Applicant Tracking Software: Four Products Sharing a Name

Search for applicant tracking software and you get a category that behaves as though it were one product. It is not. Underneath the shared name are at least four different kinds of system, built for different companies, solving problems that overlap only partially. Buying the wrong kind is the most expensive mistake available here, and it is far more common than buying a bad product.

This is a map of the category rather than a ranking. What the four kinds are, who each one fits, the questions that reveal which you are looking at, and where the whole category tends to disappoint regardless of which you choose.

What is applicant tracking software, in one paragraph?

Applicant tracking software collects applications in one place, records each candidate's stage, holds the notes and scores behind decisions, and handles the messages and scheduling around them. What it is for differs by company: enforcing process across many recruiters, or stopping a small team from losing people between the cracks.

The four kinds

1. The compliance-first system

Built for companies where hiring is audited — regulated industries, government contractors, large employers with reporting obligations. The defining features are an immutable record of who saw what and when, structured demographic reporting, approval chains for requisitions and offers, and retention rules.

These systems are unpopular with the people who use them daily, and that unpopularity is largely by design: the friction exists to make the record complete. If you have genuine compliance obligations, this friction is the product and you should not shop for a system that removes it. If you do not, buying here means paying for constraints you will spend two years working around.

2. The recruiter-productivity system

Built for people whose full-time job is hiring — in-house talent teams and agencies. The centre of gravity is throughput: sourcing, pipelines with many candidates in motion, bulk actions, templated outreach at volume, and reporting on recruiter activity.

The tell is that the interface assumes you live in it all day. Keyboard shortcuts, dense lists, many statuses. That is exactly right for someone with forty roles open and exactly wrong for a hiring manager who visits twice a week and has to relearn the layout each time.

3. The lightweight tracker

Built for small teams that hire occasionally. Few stages, quick setup, a hosted careers page, simple email templates. It does what a good spreadsheet does, plus reliability and shared visibility.

Its honest limitation is that it stores and organises without assessing. The reading of two hundred applications remains yours, in full. For a team receiving twenty applications per role that is fine and the tool earns its place on coordination alone. For a team receiving two hundred, the largest cost is untouched.

4. The assessment-first system

Built around the reading problem. Criteria come from the job description, applications are scored against them, and the output is an ordered pile with reasons attached rather than a chronological list.

Its risk is the mirror image of the lightweight tracker's: a system that produces confident scores invites you to trust them, and the quality of a score depends entirely on whether the assessment is genuinely reading the document or matching keywords. The questions in the next section are mostly aimed at telling those two apart.

Which kind are you looking at?

Vendors do not describe themselves this way, so four questions do the work.

Question	Compliance	Recruiter	Lightweight	Assessment
What happens to a CV after upload?	Stored, logged	Stored, parsed into fields	Stored	Scored against criteria
Who is the daily user?	HR and audit	A full-time recruiter	A manager, twice a week	A manager, at intake
Can a stage be skipped?	No, by design	Yes	Yes	Yes
What does the demo spend most time on?	Records and reporting	Sourcing and pipeline speed	Setup speed	What the score means

The second row is the most predictive. Software is shaped by who it expects to see every day, and a tool built for a full-time recruiter will feel like a cockpit to a founder who opens it on Tuesdays.

The gap that runs through the whole category

There is one weakness shared by most applicant tracking software, and naming it makes the rest of the shopping easier: these systems are built to move candidates between states, and they are largely uninvolved in deciding whether a candidate should move.

The word "screening" hides this. In most products, screening means knockout questions on the application form — right to work, years of experience, salary expectation. Those filter people who answer honestly. They do not read a CV.

The next tier up is keyword matching, and it is worse than no filter in one specific respect: it is confidently wrong. A candidate who describes the same work in different vocabulary scores low. A candidate who has learned to mirror the advert scores high. You end up with a shortlist selected for phrasing, and because a number was produced, the number gets believed.

So the question that separates products is narrow and answerable: what happens between the file arriving and the candidate appearing in a list? If a vendor cannot describe that in a sentence, the answer is nothing, and the five hours of reading remain yours.

What a real assessment step contains

Three components, and all three have to be present together or the result is decoration.

Criteria that existed before the applications did. Criteria written after reading forty CVs are criteria shaped by who applied. Derived from the role, before intake, they are the only defence against that drift — and they are what makes a rejection letter possible to write honestly.

An outcome that admits uncertainty. A binary yes or no on a document is a lie about how much a document tells you. Most applications sit in a middle band: not obviously right, not obviously wrong, worth twenty minutes if the top of the pile thins. A system forcing those into a binary either wastes interview slots or discards good people quietly.

A reason attached to every outcome. A score with no explanation cannot be checked, and what cannot be checked will eventually be wrong without anyone noticing. The reason is also the sentence you paste into a reply so that it says something true.

Add a fourth that almost nobody has: an explicit state for documents that could not be read at all. Scanned CVs, locked PDFs, files that are really photographs, CVs laid out entirely inside a graphic — these arrive in every real intake. A system with nowhere to put them puts them in the rejection pile silently, and a candidate is lost to a file format rather than to a decision.

How Orova Recruit is shaped

Orova Recruit is the hiring module inside Orova, and it sits in the fourth kind: assessment-first. Stating its shape precisely is more useful than a recommendation.

Positions carry their own criteria, derived from the job description. The description can be drafted inside the module and then analysed, which mostly serves to expose requirements stated so vaguely that no candidate could be scored against them. Change the description and the criteria are re-derived rather than left stale.

Applications are uploaded in bulk and assessed against those criteria. A position holds up to 500 CVs, with a ceiling of 10 MB per file. Each one ends in one of six states: pending, processing, then meets the criteria, worth considering, does not meet, or could not be read.

Interview questions are generated from the same criteria and can be refined, then applied across a whole position so every candidate faces the same core set — the cheapest fairness improvement available to a small team, and one that costs nothing to adopt.

Interviews can be recorded and transcribed. Notes can be typed, dictated, or produced from an uploaded audio file. An interviewer who is not typing is an interviewer who is listening, and the evidence for a decision becomes specific enough to quote rather than characterise.

Review is where the decision is made and written down. Candidates are scored, a recommendation is produced, decisions are recorded individually or in bulk, and the outcome exports to PDF for people who will never log in.

Scheduling covers rooms and declared slots, and the contact area holds message templates — including AI-drafted ones and image attachments — with rules for automatic sending, so replying is the default rather than an act of virtue during a busy week.

What it does not do

It does not post vacancies to job boards, and it does not syndicate to LinkedIn, Indeed or any aggregator. If your problem is that not enough people apply, this is the wrong purchase and no amount of screening quality will fix it — distribution and selection are different problems.

It has no requisition approval chains, no structured demographic reporting, and no retention policy engine. If you have genuine audit obligations, buy from the first kind. It is also not a human resources information system: no payroll, no employee records after the hire, no performance reviews. It ends at the offer.

What it costs beyond the subscription

Three costs are routinely omitted from the comparison, and together they usually exceed the licence.

Setup that never finishes. Configuration is presented as a one-off, and in companies without a talent operations function it is recurring: stages that need adjusting, templates that go out of date, criteria that stop matching the role. Any system requiring ongoing maintenance to stay useful will drift within two quarters and then be quietly abandoned. Ask what the system does with no maintenance at all for six months.

The seat problem. If every interviewer needs a paid account, the effective price for a small team is several times the headline. Some products solve it with limited-access roles; some do not. This single question moves the annual figure more than any negotiation on rate.

Getting out. Export during the trial rather than at the end of the contract. Check that the export contains the notes and the scores, not only names and stages. Products differ sharply and the difference is invisible until the day you want to leave — at which point the part that took you the most work to create is the part being held.

A selection process that fits in two weeks

Most evaluations go wrong by starting with a longlist of products instead of a description of the problem.

Days one and two: write the problem down. Two sentences. "We receive around X applications per role and nobody has time to read them" is a different problem from "candidates fall through the gaps between us" and points at a different kind of system. If the team cannot agree on which sentence is true, stop here; that disagreement will otherwise be resolved by whichever vendor demos best.

Days three to five: shortlist by kind, not by brand. Decide which of the four kinds you need and take two candidates from that kind. Reviewing six products from three different kinds produces a comparison table where the columns are not comparable.

Days six to nine: run both against real data. Thirty real CVs from a role you already hired for, including the messy files, and whose outcome you know. Compare each system's view of that pile against hindsight. This single exercise is more informative than every feature list, and almost nobody does it because it feels like work on a solved problem.

Days ten to twelve: test the awkward paths. A three-person panel. A candidate who reschedules. A file that will not parse. A rejection sent to a real address so you can read what arrives.

Days thirteen and fourteen: export, and ask the interviewers. Not the person who ran the evaluation — they are already invested. The people whose week it will change.

Six mistakes that outlast the purchase

Configuring for the hiring year you wish you had

The demo shows six stages, four roles and forty candidates in motion. Your next twelve months is two roles, sequentially, with eleven serious candidates between them. Six stages configured for that produces four permanently empty columns and a process everyone routes around.

Start with three: applied, interviewing, decided. Add a stage only after you have felt its absence twice. A process that grows from friction fits the company; one installed from a template fits nobody, and the template was written for a company ten times your size because that is who buys the enterprise tier.

Treating the ranking as the decision

Any system that assesses applications produces something that looks like an ordering, and an ordering invites a line. Interview everything above it, discard everything below. That is not what the assessment is for.

Its job is to change the order in which you read, so attention lands where it is most likely to pay and the eightieth application gets the same quality of reading as the eighth. The candidate ranked twenty-second is not worse than the one ranked eighth; they are less legible to an automated reading of a document both of them wrote under different assumptions about what a CV is for. Keep reading past the line.

Skipping criteria because the role is obvious

Every role is obvious to whoever wrote the description and to nobody else. The test takes five minutes: ask two panel members to write down separately the three things that would make them say yes. The lists differ almost every time, and that difference is the entire value of the criteria step — not overhead but the argument you were going to have anyway, held cheaply in week one instead of expensively in week seven with two finalists waiting.

Automating the rejection without reading it

Automatic replies fix silence and create a new problem. A template that is fine for a first-stage decline reads as insulting to a senior candidate who spent six hours on an exercise. Scope the automation by stage: automatic where volume is high and the message is genuinely generic, written by a person for anyone who has met a person.

Importing years of history

"In case we want to revisit them." Almost nobody revisits them, and the import fills a clean system with stale records that nobody trusts, which then contaminates every search. Carry one thing across: the list of people you have already told no. Losing that is how someone gets rejected twice for the same role, which reads to the candidate as total indifference.

Running two sources of truth

The old spreadsheet stays alive "just for this role" and then permanently. Two sources are worse than one bad source, because every question now has two answers and no way to tell which is current. People check both, which is slower, so they stop checking and ask in chat instead. Pick a date; from that date, new applications go into the system and nowhere else.

What a good implementation looks like after three months

Four signs, none of which are about features.

The pipeline is true on a Friday without anyone tidying it. Not because people are diligent, but because updating it is the easiest way to do the work rather than an extra step afterwards. If keeping the system current requires discipline, it will be current for six weeks.

Rejections go out without anyone being brave. The rule sends them, the wording is decent, and nobody has to psych themselves up on a Friday afternoon. This is the change candidates notice and the one that quietly improves your reputation in a local market.

A decision from six weeks ago can be reconstructed. Open a closed role and answer why candidate B was preferred to candidate D. If the answer is a shrug, the system recorded stages rather than reasoning, and you will discover this the day a chosen candidate declines and you need to reopen a shortlist.

Nobody talks about it. Software that works becomes invisible. Continued discussion — about who should update what, about whether the stages make sense — means the process is still being negotiated, and the tool has become a venue for that negotiation rather than a resolution of it.

Where the spreadsheet actually wins

It is worth being fair to the incumbent, because a spreadsheet beats applicant tracking software on three counts and understanding them prevents a purchase that gets resented.

It is infinitely flexible. A column can be added in four seconds to capture something nobody anticipated, and no vendor roadmap is involved. Any system with a schema is less flexible than a grid, and teams with genuinely unusual processes feel that constraint immediately.

Everyone already knows how to use it. There is no adoption curve, no training, and no colleague quietly avoiding it. For a process that runs twice a year, familiarity is worth more than capability.

And it costs nothing, which matters more than it should because it means nobody has to justify it. A subscription creates an obligation to demonstrate value, and that obligation sometimes produces worse decisions — keeping a tool because it is paid for, or over-configuring it to feel like the money was used.

What the spreadsheet loses on is exactly the three things this article has been about: it does not read applications, it does not remember why a decision was made, and it does not send anything. Those are the costs that grow with volume. Below the volume where they hurt, the spreadsheet is the correct answer, and saying so is more useful than pretending otherwise.

When the honest answer is not to buy

Three situations, and they cover more teams than the category likes to admit.

Fewer than four hires a year, under thirty applications each. A folder, a spreadsheet with an owner column, and a Friday reminder will carry you. The overhead of learning a system exceeds what it saves at that volume, and it will be unfamiliar again by the time you next open it.

Not enough applications. No system improves a pile of six. The fix is distribution, referrals, and an advert that describes the work honestly instead of listing requirements copied from a competitor. Buying screening software here converts a hiring problem into a hiring problem plus a subscription.

No agreement on what you are hiring for. If the panel cannot state three things that would make them say yes, no software resolves that; it records the disagreement more neatly. Have the argument first. It costs an afternoon and saves a quarter.

A note on the word "tracking"

The name has aged badly and it shapes expectations in an unhelpful direction. "Tracking" describes the least valuable thing these systems do — recording which stage someone is at — and it was accurate when the alternative was a filing cabinet. It is now the part any spreadsheet handles adequately.

The valuable functions are assessment, evidence, and communication: helping you decide who to talk to, keeping the reasoning behind decisions, and making sure people hear back. None of those are tracking. A buyer who takes the category name literally will evaluate products on their pipeline views and pick the one with the prettiest board, which is roughly the least consequential comparison available.

Worth keeping in mind when reading vendor sites too. The products that describe themselves primarily as trackers are usually being accurate about their scope, and that is useful information rather than a weakness — a tool that is honestly a tracker, priced as a tracker, is a perfectly reasonable purchase for a team whose problem is coordination.

The one metric worth tracking

Small teams are told to measure time-to-hire, cost-per-hire, source effectiveness and offer acceptance. At two hires a year those are averages over a sample of two, and decisions get made with them anyway.

Track one thing: the median gap between an application arriving and the candidate hearing something true from a human being. Median rather than mean, so a single forgotten candidate does not distort it.

It correlates with nearly everything else. Short, and strong candidates stay engaged and acceptance rates rise. Past a week, and you lose exactly the people who had other options. It also degrades before anything else visibly breaks, which makes it the earliest warning that the process has stopped running — long before anyone notices a role has been open for four months.

If applicant tracking software does one thing for you, it should be to keep that number small. That is a better buying criterion than any feature comparison, this one included.

How the four kinds behave as you grow

A purchase that fits today can become the constraint in eighteen months, and the direction of failure is different for each kind. Worth knowing before you commit.

The lightweight tracker fails on volume. It keeps working perfectly and the reading problem grows underneath it. The failure is not a crash but a creeping realisation that the tool is organising a workload nobody can get through. Teams usually notice when a role attracts three hundred applications and the shortlist takes two weeks to produce.

The assessment-first system fails on process complexity. It handles intake well and then a company grows into requisition approvals, headcount planning, and multiple recruiters coordinating. Those are the compliance-first system's problems, and an assessment tool has no answer to them beyond conventions people agree to follow.

The recruiter-productivity system fails on adoption. It works while there is someone whose job is hiring. If that person leaves and the work distributes back to managers, the tool becomes too heavy for anyone to open twice a week, and the process reverts to email while the subscription continues.

The compliance-first system rarely fails; it just costs. It keeps working and the friction keeps being paid, in the form of managers who avoid it and shadow spreadsheets that appear alongside it. The failure mode is a complete official record of a process that partly happened somewhere else.

Two practical conclusions. First, buy for the volume and structure you will have in a year, not in three — three-year predictions in hiring are fiction and you will pay for capability you never use. Second, check the export before you commit, because the most likely future is that you outgrow this in one specific direction and the migration is easier if the notes and scores come with you.

The questions nobody asks and should

Beyond features, six questions that change how a system feels to live with.

What does a hiring manager see when they log in after three weeks away? The most common daily user of a small company's system is someone who forgot how it works. A landing view that says "these four people are waiting on you" is a different product from one that opens on a dashboard of aggregate metrics.

How many clicks from an application to a decision recorded? Count them in the demo. Anything above about six means the fast path is not designed, and the fast path is what gets used on a busy Thursday.

What happens to a candidate nobody touches for two weeks? Silence, or a nudge? Systems that surface stalled candidates prevent the single most common failure in small-team hiring, and most do not.

Can two interviewers disagree on the record? Averaged scores hide disagreement, and disagreement is the most useful signal a panel produces. If the system collapses two views into one number, it is destroying information to produce tidiness.

Who can see salary expectations and notes? Small companies share everything by default, which is fine until it is not — a candidate is someone's friend, or a colleague's manager. Ask about field-level visibility before you need it.

What does the candidate-facing side look like on a phone? Most applications arrive on phones. Open the application form on your own phone during the demo, not on the vendor's laptop.

A note on fairness that is also practical

Structured hiring is usually presented as an ethical improvement, and it is, but the practical argument is stronger and gets less airtime.

Criteria written before intake, the same questions asked of every candidate, and scores recorded with reasons — these produce better hires as a side effect of producing defensible ones. The mechanism is simple: unstructured interviews measure how comfortable a candidate is in an unstructured conversation with you specifically, which correlates with background rather than capability. Structure narrows what is being measured to something closer to the job.

Two habits capture most of the benefit and neither requires software. Ask every candidate for a role the same core questions, in the same order, and write the score down before hearing anyone else's. Discussion before independent scoring converts a panel into one opinion held by the most confident person in the room.

Software helps by making these the default rather than an act of discipline, and by keeping the record when nobody has time to keep it. What it cannot do is supply the willingness to be constrained by your own criteria when a charming candidate scores badly — that remains a human choice, and it is the moment where every structured process is actually tested.

Where the category is heading, and what that means for a purchase now

Two shifts are visible across products, and both affect how long a decision made today will hold.

The assessment step is moving from an add-on to the centre. Five years ago, an applicant tracking system was a database with a workflow and screening meant form questions. Products increasingly compete on what happens to the document, which is a healthy direction because it addresses the largest cost. It also means the differentiator is now something hard to evaluate from a feature list — you cannot tell the quality of a reading from a screenshot, only from running your own applications through it.

The line between tracking and the rest of hiring is blurring. Interview notes, transcription, structured scoring and scheduling used to be separate purchases. Bundling them is mostly good, because the handoffs between separate tools are where candidates fall through. It also raises the cost of a mistake: replacing a bundled system means replacing five habits rather than one.

Two practical implications for a decision you are making now. First, weight the export question more heavily than you would for a single-purpose tool, because the switching cost is higher when more of the process lives inside. Second, be sceptical of long contracts in a category that is changing this quickly — a two-year commitment locks in a product's current answer to a question the whole category is still working on.

None of which is an argument for waiting. The teams who lose most are not the ones who bought the wrong system; they are the ones who spent a year evaluating while candidates fell through a spreadsheet. A tool chosen reasonably and adopted properly beats a better tool chosen eventually, and the difference between the top few products in any kind is smaller than the difference between adopting something and not.

Migrating without losing the thread

The migration is where a surprising share of first deployments fail, and the failure is always the same shape: the old process keeps running in parallel because the new one is not quite ready, and parallel becomes permanent.

Pick a date. From that date, every new application enters the system and nowhere else. Candidates already in flight finish where they started, the old sheet is marked closed when the last one resolves, and then it becomes read-only. This feels untidy — for a fortnight there are two places to look — and it is the only version that terminates. Attempts to migrate everything first produce a system full of records nobody trusts and a launch date that keeps moving.

Three artefacts are worth writing down on the first day, none of which live inside the software and all of which determine whether it helps.

Who owns the pipeline being true. One name, not a team. Fifteen minutes on a Friday.

What each stage means, in one sentence each. Particularly the difference between "reviewed" and "decided", which is the ambiguity that loses the most candidates.

The reply commitment. How fast, from which stage, written by whom. If nobody will commit to a number, say that out loud rather than implying a different one in the careers page copy.

Those three take twenty minutes and outperform every configuration decision you will make inside the product.

The part people underestimate

The hardest part of hiring well at a small company was never the software. It is the discipline to decide what you are looking for before you look, and the nerve to reply honestly to people you will not hire. Applicant tracking software makes both easier to sustain through a busy quarter — that is a real and underrated benefit, because busy quarters are exactly when good candidates get lost.

What it cannot do is supply either quality on your behalf. Any vendor implying otherwise is selling a story about your hiring that will not survive its first contact with a real shortlist, and the disappointment that follows usually gets attributed to the product rather than to the story.

The short version

Applicant tracking software is four different products sharing a name: compliance-first, recruiter-productivity, lightweight tracker, and assessment-first. Most bad purchases are a mismatch of kind rather than of quality, and the most predictive question is who the software expects to see every day.

The gap running through the category is that storing is not screening. Ask what happens between the file arriving and the candidate appearing in a list, and treat a vague answer as "nothing". A real assessment step needs criteria written before intake, an outcome that admits uncertainty, a reason attached, and somewhere to put files that could not be read at all.

Configure less than you think, keep reading past the line the ranking draws, never run two sources of truth, and measure the median hours from application to a true human reply. If you hire four times a year and get thirty applications each, the honest answer is a folder and a Friday reminder.

If you remember one question from all of this, make it the narrow one: what happens between the file arriving and the candidate appearing in a list? It separates the four kinds faster than any feature comparison, it exposes the difference between storing and screening, and a vendor who answers it precisely is describing a product that has met real intake. A vendor who answers it with adjectives is describing a database.

Related reading: what an applicant tracking system actually does, a job description template that doubles as your scorecard, and what actually works with AI in recruitment. The product is at orova.vn.