TL;DR
- The core decision: whether the document you publish is the one written to get the role approved, or one written to make a working person apply.
- When doing nothing is right: when your postings attract people who can do the job and you're choosing between good candidates rather than hoping.
- What has to be true: somebody can describe an ordinary week in the role, in specifics, before a word of the advert is written.
- How the options split: by what the document leads with, responsibilities, a week of work, outcomes, a problem, or evidence.
- Decision rule: every requirement you list is a filter. If you can't say who it removes and why that's correct, delete it.
- Outcome to expect: fewer applications, more of them from people who could actually do the work.
Two Documents Wearing One Name
A hiring manager writes a job description to get a role approved. It has to survive a levelling conversation and a budget review, so it's written in the language those conversations use: scope, reporting lines, bands, a list of responsibilities long enough to justify the level. It does its job. The role is approved.
Then that same document gets posted as the advert, because it exists and writing another one is work nobody has budgeted for. Now it has a completely different reader: somebody currently employed, reasonably content, deciding on a Tuesday evening whether to spend an hour applying. The document was never written for that person, and it shows. It describes a position in an org chart rather than work somebody might want to do.
The real problem isn't that job adverts are badly written. It's that most of them are the wrong document entirely, repurposed because it was there. Everything else follows from that: the requirement lists that filter out the people you wanted, the responsibilities that describe a level rather than a job, the postings that draw plenty of applications and almost nobody who can do the work.
Best tools for Recruitment & Hiring
When You Genuinely Do Not Need to Act Yet
Your current setup is genuinely fine. Your postings bring in people who can do the job, and your problem is choosing between them rather than finding them. Whatever you're writing works. Rewriting it because a template says you should is effort spent on a thing that isn't broken.
Friction is starting to show. You've noticed applications arriving from people who match the words and not the work, or a hiring manager has said the pool was disappointing without being able to say why. That's the first sign the document is describing the wrong thing. The cheapest check is to read your last posting and ask whether it tells anyone what a Tuesday in the role looks like.
It has become a real cost. Screening time is now significant, and most of it goes to rejecting people the advert should never have attracted. Or the opposite, which is harder to see: a role sits open with few applicants, and everyone concludes the market is tight without checking whether the posting is deterring people who could do it. At this point the document is costing you either attention or candidates, and both are expensive.
The edge case that forces it. You're hiring for something new, where no previous posting exists to copy. This is where organisations discover they've never written a job advert from scratch, only edited inherited ones, and that nobody can describe the actual work in specifics. The same thing happens when a role has drifted, and the posting describes what the job was two years ago rather than what it has become.
Five Questions This Reader Asks at 11pm
Could someone tell what a Tuesday looks like from this? Read your posting and try to describe an ordinary day from it. If you can't, neither can a candidate, and they'll fill the gap with assumptions drawn from the last similar job they had. That's how you get people who accept and then discover the role isn't what they pictured.
Which of these requirements have I actually checked? Go down the list and mark the ones you'd genuinely test at interview. The unmarked ones aren't requirements, they're decoration, and each one is quietly deterring people. A list that nobody verifies is a list that filters on confidence rather than capability.
Am I describing the role or the level? Responsibilities written to justify a band read as generic because they are: they describe a category of position rather than this particular job. The test is whether the text would fit any similar role at any similar company. If it would, it isn't telling anyone anything.
Who reads this and decides not to apply? Every line has that effect on somebody. Some of that is intended and useful. The question is whether the people it deters are the ones you meant to deter, and for most requirement lists the honest answer is that nobody has ever asked.
Do I know what to say about pay? Whether you must publish a range differs by jurisdiction, and the rules have been moving. Find out what applies where you're hiring before you write, rather than after the posting is live, and take local advice rather than copying what a competitor does.
Three Honest Categories the Approaches Split Into
Position-led, describing the role's place in the organisation. Reporting lines, scope, a list of responsibilities, the requirements that define the level. It's right for the internal document, where the reader is deciding whether to approve a headcount and needs to see how the role fits the structure. It's also right where regulated or unionised environments require a formal specification. It fails as an advert because it answers a question candidates aren't asking. Nobody employed elsewhere reads a scope statement and feels compelled to apply, and the responsibilities list, written to justify a band, is the section candidates skim hardest.
Work-led, describing what the person actually does. A week in the role, the problems they'd own, who they'd work with, what a good first quarter looks like. It's right when you're competing for people who have options, because it's the only form that lets somebody picture themselves doing the job. It also self-selects usefully: people who read an honest description of the work and don't want it were never going to stay. It fails when nobody can describe the work concretely, which is more common than it sounds, and it fails when the description is aspirational rather than accurate, which produces early departures.
Evidence-led, describing what you need someone to have done. Specific capabilities with the evidence that would demonstrate them, rather than years and titles. It's right when the capability matters more than the background, and it opens the role to people whose route in was unusual. It gives your interviewers a clear brief, since the advert and the assessment then test the same things. It fails when it's written as a wish list, because a long evidence list is just a requirements list with better grammar, and it fails when the evidence you ask for is only obtainable at a certain kind of employer.
Five Diagnostic Questions You Can Self-Assess Against
Read your last posting aloud. Where does it get boring? Whatever section you want to skip is the section candidates skip. Usually it's a responsibilities list of near-identical bullets, each starting with a verb, none of which says anything specific. That section is doing no work, and it's usually the longest one.
Count the requirements, then count the ones you'd reject over. If the second number is much smaller, the list is deterring people for nothing. Move the unverified ones into a short "helpful but not required" note, or cut them. The list you'd actually reject over is usually three or four items.
Ask the person doing the job now what's missing. They'll name something the posting doesn't mention, and it's usually the part that consumes the most time: the coordination, the stakeholder work, the recurring meeting nobody enjoys. That omission is why the last hire was surprised, and it's the single highest-value addition available.
Compare the posting against the interview. If the advert asks for one thing and the interview tests another, the mismatch tells you which one is real. Usually the interview is real and the advert was inherited, which means candidates are being selected against criteria that were never published.
Look at who applied last time, and who didn't. If applications cluster in one background, ask whether that's the work or the wording. Requirements phrased as years and titles reproduce whoever already holds those titles, which narrows the pool in a way that has nothing to do with capability.
Six Job Description Structures, Reviewed
The responsibilities list
The traditional form: a bulleted set of duties, usually a dozen or more, each beginning with a verb. It earns its place as an internal document. Approvers can see the scope, levelling conversations have something concrete to weigh, and where a formal specification is required this is the form that satisfies it. It's also fast to produce, because most organisations have a comparable one to adapt.
Where it genuinely falls short is as an advert. A list written to justify a band describes a category rather than a job, so it reads as interchangeable with every similar posting, and candidates learn nothing that helps them decide. It also invites length: each bullet feels cheap to add, so the list grows until it describes a role no single person holds. The result attracts people who match the vocabulary and tells you nothing about who can do the work.
It's worth being precise about why this form persists, because the reason isn't laziness. The list is the artefact that survives the approval process, so it's the one that exists when somebody asks for the advert. Nobody is choosing it over a better document; it's simply the only document in the folder. That's also why editing it rarely helps. Trimming a responsibilities list produces a shorter responsibilities list, not a description of the work, and the reader still can't tell what they'd be doing. If you keep this form for the advert, the honest version is short: three or four duties that genuinely define the role, and the rest left in the internal file where it belongs.
The day-in-the-life narrative
A description of an ordinary week: what the person does, who they work with, what interrupts them. It earns its place by being the only structure that lets a candidate picture themselves in the role. It's also the most honest, and honesty here is a filter that works in your favour, since somebody who reads an accurate description and isn't interested has saved you both a process.
It falls short when nobody can write it, which is the usual obstacle. If the role is new, or the manager has never done it, the narrative becomes invention and the invention is optimistic. It also fails when the week described is the good week rather than the ordinary one. If the actual job is substantially incident response and recurring coordination, and the advert describes design and building, the person who accepts will discover the difference quickly and act on it.
The practical way to write one is to interview rather than compose. Sit with whoever does the job and ask what they did last week, in order, including the parts that were tedious. Ask what interrupted them, what they were waiting on, and which meeting they'd cancel. Then write from those notes. The result reads as specific because it is, and the specificity is what makes it credible to a reader who has been told by four other adverts that the role is fast-paced and impactful. Keep the ordinary parts in. A candidate who reads about the recurring coordination work and applies anyway is a candidate who won't be surprised in month two.
The outcomes-and-objectives format
The role defined by what it's accountable for achieving rather than what it does daily. It earns its place for senior and ambiguous roles where the method is genuinely the hire's to determine, and it attracts people who think in terms of results. It also gives you a natural structure for the first review conversation, because the advert already states what good looks like.
The limitation is that it works poorly for roles where the work is the point. Much valuable work isn't outcome-shaped: it's maintenance, coordination, or the steady application of judgement, none of which reduces cleanly to an objective. Written for those roles, the format produces vague targets that nobody can assess. It also tempts you to publish objectives nobody has agreed internally, which becomes visible at the first review.
There's a subtler failure worth watching for. Objectives written for an advert tend to be written optimistically, because they're doing marketing work as well as descriptive work. The person who accepts then holds you to them, reasonably, and discovers that the resourcing behind the objective was never agreed. So the test before publishing an objective is whether you'd be comfortable being held to it by the hire in their first month, with the support they'd need actually in place. If the answer is no, either fix the resourcing or write a smaller objective, because publishing the larger one only moves the disappointment later.
The problem-statement advert
The posting leads with the specific problem the role exists to solve. It earns its place by being genuinely interesting to the people you most want, because strong candidates are usually motivated more by an interesting problem than by a title. It also communicates a great deal quickly: the problem implies the scope, the constraints and the kind of person who'd enjoy it.
It falls short when the problem is dull or can't be described without disclosing something confidential, which is common. It also risks over-promising: a candidate drawn by a problem statement expects to work on that problem, and if the reality is that the role spends most of its time elsewhere, the mismatch surfaces early and expensively. Only use it where the stated problem really is the substance of the role.
It also asks more of the writer than the other forms. A problem stated vaguely reads as a slogan, and the reader discounts it immediately, so the problem has to be described at a level of detail that makes it recognisable to somebody who has worked on something similar. That usually means naming the constraint rather than the goal: not that you want to grow, but what specifically stops you growing at the pace you'd like. Where the constraint can't be described publicly, this form doesn't work, and forcing it produces exactly the slogan you were trying to avoid.
The skills-and-evidence structure
Each capability stated with the evidence that would demonstrate it, replacing years and titles. It earns its place by widening the pool without lowering the bar. Somebody who developed a capability in an unusual setting can recognise themselves in evidence-based wording where a years-and-titles list would have excluded them. It also aligns the advert with the interview, since the evidence you ask for is the evidence you test.
The limitation is discipline. An evidence list grows exactly as a requirements list does, and a long one is no better for being better phrased. It's also easy to specify evidence only obtainable at a certain size or type of employer, which reintroduces the narrowing you were trying to avoid. Keep it to the few capabilities that genuinely decide the hire, and be precise about what would count.
The discipline that makes it work is writing the evidence and the assessment at the same time. If you state that you need somebody who has done a particular kind of work, write down in the same sitting how you'd check it at interview. Where you can't describe the check, the capability is probably aspirational rather than required, and it belongs off the list. Doing this in one pass also fixes the mismatch problem, because the advert and the interview stop being written by different people at different times against different mental models of the role.
The minimal advert with a linked brief
A short posting covering the essentials, linking to a fuller document for anyone who wants detail. It earns its place by respecting how people actually read. The short version does the work of attracting attention; the brief serves the smaller number who are seriously considering it, and can carry the honest detail a posting has no room for.
It falls short when the brief doesn't exist or isn't maintained, which is the usual outcome after the first few roles. A link to a stale document is worse than no link. It also assumes the reader will click, and many won't, so anything genuinely disqualifying has to sit in the short version regardless. Treat the brief as an addition for serious readers, not as somewhere to put the parts you'd rather not lead with.
Where it works best is for roles you hire repeatedly, because the brief then earns its maintenance. A team that hires the same shape of role several times a year can keep one good document current and link every posting to it, which is a much better return than writing a long advert each time. For a role you'll fill once, the brief is usually overhead, and the effort is better spent making the short version accurate. The deciding question is whether anyone will own the document after the current role closes.
The Decision Table
| Situation | Scale | Setup | Primary Pain | Recommended Starting Point |
|---|---|---|---|---|
| Postings attract capable people | Any | Any | None | Leave it alone |
| Applications match words, not work | Any | Any | Screening load is wasted | Day-in-the-life narrative |
| Few applicants, market blamed | Under two hundred | Any | Requirements are deterring people | Cut the list to what you would reject over |
| Role is senior or ambiguous | Any | Any | Method is the hire's to choose | Outcomes and objectives |
| Competing against better-funded employers | Any | Any | Cannot win on package | Problem-statement advert |
| Pool is narrower than the work requires | Any | Any | Years and titles are filtering | Skills and evidence |
| Regulated or formal specification needed | Any | Any | Compliance form is mandatory | Responsibilities list internally, separate advert externally |
Most organisations sit in the last row without realising the two documents can differ. The formal specification can exist, be filed, and satisfy whoever needs it, while the advert is written for the person deciding whether to apply. Nothing requires them to be the same file.
Requirements That Filter the Wrong People Out
Every line in a requirements list is a filter, and most were added without anyone asking who it removes. The mechanism is simple and mostly invisible: a long list of firm-sounding requirements gives a reader many opportunities to disqualify themselves, and readers differ in how readily they take those opportunities. That variation has little to do with capability.
| Requirement type | Who it deters | Is that intended? | What to write instead |
|---|---|---|---|
| Years of experience | People who developed the capability faster, or in an unusual route | Rarely | The capability, and what evidence of it looks like |
| A specific job title | Anyone who did the work under a different name | Rarely | The work itself, described plainly |
| A named degree or certification | People qualified by practice rather than credential | Sometimes, and check whether it is genuinely required | "Required" only where it truly is, otherwise omit |
| A named tool or platform | People fluent in an equivalent | Rarely | The problem the tool solves |
| Industry background | Anyone solving the same problem elsewhere | Sometimes | The constraint that makes your context distinctive |
| A long list of desirables | Readers who take an unmet item as disqualifying | Almost never | Cut it, or label it clearly as not required |
The practical rule is to keep only what you'd genuinely reject over. For most roles that's three or four things. Everything else belongs either in a short line saying it's helpful and not required, or nowhere.
The resistance you'll meet is that cutting a requirement feels like lowering the bar. It isn't, and the distinction matters. The bar is what you reject over at interview, and that doesn't move when you delete a line from a document nobody was checking. What moves is who applies. A hiring manager who worries that a shorter list will bring weaker candidates is usually describing a screening problem, and the answer to a screening problem is a better assessment rather than a longer advert. The list can't do the assessment's job, and asking it to means it does neither well.
One more thing worth checking. Requirements phrased as years and titles reproduce the population that already holds those titles, which is a narrowing unrelated to whether somebody can do the work. Where requirements touch on protected characteristics, or could operate as a proxy for one, the rules differ considerably by jurisdiction. Have your standard wording reviewed locally rather than deciding case by case.
The Internal Version and the Advert Are Not the Same Document
Treating these as one file is the root cause of most of what's wrong with job adverts, and separating them costs less than people expect. The internal document already exists; the advert is a shorter piece of writing with a different job.
| Element | Belongs internally | Belongs in the advert | Must agree between them |
|---|---|---|---|
| Level and band | Yes | Only where a range is published | Yes, always |
| Reporting line | Yes | Usually, briefly | Yes |
| Full responsibility list | Yes | No | The substance, not the wording |
| A week of actual work | Optional | Yes | Yes |
| Requirements you would reject over | Yes | Yes | Yes, exactly |
| Desirables and nice-to-haves | Yes | Cut, or clearly labelled | No |
| The problem the role exists to solve | Optional | Yes | Yes |
| Working arrangement and location | Yes | Yes | Yes, precisely |
The final column is the one that causes trouble when ignored. Where the two documents disagree, candidates find out at interview or later, and the discovery reads as either disorganisation or bait. The most damaging version is the working arrangement: an advert implying flexibility that the internal document doesn't support produces an offer declined at the last moment, or an early departure.
Write the advert after the internal document is settled, not before, and check the two against each other on the rows marked yes. It takes a few minutes and prevents the whole class of problems where a candidate was told one thing by a document and another by a person.
What to Put in Writing
Adverts get written by whoever is free and inherited by whoever comes next. The reasoning behind each requirement disappears immediately, which is why lists only ever grow.
| Artefact | Who owns it | When it is written | What it prevents |
|---|---|---|---|
| The ordinary week, in specifics | Hiring manager, with the current post-holder | Before the advert is drafted | An advert describing a level rather than a job |
| Requirements you would reject over | Hiring manager | Before the advert is drafted | A list nobody can later prune |
| Why each requirement is there | Hiring manager | At the moment it is added | Inherited filters nobody can justify |
| What was published about pay, and on what advice | Talent lead | Before the posting goes live | Inconsistency across jurisdictions |
| Where the advert and the internal document differ | Recruiter | Before the posting goes live | Candidates discovering a contradiction later |
| Which requirements the interview actually tested | Hiring manager | After the first interviews | Publishing criteria nobody assesses |
The third row is the one that keeps lists short. If adding a requirement means writing down why it's there and who it removes, most additions don't survive the sentence.
Questions to Ask Before You Commit
On the week. Can you describe an ordinary Tuesday in this role, specifically? If nobody can, the advert will be generic whatever structure you use. A bad answer describes the role's purpose rather than its days.
On requirements. Which of these would you actually reject over? A bad answer is that they're all important, which means none has been examined.
On the omission. What does this role involve that the posting doesn't mention? Ask the current post-holder. A bad answer is that the list is comprehensive.
On pay. What must we publish where we're hiring, and who confirmed it? A bad answer is what a competitor does.
On the two documents. Where do the internal version and the advert differ, and is every difference deliberate? A bad answer is that they're the same document.
On the test. Will the interview assess what the advert asked for? A bad answer is that the interviewers know what matters.
What Getting This Wrong Costs
The visible cost is screening time, and it lands on whoever reads applications rather than in any budget. An advert that attracts the wrong people generates work in direct proportion to how well it's distributed, so improving your reach makes the problem worse. That's why organisations sometimes find that spending more on sourcing produces more applications and no more hires.
The second cost is the candidates you never see. A requirements list that deters capable people does so silently, and the absence is invisible: you can't review the applications you didn't receive. So the posting gets judged on the quality of who applied, which is exactly the population it selected, and the conclusion is always that the market is tight rather than that the document was.
The third cost arrives months later. An advert that describes an aspirational version of the job produces a hire who accepted one role and found another, and that mismatch is among the most reliable causes of an early departure. The cost isn't only the replacement. It's that the team absorbed a hiring process, an onboarding, and a resignation, and is now back where it started with less appetite for the next attempt.
There's a compounding version of this that's harder to notice. Each of these costs makes the next hire worse. A team that has been burned by a mismatch writes a more defensive advert next time, with a longer requirements list, which narrows the pool further and makes the following hire harder still. The document gets more cautious in exactly the situations where it needed to get more honest. Breaking that cycle usually means somebody sitting down with the last three postings and asking, for each requirement, who added it and what they were worried about. The answers are almost always about a previous bad hire rather than about this role.
So before you write: are you solving an attraction problem, a filtering problem, or an accuracy problem? Attraction means capable people aren't applying. Filtering means they are, alongside many who can't do the work. Accuracy means the right people apply, accept, and find the job isn't what the advert described. Three problems, three different documents, and only one of them is fixed by writing more.
When You Are Ready to Go Further
The work above needs no tooling and no budget. It needs somebody to describe an ordinary week in specifics, cut the requirements to what they'd genuinely reject over, and check the advert against the internal document before it goes live.
The next step, for a reader who wants to know whether their postings are unusual for the roles they hire, is comparison work. That's the part you can't judge from inside, because you only ever read your own.
HROpsLab publishes independent comparison work across HR tooling, applicant tracking and payroll. We sell nothing, we take no vendor money, and we publish no paid placements. If the next step is testing your assumptions against the wider market, our comparison work is one place to start.
Frequently Asked Questions
What should a job description contain?
For the internal document: the level, the reporting line, the scope, and the responsibilities that justify the band, since its reader is deciding whether to approve a headcount. For the advert, which is a different document with a different reader, the essentials are what the person will actually do in an ordinary week, the problem the role exists to solve, the few requirements you'd genuinely reject over, and the working arrangement stated precisely. The common error is publishing the first document as the second, which produces a posting that describes a position in an org chart rather than work somebody might want.
How long should a job advert be?
Short enough to be read by somebody deciding on a Tuesday evening, which in practice means the substance fits on one screen with detail available for anyone who wants it. Length isn't the real variable though. A short advert made of generic bullets communicates less than a longer one describing a real week. Judge it by whether a reader could picture an ordinary day and could tell whether they'd be rejected, rather than by word count. If a section bores you when you read it aloud, it's costing you attention rather than earning it.
Should I publish a salary range?
Whether you must differs by jurisdiction and the rules have been changing, so establish what applies where you're hiring and take local advice rather than following what a competitor does. Beyond compliance, publishing a range tends to reduce wasted process on both sides, because candidates whose expectations are incompatible self-select out before anyone spends an hour. If you publish one, make sure it's the range you'd actually offer within. A range that turns out to be aspirational at the top does more damage than not publishing at all.
What is the difference between a job description and a job advert?
The job description is an internal document written to get a role approved and to define it formally, with a reader who cares about scope, level and structure. The advert is external, written to make an employed person consider applying, with a reader who cares what the work is and whether they'd want it. Nothing requires them to be the same file, and treating them as one is the root of most bad postings. Keep the substance consistent, particularly on requirements, level and working arrangement, but write each for its own reader.
How many requirements should I list?
Only the ones you'd genuinely reject over, which for most roles is three or four. Everything else belongs in a clearly labelled note saying it's helpful rather than required, or nowhere at all. The reason to be strict is that each firm-sounding requirement gives a reader another opportunity to disqualify themselves, and readers vary in how readily they take those opportunities for reasons unconnected to capability. A quick test: go down the list and mark the ones you'd actually test at interview. The unmarked ones are decoration.
Should I list nice-to-haves at all?
Only if you label them unmistakably, and even then consider whether they earn their place. In practice a "desirable" section is read by many people as a longer requirements list, so it does filtering work you didn't intend. If something genuinely would help but isn't necessary, one short sentence saying so is safer than a bulleted section that looks identical to the requirements above it. The safest option for most roles is to cut them and let the interview discover what else somebody brings.
How do I write an advert for a role I do not understand technically?
Get the specifics from someone who does, then write it yourself. Ask the current post-holder or a practitioner to describe an ordinary week, the problems the role owns, and what separates somebody strong from somebody adequate. Your job is to turn that into something readable, not to invent it. What you should avoid is copying a posting from another company for a similar title, because you'll import their requirements without knowing which ones matter, and you'll have no way to tell whether the advert matches your actual role.
Who should sign off the advert?
The hiring manager owns the content, because they're accountable for whether it describes the job accurately. Whoever runs hiring should check it against the internal document for consistency on level, requirements and working arrangement, since that's where contradictions cause the most damage. Where pay is published or requirements touch on anything legally sensitive, that wording should be reviewed once and reused rather than decided per role. Keep the approval light. A posting that needs four sign-offs won't be rewritten when it stops working.
Write the advert for the person deciding whether to apply, not for the person approving the headcount.
