What an ATS actually does
Underneath the feature lists, an applicant tracking system does four things.
- Holds requisitions, so there is one authoritative statement of what is open, at what grade, approved by whom.
- Receives applications from wherever they originate and puts them in one place with their source recorded.
- Moves candidates through defined stages, keeping the history of who reviewed, who decided and when.
- Reports on all of the above, which is only possible because the first three exist.
Everything else, and there is usually a lot of everything else, is built on those four. A system that does them cleanly and little else is more useful than one with a long feature list and a messy record.
The reason the record is the point is that hiring is otherwise unmeasurable. Applications arriving in several inboxes, decisions taken in conversations and offers tracked in a spreadsheet cannot produce a reliable time to hire, a source comparison or an answer to why a candidate was rejected.
What it will not fix
An applicant tracking system enforces a process. It does not design one, and this is where most disappointing implementations come from.
If it is unclear who approves a requisition, how many interview stages there are, what each stage is for, or who makes the final decision, the system will faithfully reproduce that ambiguity with timestamps on it. Hiring managers will still not respond, panels will still duplicate questions, and offers will still be delayed by an approval nobody knew was required.
The order that works is to settle the process first, at least in outline, and configure to it. Organisations that configure first tend to build their existing confusion into a system and then blame the system.
The second thing it will not fix is a shortage of applicants. A tracking system manages flow; it does not create it. Where the problem is that nobody is applying, the answer lies in the role, the pay, the description and where it is advertised.
Keyword filtering, and its cost
Most systems can filter or rank applications automatically, and the feature is popular because application volumes are high and reviewing them is slow.
It is worth being clear-eyed about what filtering does. It rejects on the presence or absence of terms, which means it rejects candidates who describe the same experience in different words, who have non-linear careers, who worked in an industry with different vocabulary, or whose formatting the parser handled badly. Those rejections are invisible: nobody reviews what the filter removed.
| Filter type | Reasonable use | Where it causes damage |
|---|---|---|
| Hard requirements | A licence, certification or legal right to work that the role genuinely requires | Listing preferences as requirements, which removes candidates who could do the job |
| Keyword ranking | Ordering a large pile for human review | Auto-rejecting below a threshold, so nobody ever sees the tail |
| Experience thresholds | A rough sort where volume is unmanageable | Excluding career changers and returners, who rarely match on years |
The discipline that helps is to audit the filter against outcomes: sample what was rejected automatically and check whether any of them should have progressed. Filters set once and never reviewed drift into removing exactly the candidates the organisation says it wants more of.
The audit trail
The least discussed and most valuable part of an applicant tracking system is that it records who did what.
Where a hiring decision is questioned, whether by an unsuccessful candidate, a regulator or an internal review, the useful evidence is a contemporaneous record: the criteria applied, the assessment recorded against them, and the reason given at the time. A reconstruction assembled afterwards from memory and email is much weaker.
This matters more where hiring touches protected characteristics. A rejection recorded as a considered assessment against stated criteria is defensible. One recorded as a single word, or not recorded at all, is not, and the absence of a reason tends to be read unfavourably.
It is also what makes it possible to answer questions about the funnel honestly, including uncomfortable ones about where particular groups of candidates stop progressing.
Judging one
The questions worth asking when choosing a system are mostly not about features.
- What does the candidate see, and how many steps does it take them to apply? Application completion rates fall sharply with each additional screen.
- Can a hiring manager use it without training? If not, they will email you instead and the record will fragment again.
- Does it hand over to onboarding and the employee record, or does someone re-key the new joiner? Re-keying is where new-starter payroll errors originate.
- What can it report without an export? A system whose reporting requires a spreadsheet is a filing cabinet.
- What happens to candidate data, and for how long is it kept? This is a data protection question with a real answer, not a formality.
The last one deserves attention. Applicant data is personal data belonging largely to people who never became employees, retention periods are rarely configured deliberately, and an unreviewed system accumulates it indefinitely.
Frequently asked questions
What is an applicant tracking system?
Software that holds job requisitions, applications and candidates through defined hiring stages in one record, replacing inboxes and spreadsheets. Its core value is a single place where every application and decision lives, which is what makes hiring measurable and auditable.
Will an ATS fix a slow hiring process?
Only if the process is defined. The system enforces a process rather than designing one, so an unclear approval path or an undefined interview stage is reproduced faithfully with timestamps attached. Settle the process in outline first, then configure to it.
Do applicant tracking systems reject good candidates?
Keyword filters do, and invisibly, because nobody reviews what was screened out. Candidates who describe the same experience in different words, career changers and returners are the usual casualties. Audit the filter against outcomes rather than trusting it.
What should we look for when choosing one?
What the candidate experience looks like and how many steps an application takes, whether a hiring manager can use it untrained, whether it hands over to onboarding without re-keying, what it reports without an export, and how long candidate data is retained.
Why does the audit trail matter?
Because it is the evidence if a hiring decision is challenged. A contemporaneous record of the criteria applied and the reason given at the time is defensible; a reconstruction assembled later from memory and email is much weaker, and a missing reason tends to be read unfavourably.
How Engage tracks applicants
Engage keeps requisitions, applications, stage history and decisions in the same system as onboarding and the employee record, so a hire moves to a joiner without anyone re-keying the details that payroll then depends on. Source, stage timing and rejection reasons are reportable directly, which is what makes a funnel readable rather than reconstructed.
See recruitment software in Engage