prices are calculated before signing up. Small-business tools often publish monthly prices, enterprise ATSs are quote-based, and AI features are increasingly sold on credits.

Applicant tracking systems in the UK: what’s different

A few UK laws overlap with what an ATS does, so confirm any grey areas with your legal team before going live.

Automated decisions are allowed (with caveats). Changes under the Data (Use and Access) Act 2025 came into effect on 5 February 2026. There’s no longer a general prohibition on solely automated significant decisions, provided they don’t use special-category data. You must, however, tell candidates it’s happening, let them make representations, offer meaningful human intervention and let them contest the decision.. Increasingly common automated screening features like auto-shortlisting and auto-rejecting count as significant decisions, so these safeguards must be on.

The regulator is taking an interest. In its November 2024 report on AI recruitment tools, the Information Commissioner’s Office (ICO) made 296 recommendations and 42 advisory notes. In short, it wants candidates told how AI uses their data, assurances that only necessary information is collected and data protection impact assessments (DPIAs) are done early, and clarity on who is controller and who is processor. In March 2026, its draft automated decision-making guidance made it clear that human involvement must be “active rather than tokenistic”, but final guidance hadn’t been published at the time of writing.

The Equality Act 2010 covers key ATS features. Indirect discrimination (section 19) and reasonable adjustments (section 20) apply to both knockout questions and AI match scoring. Even if the breach is due to the behaviour of a vendor’s tool, liability still rests with the employer.

Right-to-work checks carry real penalties. Certified identity service providers can run digital checks for British and Irish citizens with a valid passport; everyone else needs a Home Office share code. Get it wrong and civil penalties reach £45,000 per illegal worker for a first breach and £60,000 for repeat breaches. Many UK ATSs build these checks in – make sure yours does.

Check data residency before you buy. Always ask where candidate data is hosted. That’s particularly true of public bodies, who need to know whether an ATS is available through the government’s cloud services buying framework G-Cloud.

UK firms that also hire in Europe should also bear in mind that the EU AI Act’s high-risk obligations for recruitment AI have been delayed to 2 December 2027.

How to choose an applicant tracking system

As you can see, there’s a lot of factors that come into play when choosing an ATS. If you’re feeling overwhelmed, don’t worry. Here’s the most important questions to ask alongside a selection process you can crib from.

What to ask

1. Does it fit your hiring model? Professional, high-volume, public-sector and agency hiring need different things from an ATS. The types table above will help you filter the options quickly.
2. Does it integrate with your stack? Check compatibility with your HR system, job boards and calendars, and whether assessment partners can input their scores back into the ATS.
3. Does it have strong AI governance? Ask how AI match scores are generated, what evidence the vendor has published, and whether there are independent audits and human oversight.
4. Is the candidate experience accessible? A strong candidate experience means mobile, short applications and easy adjustments.
5. Does it comply with UK laws and data rules? Check DPIA support, automated-decision safeguards, hosting location and right-to-work checks.
6. Does your recruitment team find it user friendly? Let the people who’ll use it most try it first.

The five-step selection process

1. Separate the ATS from how candidates are evaluated. Choosing an ATS and choosing how you assess people are different decisions within your recruitment strategy. Don’t just default to whatever the ATS bundles.
2. Integrations and UK compliance come first. Reject any tool that fails either, however good its features.
3. Question every vendor about its AI approach. What does the candidate produce? How is the match score built? What validity evidence is published? Is there an independent audit of the feature in question? What human-oversight controls exist?
4. Test score write-back in the demo. Ask the vendor to move a partner’s assessment score into a candidate record on the ATS platform, then sort by it.
5. Pilot on one role family. Track completion, candidate satisfaction, adverse impact at role level, time to hire, and early retention.

Why your ATS is only as good as the interview data going into it

Computing’s oldest rule is “garbage in, garbage out,” and recruitment has spent two decades trying to dodge it. An ATS can never be better than the CVs and profiles fed into it – self-reported summaries of the past that were never designed to be scored. Buying a shinier ATS to fix bad inputs is like buying a faster car because you’re lost.

Moving away from these inputs to a structured interview provided at the point of application helps employers and creates a more positive experience for candidates. Instead of scanning CVs for vague proxies for competence like past employers and job titles, which can invite unconscious bias, specific, job-relevant answers can become the new sorting metric. But getting that information into an ATS isn’t always easy.

Enter Sapia.ai: an untimed, text-based chat interview that candidates complete on their phone. Scores are built from the interview answers only, so nothing on the CV – name, school, previous employers – feeds the score. Sapia is not an ATS. It runs inside major ATSs like Workday, SAP SuccessFactors, SmartRecruiters, iCIMS, Greenhouse, Teamtailor, Tribepad, eArcu and others.

An independent disparate-impact audit by BLDS (late 2022, 72 tests, US and Canada models only) found no evidence of practically significant disparate impact. Sapia also has built-in detection of AI-generated answers. There’s field evidence too: LNER integrates Sapia with its eArcu ATS and reports ethnic-minority representation holding at 30% through every hiring stage.

Conclusion

For an ATS to be an effective system of record for recruitment, it must be chosen to fit your hiring model and HR stack. Those in the UK must pay special attention to the ramifications of switching on AI features, with safeguards in place for automatic filtering.

Then stop asking your ATS to do a job it was never built for. It’s a system of record, not a judge of talent. Feed it CVs and it ranks guesswork; feed it structured interview evidence and its rankings start to mean something. 

Book a demo with Sapia to see how you can get more out of your ATS.

FAQs

About Author

Barb Hyman

CEO & Founder