How to Read a Job Posting Like a Forensic Accountant
A job posting is a disclosure document written by a party with an interest in the outcome. Nobody taught you to audit one — so here is the method, including what a salary range 60 grand wide is actually telling you.
The range was $95,000 to $155,000.
I read it three times on a Tuesday morning, and the third time I finally saw it for what it was — not a salary range but a legal artifact, a sixty-thousand-dollar-wide band of nothing, sitting in a posting because a statute required a number to be present and no statute required the number to mean anything.
Sixty thousand dollars is not a range. It's the distance between two different jobs. Somewhere in that gap is the actual budget line for this role, known precisely by at least four people at that company, none of whom were going to tell me before I'd spent six hours in their process.
And here's the thing that annoyed me most, sitting there with my coffee going cold: I had been reading these documents wrong my entire career. I'd been reading them as descriptions. They are not descriptions. They are disclosures — filed under compulsion, drafted by someone protecting an interest, and structured exactly like every other document produced under those conditions.
You already know how to read one of those. You just haven't been told that's what this is.
Disclosure regimes produce compliance, not honesty
This is the whole frame, and it's borrowed wholesale from securities filings, where it has been true for ninety years.
When you compel a party to disclose something, you don't get honesty. You get the minimum artifact that satisfies the rule, drafted by someone whose job is to satisfy the rule at the lowest cost. This isn't cynicism about hiring specifically; it's the observed behavior of every entity that has ever been required to publish a number it would rather not publish.
Pay transparency law is a textbook case, and a young enough one that you can watch the adaptation happening in real time. Eighteen states plus DC now mandate a salary range in the posting. What almost none of them mandate is a ceiling on the width of that range. So the compliant response — the rational, legal, entirely predictable response — is a range wide enough to contain every outcome the company might want.
$95,000 to $155,000 is not a company being evasive. It's a company being compliant, in precisely the way the rule permits.
The forensic reading
An auditor doesn't primarily ask is this true? They ask three sharper questions:
- What was disclosed, and what was omitted from the same category? (A range with no midpoint. Requirements with no seniority. Benefits with no employer contribution.)
- What is the structure of the disclosure hiding? (Width. Vagueness. Passive voice.)
- What did they disclose by accident? (This is where everything useful lives.)
Apply those three to a job posting and it stops being an advertisement and becomes evidence.
What the width is actually telling you
The width isn't noise. It carries information — just not the information it appears to.
A tight range ($112,000–$124,000) usually means a real, budgeted, banded role. Someone has done compensation work. The number exists.
A very wide range usually means one of three things, and you can often tell which:
- The level isn't decided. They'll hire a senior person at the top or a junior at the bottom and shape the job around whoever they get. Legitimate, common, and worth knowing — because it means the role's scope is negotiable in a way most aren't.
- The range spans multiple internal bands and they've published the union of all of them rather than the one for this job.
- They intend to anchor low and the top number is doing marketing.
Here's the part that turns this from a nuisance into something with a measurable cost. A team at Cornell's School of Industrial and Labor Relations — Alice Lee, Tae-Youn Park and Sungyong Chang — published four studies on exactly this in the Journal of Applied Psychology in February 2026, beginning with an archival analysis of nearly 10 million U.S. job postings.
Their finding, in Lee's own words: "women show a stronger preference for jobs with narrower salary ranges compared to men."
And the effect doesn't stop at whether you apply. Applicants who chose narrower-range positions "negotiated less assertively, and they were more satisfied with a midpoint salary offer."
So the width of a band that exists purely as a compliance artifact is shaping who applies and how hard they push once they do. Nobody designed that. It's a second-order consequence of a well-intentioned law meeting a drafting incentive, which is how most of the world actually gets built.
“When that information was provided, we no longer observed the gender gap in application decisions, and it also eliminated the gap in negotiation behaviors.
”
That's the practical payload of the whole study: the fix isn't narrower ranges, it's context — the typical starting salary and the factors that determine the final offer. Which is also precisely what you should ask for in a first call, and now you can ask for it citing a reason rather than sounding like you're haggling.
An honest correction about "ghost jobs"
You have seen the statistic. A third of job postings are fake. It circulates constantly, and it is emotionally satisfying in a way that should make you suspicious.
The common construction goes: employers reported roughly 7.4 million openings but only about 5.3 million hires, so the missing two million are phantom listings. I've seen that arithmetic in trade press repeatedly.
The arithmetic is invalid, and I want to be exact about why, because the underlying situation is bad enough without inventing a number.
Stock versus flow — the error in one paragraph
BLS reports job openings as a stock: a snapshot of positions open on the last business day of the month. It reports hires as a flow: every hire made across the whole month.
Subtracting a flow from a stock is a category error. A job posted on the 3rd and filled on the 20th appears in the hires figure and has already vanished from the openings snapshot. The two numbers aren't the same kind of thing, so their difference isn't "unfilled jobs" — it isn't anything.
The actual BLS release for June 2026 states it plainly: "The number of job openings was little changed at 7.4 million in June... Hires were unchanged at 5.3 million, while total separations changed little at 5.4 million."
Look at the third number instead, because it's the one nobody quotes. Separations (5.4M) exceeded hires (5.3M) that month. That's a labor market where slightly more people left jobs than entered them — a genuinely useful fact about your leverage, arrived at without inventing a conspiracy.
Do postings exist that nobody intends to fill? Yes — pipeline-building, internal candidates already chosen, roles frozen after posting, compliance postings for positions effectively spoken for. Any recruiter will tell you so. But the honest version is "some, and you can't derive the number this way," and a piece that tells you a fake number to make you feel better about a rejection has picked its own convenience over your interests.
The turn
Here's what I actually think is going on, and it took me longer than it should have to see it.
Every one of these behaviors — the sixty-thousand-dollar range, the requirements list nobody meets, the title one level above the work — is a rational response to the posting being a document with more than one audience.
That posting is read by candidates. It's also read by the current team, who will notice if the range implies a new hire out-earns them. By the internal candidate who was privately promised this role. By the compensation team enforcing bands. By a lawyer checking statutory compliance in eighteen jurisdictions. By a competitor mapping the org.
No single sentence can be optimized for all of those readers, so the document optimizes for the ones with power inside the building. You, the applicant, are the reader with the least leverage and the least recourse, and the text reflects that with total honesty.
That isn't a moral failing by whoever drafted it. It's the same mechanism that makes the published org chart a fiction: a formal artifact compresses a political reality, and the compression is lossy in whichever direction protects the people doing the compressing. The posting is the org chart's cover letter, and it lies for the same structural reasons.
Once you stop reading it as an offer and start reading it as a filing, the frustration mostly drains out. You are not being personally deceived. You are reading a document that was never written to you.
The audit
- 1
Measure the range as a percentage, not a dollar gap
Divide the width by the midpoint. Under 15% is a real band. Over 40% means the level is undecided — which is information, not an insult, and it means scope is negotiable. Then ask the Cornell question directly on the first call: what's the typical starting salary for this level, and what determines where in the band an offer lands? That's the exact intervention that erased the effects in the study.
- 2
Read the requirements for grammar, not content
"Experience with X" is a filter. "Deep expertise in X" is someone describing the person who just left. "Familiarity with X" means nobody on the team has it. A list of twelve technologies means the role absorbed the work of two departed people — count the clusters and you can usually see the seams where one job was welded onto another. What you are reading, in that case, is a handover that never actually happened, transcribed into a requirements list.
- 3
Compare the title to the reporting line
Title inflation is nearly free, so titles run ahead of scope. The reporting line isn't free, because it maps to actual authority. "Director, reporting to a Senior Manager" describes a title given instead of a raise. If the posting omits the reporting line entirely, that omission is the disclosure.
- 4
Look for what's missing from a disclosed category
The forensic move. If they list benefits but never the employer's contribution percentage, that's an omission from a category they chose to open. Equity mentioned with no vesting terms, no strike price, no total-shares denominator is not a compensation disclosure — it's the word "equity."
- 5
Date the posting and check for reposts
A role live for four months is either frozen, unfillable at the offered number, or a pipeline. All three change your strategy — and all three are the visible surface of a decision somebody keeps deferring at a small recurring cost. A role reposted with a quietly revised range tells you what the market said back to them, which is the closest thing to an audited figure you will get.
- 6
Price the whole thing against your real hourly rate
Base salary is one line in a longer calculation involving commute, on-call, and hours the posting will never state. Run the actual arithmetic on what your time costs before the number impresses you — the same discipline that stops a freelancer underpricing a project stops an employee misreading an offer, and the real hourly rate is almost always worse than the headline once you count the hours nobody invoices for.
Isn't this just teaching people to be paranoid about employers who are mostly acting in good faith?
The strongest objection, and mostly I agree with its premise — the person who drafted that posting is usually a recruiter with sixteen open roles and a template, not a schemer. But good faith and structural distortion coexist comfortably. A wide salary range is what an honest person produces when the rule says publish a range and the compensation team says don't commit. Nothing here requires anyone to be lying; it requires only that the document serve readers other than you, which it demonstrably does. Reading it accurately isn't paranoia, it's literacy.
If the fake-jobs number is wrong, aren't you downplaying a real problem?
No — I'm refusing to defend a real problem with a broken statistic, which is how real problems get discredited. Postings that nobody intends to fill exist and they waste enormous amounts of people's time. What doesn't exist is a defensible route from BLS openings-minus-hires to a count of them, because those two figures aren't the same kind of measurement. If you want the strong version of this argument, make it from recruiter surveys and platform data with disclosed methodology, and quote their actual sample sizes.
Doesn't pay transparency law still help, even with wide ranges?
Yes, and the piece shouldn't be read as saying otherwise. A wide range beats no range: it establishes an order of magnitude and a ceiling someone has to defend. The criticism is narrow — mandating disclosure without constraining its form produces predictable evasion, and the Cornell result shows that evasion has costs falling unevenly. That's an argument for better-drafted rules, not for repealing them.
What if I'd rather just apply and find out?
Reasonable, and for a role you actively want, apply — this method is for triage when you have forty tabs open and six hours. Its real value isn't screening jobs out. It's walking into the first call already knowing which two questions to ask, which is the difference between a conversation you're being processed through and one you're participating in.
Back to the range
I asked. That's the whole ending — I got on the call and asked what the typical starting salary was for the level, and what determined where in the band an offer landed.
The answer took nine seconds: most people at this level start around $118,000, and the top of the band is for candidates already doing the job's next level at their current employer. Sixty thousand dollars of ambiguity collapsed into one number and one condition, instantly, because someone had simply never been asked and had no particular interest in withholding it.
That's the part I keep turning over. The information wasn't protected. It was undisclosed by default, sitting one question away, in a document optimized for readers who weren't me — and I had spent years treating that document as though it were the whole story rather than the cover sheet.
It's a filing. Read it like one. Then ask the question it was drafted to avoid answering, and notice how rarely anyone refuses.
It's just business.
Sources
- U.S. Bureau of Labor Statistics — Job Openings and Labor Turnover Summary, June 2026 (released 4 August 2026)
How this was checked
Openings 7.4 million (stock, last business day of month); hires 5.3 million and total separations 5.4 million (flows, whole month). The stock/flow distinction is why openings-minus-hires does not yield a count of unfilled or fake postings.
- Cornell Chronicle — 'Job listings with wide pay ranges may deter female applicants'
How this was checked
Alice Lee, Tae-Youn Park and Sungyong Chang, 'The Implications of Pay Range Transparency on Job Application Preferences and Negotiations', Journal of Applied Psychology, 16 February 2026. Four studies, beginning with an archival analysis of nearly 10 million U.S. job postings.
- Center for American Progress — Quick facts about state salary range transparency laws
How this was checked
Background on which jurisdictions mandate salary ranges in postings, and the general absence of any constraint on how wide a disclosed range may be.
- HR Dive — 'US job listings go nowhere, creating a ghost job economy'
How this was checked
Example of the widely-circulated openings-minus-hires construction this piece declines to use. Cited so readers can check the reasoning rather than take my word for the objection.
Keep reading
This article is educational and satirical content from Business Dog. It is not financial, legal, or tax advice. It's just business.