How to Read Compensation Data Before You Negotiate (2026)

Three Sources, Three Different Numbers

Look up what your job pays and you will get three answers that do not agree. A federal dataset says one thing, a job board says another, and a crowd-sourced salary site says a third. None of them is lying. They are measuring different populations with different methods, and using the wrong one is how people talk themselves out of thousands of dollars.

This guide is about reading the data, not about quoting a number. Wage figures move; the reading skill does not.

Start With the Government Data

The Bureau of Labor Statistics runs the Occupational Employment and Wage Statistics program. It surveys employers — not workers — across every metropolitan area and occupation code, and publishes wages at the 10th, 25th, 50th, 75th, and 90th percentiles.

Three reasons to start there:

• Employers report it. No self-selection bias from people who felt good enough about their pay to post it.
• It is geographic. National medians are close to useless for negotiation. Metro-level data is the number your employer's own compensation team is looking at.
• It shows the spread. A median tells you the midpoint. The 75th and 90th percentiles tell you what the top of the range looks like, which is where a negotiation actually happens.

The trade-off is lag. Survey data is collected and processed over months, so it describes the recent past rather than today's posting. Treat it as your floor and your structure, not your ask.

Then Read Job Postings as Live Market Data

Pay-transparency laws in a growing number of states require employers to publish a good-faith pay range on the posting. That is the closest thing to real-time compensation data available, and it is free.

How to read a posted range properly:

• The top of the band is usually reserved. Most employers hire in the lower-middle of the posted range and keep headroom for promotion.
• Wide bands hide levelling. A range spanning twenty thousand dollars usually covers two or three internal levels. Ask which level the role sits at before treating the top as reachable.
• Collect ten, not one. A single posting is an anecdote. Ten comparable postings in your metro is a distribution.

Treat Crowd-Sourced Averages With Care

Self-reported salary sites have real value for job titles that government occupation codes describe poorly — anything new, niche, or fast-moving. But three biases run through them:

• . Self-selection. People with unusual pay, high or low, are likelier to report.
• . Title inflation. Titles are not standardised across employers, so the sample mixes seniority levels.
• . Staleness. Entries can sit for years with no timestamp shown.

Use them to sanity-check a range, never as your primary source in a conversation with a hiring manager.

Compare Total Compensation, Not Base Pay

Base salary is the number everyone quotes and the least complete one. Employer-paid benefits — health premiums, retirement match, payroll taxes, paid leave — add a substantial share on top of wages, and the share differs enormously between employers.

When you compare two roles, build the same four lines for each:

• . Base pay, annualised
• . Expected variable pay — overtime, shift premiums, bonus at realistic attainment
• . Employer benefit contributions
• . Costs the job imposes: commute, parking, required tools, unpaid time

Then divide by the hours the job actually consumes. The pay comparison calculator does this arithmetic for two offers at once, including the overtime line that base-pay comparisons always miss.

Do Not Forget the Overtime Line

For hourly and non-exempt salaried roles, overtime is often the difference between two offers, and it never appears in a salary survey. A role averaging six overtime hours a week at time and a half adds roughly 22% to annual earnings above base — real money that a base-pay comparison treats as zero.

Two cautions. Model it at last quarter's actual hours, and model it again at zero: if an offer only wins with heavy overtime, your income is exposed to the next slow season. And confirm the role is genuinely non-exempt before counting the premium at all — a salaried exempt role produces no overtime no matter how long the weeks run. To see what those hours are worth after withholding, run them through the overtime tax calculator.

Normalise Before You Compare

Two offers in different units cannot be compared until they share a scale. Convert everything to annual and to an effective hourly rate using your real weekly hours — the salary calculator handles every direction of that conversion.

Then apply a cost-of-living sanity check across metros. A twelve-percent raise into a market with materially higher housing costs can be a cut in practice. Compare after-tax, after-housing income rather than the headline.

Building Your Number

Four steps produce a figure you can defend out loud:

• . Anchor on the metro percentile. Find the 50th and 75th percentile for your occupation code in your metro.
• . Adjust for evidence. Move up for scarce skills, certifications, a documented track record; move down for a career change into the field.
• . Ask at the upper-middle. Not the ceiling, not the median. High enough to have somewhere to move.
• . Bring the source. "Metro-level BLS data for this occupation puts the 75th percentile near X, and eight current postings here range from Y to Z" is a fundamentally different conversation from "I think I deserve more."

What Data Cannot Tell You

Whether the hours are predictable. Whether the health plan covers your family's actual needs. Whether the manager keeps people. Whether the overtime is optional in practice or optional on paper. Use data to eliminate offers that are clearly below market, then decide the close calls on stability — because the offer that pays four percent more and burns you out in ten months was not the better offer.

For the pay mechanics behind these comparisons, see hourly vs salaried pay and overtime pay by profession.