> This is the markdown version of https://www.classet.ai/blog/how-to-write-a-job-posting
> Learn more at https://www.classet.ai



![Abstract low-poly illustration of a job posting with blank spaces](/_next/image?url=%2Fimages%2Fblog%2Fhow-to-write-a-job-posting.png&w=3840&q=75)

# How to Write a Job Posting Candidates Believe

Researchers shuffled the pay and perks on 6,000 job ads and asked workers what else they assumed about the job. People guessed. Here is how to write a job posting that answers the guess before they make it.

[![Paul Jones](/_next/image?url=https%3A%2F%2Fassets.basehub.com%2Fe0b5701f%2F6599306507912123f90f150a8bfaaf6c%2Fscreenshot-2026-01-28-at-10.53.16-am.png%3Fwidth%3D100%26height%3D100%26quality%3D100&w=96&q=75)

Paul JonesHead of Growth at Classet

](/blog/authors/paul-jones)

September 10, 2026

Hiring Tips

Quick answer

Write your job posting to close the gaps. In a 2026 experiment with 6,000 workers, people inferred unlisted things about a job from whatever the ad did mention, and a higher posted wage made them expect more stress. So name the schedule, the supervisor, the overtime rules and what the first week looks like. Candidates fill silence with the worst job they have ever had.

Most advice about how to write a job posting is about what to add. Add a benefits section. Add culture. Add an emoji, apparently.

Five economists just showed that the part of your posting doing the most work is the part you never wrote.

## Candidates grade the blanks

Pawel Adrjan, Mária Balgová, Simon Jäger, Jonas Jessen and Jason Sockin ran a pre-registered experiment on 6,000 workers, published as [RFBerlin Discussion Paper No. 184/26](https://www.rfberlin.com/wp-content/uploads/2026/07/26184.pdf) in July 2026 and pre-registered as AEARCTR-0017871. Each person saw job ads where the pay and the listed perks had been randomly shuffled. Then the researchers asked what else they believed about the job: whether it was stressful, whether there was a pension, what the atmosphere was like, whether the employer was family friendly.

None of that was on the page. Everyone answered anyway.

3-7 pts

Belief shift across all ten unlisted attributes when remote work was advertised

40 pts

More likely to call the employer advertising remote work 'family friendly', head to head

21 pts

More likely to assume a pension exists when the ad mentioned training

Mention one thing and workers revise their opinion of everything else. Advertise a flexible schedule and you gain 28 points on family friendliness. Advertise a subsidized lunch and people become 11 points more likely to believe there are employee discounts, which nobody claimed.

The sample is a representative panel of German workers reacting to hypothetical ads in a survey rather than hourly warehouse workers in Ohio, and it has not been through peer review yet. What travels is the mechanism, because reading a workplace off a job ad is something every applicant does, and it gets stronger the less someone knows about you. A VP applying to your company has a network and a Glassdoor rabbit hole. A CDL driver comparing four postings on a phone at a truck stop has your ad and nothing else.

## Your wage is saying something you did not intend

Post a higher number and workers do expect better benefits. They also expect the job to grind them, because higher pay made people more likely to believe the work would be stressful.

The free-text answers are worse than the charts. One worker shown a high salary wrote that a lot is probably expected of employees. Another, shown a low one, guessed it was either a greedy boss or a family business, and either way high pressure. A third saw a short vacation allowance and decided the place was understaffed.

So if you raised your starting wage to compete, you may have bought yourself a reputation for burnout at the same time. The fix is cheap: give the number a reason. "$28/hr because this is the overnight shift and nights carry a $4 differential" reads differently from "$28/hr" sitting alone at the top of a posting.

## You cannot ask them to stop guessing

Adrjan and his co-authors tried to switch the guessing off. Half the participants were told, in plain language, to assume everything else about the two jobs was identical. It changed nothing. The two groups came out the same.

The guessing comes from what people know about how jobs tend to go together, applied to your posting whether you like it or not. Your only move is to answer the question before they ask it.

## How to write a job posting that closes the gaps

Six blanks account for most of what candidates invent about an hourly job: the schedule, the supervisor, overtime, the physical demands, the first week, and why the last person left. Write one line on each, attach a reason to the pay, and cut the benefits list to the two that are true. On a posting you already have, that rewrite takes about twenty minutes.

### Answer the six things people guess about

Each line removes a blank that a candidate was about to fill in with their last bad job, so write them plainly and skip the adjectives.

What they guess about

A line that closes it

Schedule

"Nights, 6pm to 6am, seven on and seven off. The rotation is posted a month out."

Supervisor

"You report to Marcus, who runs second shift and has been here nine years."

Overtime

"Overtime is voluntary and posted a week ahead, usually 4 to 8 hours in peak season."

Physical demands

"You are on concrete the whole shift and lifting up to 50 lbs."

First week

"Week one is paid orientation, forklift certification, and shadowing on days."

Why the seat is open

"The last person in this role moved up to lead in March."

### Put the schedule above the fold

For hourly roles this outranks almost everything except pay. "Four tens, Monday to Thursday, 6am start, no weekend call-ins" answers a question your competitor's posting leaves hanging.

### Give the wage a reason

Because pay signals stress, an unexplained number invites the worst reading. Attach the why: a differential, a certification, a busy season.

### Cut the perks parade

A wall of benefits reads as compensation for something. Pick the two that are true and specific, then spend the space you saved on the day-to-day.

### Replace adjectives with facts

"Fast-paced environment" tells a candidate you are hiding something. "You will move about 400 cases a shift" tells them whether they want the job, which is the whole point.

### Test one requisition at a time

Rewrite a single high-volume posting, leave the rest alone, and compare completed applications rather than clicks. Clicks only tell you the headline worked, while a completed application is the closest thing you get to a candidate saying they believed the rest of it.

## The gap your posting still cannot close

Even a good posting runs out of room. Somebody will apply at 11pm with a question you did not anticipate, and what happens next decides whether they stay interested or go back to guessing.

That is the part we built [Joy](/product) for. She calls applicants within minutes of them applying, in 25+ languages, answers what the posting could not, asks the same job-relevant questions every time, and hands your recruiter a summary. No resume screen, no camera, and a person makes every hiring decision. Teams running [light industrial](/use-cases/light-industrial) and [logistics](/use-cases/warehouse/phoenix) reqs use it to catch applicants in the window where they are still paying attention.

## Common questions

What should a job posting include for hourly roles?

Pay with a reason attached, the exact schedule, who you report to, overtime rules, physical demands, and a short description of the first week. Those are the items candidates guess about when you leave them out, and the guess is rarely generous.

Does listing more benefits get more applicants?

Not reliably. The RFBerlin experiment found that every amenity you list also shifts beliefs about the things you did not list, so a long benefits section can read as compensation for a hard job. Two specific, true benefits plus a concrete description of the work does more than a wall of perks.

Should I put the salary in the job posting?

Yes, and give it context. Posted pay moves applications, but the same research found higher advertised wages made workers expect a more stressful job. Naming why the number is what it is, such as a shift differential or a required certification, keeps the wage from reading as hazard pay.

How long should a job posting be?

Long enough to answer the six things above, which usually lands between 250 and 450 words for a frontline role. What matters is whether it answers real questions. A short posting that does beats a long one padded with adjectives.

How do I know if my job posting is working?

Track completed applications per view rather than clicks, and run one rewritten requisition against your standard template for two weeks. Clicks tell you the headline worked. Completions tell you the posting was believed.

## What to change on your next requisition

Open your highest-volume posting and read it as a stranger who has never heard of you. Every place it goes quiet, a candidate is writing their own version, and their version comes from wherever they worked last. Close three of those gaps this week and leave the rest of the posting alone, so you can tell what moved.

[See how Joy answers the rest](/demo)

[![Paul Jones](/_next/image?url=https%3A%2F%2Fassets.basehub.com%2Fe0b5701f%2F6599306507912123f90f150a8bfaaf6c%2Fscreenshot-2026-01-28-at-10.53.16-am.png%3Fwidth%3D100%26height%3D100%26quality%3D100&w=128&q=75)

Paul Jones

Head of Growth at Classet

Paul comes from an operator background running an Alpine-owned company, and brings firsthand experience with the hiring challenges Classet was built to solve. He's driven by a belief that the right technology can make meaningful work more accessible.

](/blog/authors/paul-jones)

Follow Classet in Google

Add classet.ai as a preferred source and our hiring research is more likely to show up in your Top Stories and AI Overviews.

[Add as preferred source →](https://www.google.com/preferences/source?q=classet.ai)

## Explore More

### Use Cases

-   [RPO / BPO Recruiting](/use-cases/call-centers-bpo)
-   [Data Centers Recruiting](/use-cases/data-centers)
-   [Healthcare Recruiting](/use-cases/healthcare)

### Integrations

-   [Greenhouse](/integrations/greenhouse)
-   [Bullhorn](/integrations/bullhorn)
-   [Lever](/integrations/lever)