What your job ad looks like next to everyone else's

Most arguments about publishing salary ranges are arguments about principle. This one is not. We hold 9,188 live job postings from 518 employers across 51 countries, gathered between 2026-07-10 and 2026-09-25, so we can tell you what the page your advert appears on actually looks like.

Between 24% to 31% of those postings state what the job pays. Split by country it stops being one number.

• India: 1,426 live postings, 0% to 1% state pay.
• United Arab Emirates: 195 postings, 0% to 2%.
• Germany: 1,624 postings, 0% to 1%.
• United States: 2,635 postings, 69% to 78%.

The positioning argument, stated plainly

If you are hiring in India or the Gulf and you publish a band, your advert is one of a very small number on that page that answers the question every candidate opens with. That is not a moral victory, it is scarce inventory. Differentiation usually costs money. This one costs a sentence.

We are going to be careful here, because the obvious next claim is the one we cannot support. We do not have enough hiring volume of our own to tell you that publishing a band raises your application count, and we are not going to borrow somebody else's survey and present it as if we had measured it. What we can show you is the competitive picture, and the competitive picture is that the field is empty.

The obligation probably already applies to more of your roles than you think

Across our corpus, 1,128 remote postings state pay at 39% to 47% against 22% to 29% for 8,018 onsite ones. That gap is not culture. It is jurisdiction leaking.

A pay range requirement attaches to where the work can be done, not to where your company is registered. Post a remote role that somebody in California, New York or Washington can take, and you are inside a regime that wants a range in the advert. Employers with a single remote listing and no US entity are the ones who find this out late.

Industry norm is not a real constraint, and your own peers prove it

The usual objection is that nobody in the sector does this. Hold the country and roughly the sector constant, take United States postings from employers with at least 40 live roles in our corpus, and the spread makes that hard to sustain.

• Datadog: 212 postings, 92%.
• Databricks: 456 postings, 89% to 97%.
• Okta: 108 postings, 85% to 98%.
• GitLab: 84 postings, 56% to 74%.
• Stripe: 329 postings, 6% to 8%.
• Jane Street: 101 postings, 0%.

Optiver and Jane Street compete for the same graduates out of the same universities. Optiver states pay on 68% of its 57 United States postings and Jane Street on 0% of its 101. Whatever explains that, it is not an industry norm, because they are the same industry.

The one that should bother you

Of 946 live internships in our corpus, 1% to 2% state what they pay. Of 353 apprenticeships, 0%.

If you run an internship programme and you are trying to be the employer students talk about, this is the cheapest reputational win on the table, and it is sitting there because essentially nobody in the category has taken it.

The real objection, which is not the one people say out loud

The stated reason for leaving pay out is usually market sensitivity or flexibility. The actual reason, most of the time, is internal equity. Publish a band for a role you already employ four people in, and you have published something about those four people. If two of them are inside the band and two are below it, you have not created that problem, you have surfaced it, and the surfacing lands on whoever has to hold the conversation.

That is a real cost and it deserves to be named rather than dismissed, because pretending it is about competitors is how the decision gets deferred forever. It is also finite. The set of people the band exposes is knowable before you publish, the conversations are schedulable, and every quarter you wait adds names to the list. The cost is front loaded: it falls on the first cycle, and it does not recur at the same size.

What a usable band looks like

• A range you would actually pay across. If the top of the band is a number no current holder of the role has reached, candidates work that out quickly and the disclosure costs you credibility instead of buying it.
• Narrow enough to be information. A span wide enough to cover three seniorities answers nothing, and reads as a refusal written in the format of an answer.
• Currency, period and what is included, stated once. A number without a period is ambiguous in exactly the markets where disclosure is rarest, and equity or commission folded silently into a base figure is the complaint candidates raise most.
• The same band in the advert and in the first recruiter conversation. A range that moves between the posting and the call is the specific failure that makes candidates distrust every published band afterwards, including other employers' bands.

What to do about it this quarter

• Decide the question once, at the level of policy, rather than per requisition. A band that appears on some adverts and not others tells candidates something you did not intend to tell them.
• Audit your remote listings first. That is where an obligation may already exist, and it is a smaller job than auditing everything.
• Publish a real band, not a legal one. A range wide enough to be safe is read as a refusal, and a refusal dressed as a disclosure is worse than silence.
• Start with the internships. Lowest stakes, highest signal, and the category average is close to zero.

Our count is reproducible: it comes from one script in our repository against the live table, and re-running it moves the numbers in this post. It is our ingested corpus rather than the whole market, it leans towards employers who publish structured postings, and our own parser misses some real disclosures, which is why every rate above is a range rather than a point.