Someone asks what you expect to earn, and your stomach drops. You throw out a number pulled from a friend's cousin's situation, hope it lands, and spend the next week wondering if you just left money on the table.
I've been on both sides of that table. I've lowballed myself badly (more on that later), and I've also walked into a negotiation with a spreadsheet so detailed the hiring manager asked me to send it over. The difference between those two conversations had nothing to do with confidence. It came down to one thing: I knew how to research market salary ranges before the negotiation started, and I wasn't guessing anymore.
This is the method I now use every single time. It's boring, it takes about two hours, and it works.
Key Takeaways
- Benchmarks are raw material, not answers. You need at least three sources before you trust any number.
- Auto-reported salaries skew low in some roles and high in others. Always cross-check against posted job ranges.
- Build three figures, not one: your target, your walk-away, and the number you open with.
- Geography and company stage move the range more than your job title does.
- A 6% ask is normal; a 20% counter is aggressive and needs evidence behind it.
- The best answer to "what are your expectations?" is a researched range, never a single figure.
Why salary research changes the conversation entirely
A number you can defend behaves differently in a room than a number you hope for. When you say "the range for this role at this level, in this city, sits between X and Y, and here's why I'm anchoring at the upper end," the negotiation shifts from a haggle to a discussion. The other person has to engage with your logic instead of your feelings.
Most candidates never do this. They come in with a vibe, an outdated figure from their last job, or whatever a recruiter mentioned on the phone. Which means the person across from you is usually arguing against someone who hasn't prepared. Do the preparation and you're already ahead.
The real problem with benchmark sites
Glassdoor, Payscale, and the NACE Salary Calculator are all useful, and I use them. But understand what they're actually showing you. Glassdoor figures are largely self-reported and often years old for niche roles. Payscale gives you a decent median but blurs seniority bands together. The NACE tool is genuinely good for recent graduates because it's built around entry-level data, not mid-career averages.
The catch? These numbers disagree. Sometimes by 20%. And that's not a bug, it's the whole point. The disagreement tells you where the uncertainty is, and uncertainty is where you negotiate.
How to build a range you can actually defend
Here's the process I follow, in order. It takes one evening.
Step one: cross-reference three different source types
Don't just stack three salary sites. Pull from three categories that fail in different ways:
- Aggregator sites (Glassdoor, Payscale, Indeed) for a rough median, knowing the data is self-reported
- Live job postings with published ranges, which are the single most honest signal because companies now have to post them in many jurisdictions and tend to post realistic bands
- Industry-specific reports from staffing firms or trade associations, which weigh by seniority properly but may be a year behind
If you're in a unionised or collective-bargained sector, skip half of this and find the pay grid. It's published, it's binding, and it ends the debate instantly.
Step two: adjust for geography and company stage
A title pays differently in a capital city than in a regional town, and differently at a 40-person startup than at a listed company. Two adjustments matter most:
- Cost of living: use a published index to shift the national median toward your location.
- Company stage: early-stage firms often pay below market on base and make up for it in equity or flexibility. If you value the cash, you need to say so and price it in.
I once took a role at a small firm and accepted a base 12% below market because the equity story sounded good. The equity was worth nothing. That's a mistake I don't repeat: now I ask for the cash equivalent of any equity offer before I compare.
Step three: set your three numbers
This is where most people stop too early. A range isn't enough. You need three distinct figures:
| Figure | What it is | How to set it |
|---|---|---|
| Target | The number you'd be genuinely happy with | Upper third of your researched band, adjusted for your specific experience |
| Walk-away | The number below which you decline | Based on your actual financial needs and alternative options, not pride |
| Opening ask | The figure you name first | Target plus a margin you can justify with evidence |
Write them down before the conversation. Seriously, on paper. The act of committing to the walk-away number in advance is what stops you from caving at minute forty when everyone's tired.
The questions everyone asks before the call
What is the #1 rule of salary negotiation?
Never name a number first if you can avoid it, and if you must, never name a single figure. The first number anchors everything that follows, so the person who speaks first sets the ceiling. When pushed, give a researched range and add that it depends on the full package. That keeps you in the conversation without handing over your floor.
How to negotiate salary based on range?
Use your range as the frame, not as a demand. Say something like: "Based on my research for this role and level in this market, I'm seeing a band between X and Y. Given my experience with [specific thing], I'd be targeting the upper end." Then stop talking. The silence is doing the work. If they counter below your target, respond with your evidence rather than your disappointment.
Is a 6% raise too much to ask for?
No. A 6% increase sits comfortably within normal annual movement for most roles, and it's a reasonable ask when you're staying put and asking for a review. It gets harder to justify only if your role's market band hasn't moved and your responsibilities haven't changed. Bring your research and the ask stands on its own.
Is a 20% counter offer too much?
It's aggressive, not impossible. A 20% jump usually only lands when you're underpaid relative to market, you're moving into a new title, or you have a competing offer. If you're asking for 20% purely because you'd like it, you need to be ready for a firm no. If your research shows the market genuinely pays that much more, say so and cite the band.
What to say when they ask what you expect
The salary expectations question trips up more people than any other. Here's how to handle it without a script that sounds rehearsed.
If you have no experience, don't invent a number. Say you're focused on learning the role and ask what range is budgeted for the position. Then use their figure as your starting point, not your ceiling. Recruiters almost always have a band, and asking for it is not rude.
If you have experience, answer with your range and one sentence of justification. That's it. You don't need a paragraph. And whatever you do, don't answer with a number you'd be unhappy with just to keep the peace.
Where the hidden numbers live
Beyond the obvious sites, a few sources punch above their weight. Public visa filings in some countries list actual offered salaries for sponsored roles, which is real compensation data, not self-reported. Government labour statistics give you sector-level medians you can trust structurally. And recent postings from competitors often reveal the top of the band, because companies advertise their best case.
In my experience, cross-referencing a visa filing against a job posting for the same role has resolved more disagreements than any salary site ever did.
The part nobody mentions
Research doesn't remove the discomfort. It just gives you something to hold onto when your hands are sweating and the hiring manager says "that's above our band." You'll still feel the urge to fold. The difference is you'll have three numbers written down and a reason for each one, and that's usually enough to keep your voice steady.
The candidates who negotiate well aren't the boldest ones. They're the ones who did the homework and trust it more than they trust the pressure of the moment. Two hours of research. Three figures on a page. Then let the conversation happen.
