How to Actually Use a Salary Guide When You're Building Compensation Ranges
Most people treat a Salary Guide 2022 like a pricing menu. It is not. It is raw survey data that needs adjustment before it means anything in your organization, and if you skip that step you will offer too little to your actual market or too much and bleed budget for no reason.The fundamental mistake I see repeatedly is taking the median line and plugging it directly into a job posting. The median is an aggregate of every company in that survey, including startups, enterprise, and everyone in between. It also assumes the target role matches the surveyed description exactly. In practice, job descriptions drift. A "Senior Software Engineer" at one company is a mid-level engineer at another, and the salary guide cannot tell the difference. There are a few reliable sources. Radford by Amcat, Mercer, and WIllis Towers Watson are the big three for tech and general professional roles. Robert Half publishes annually and is more accessible for smaller companies that cannot justify the consulting budget. If you need something free, Payscale and Glassdoor offer aggregated data, though the signal-to-noise ratio is lower and the sample sizes for niche roles can be thin. For my own reference, I typically pull from Radford for engineering and product roles, and Mercer for broader business functions. The Radford data runs about six months behind the publication date, which matters if you are in a high-inflation year like 2022 was. Everything in those guides was collected before the market started repricing aggressively, so the numbers are already conservative by the time they land on your desk.
Building Your Own Range from the Raw Data
Here is the method I use and have used for years, because the default percentile spreads in these guides are too wide for actual use. Take the 25th, 50th, and 75th percentiles from the guide for your specific role and location. Then compress them. I apply a 10-to-15 percent spread below the median and a 20-to-25 percent spread above it. That gives you a practical range instead of the often 40-percent-plus band the guide itself suggests. Most companies do not pay above the 75th percentile for any single role, and if you are, you have a compensation philosophy problem, not a hiring problem. One thing that catches people off guard: the guide data is usually self-reported. Companies round up. They round into favorable brackets. The published medians are systematically higher than what actually lands on pay stubs. I adjust the medians downward by roughly 5 percent to account for this reporting bias. It is a blunt tool but it prevents the embarrassing situation where your upper band is still below what your actual top performers are earning.
A Specific Problem I Faced
Last year I was building ranges for a data engineering team and the guide showed a clean median for "Data Engineer II" at a specific city. What the guide did not show was that half the companies in that survey bucket were remote-first, and the other half required on-site presence. The median blurred two completely different labor markets into one number. I ended up with a range that was adequate for the on-site candidates but uncompetitive for remote, and we missed a few strong applicants who had offers elsewhere that our range could not match. The workaround was simple but tedious. I pulled the guide data, then segmented it by work model using internal HRIS data and a few public job postings to estimate what fraction of surveyed companies operated remotely. I gave the remote segment a 7 percent uplift and the on-site segment a 3 percent adjustment. It took about an afternoon of spreadsheet work instead of ten minutes, but the resulting ranges actually competed in both segments.
Get the Full Details
What Beginners Miss About These Guides
The biggest blind spot is total compensation. A Salary Guide 2022 entry showing a $95,000 median looks reasonable until you check whether that includes equity, bonus, and benefits. Tech guides tend to fold equity into the base number for senior roles, which makes the base look inflated compared to the non-tech guide for the same title. Compare apples to apples by confirming the comp composition before you make any offer. Another thing: title inflation. The guide will have a band for "Principal Engineer" that overlaps significantly with the "Senior Engineer" band at the high end. This is not an error in the guide. It is a feature of how titles work across companies. You need to map titles to actual responsibilities, not to the alphabetical label in the survey. I use a simple leveling rubric that describes scope, autonomy, and impact. Once you have that, the guide data slots in without the confusion of competing title semantics.
When a Salary Guide 2022 Does Not Help You
It fails in a few clear scenarios. Emerging roles with insufficient survey samples, highly specialized domains where fewer than fifty companies reported data, and niche geographies outside the major metro areas covered in the guide. If your role does not appear explicitly, do not force a proxy. The error bars on small samples are large enough that you are more likely to be wrong than right. For those cases, I fall back to three signals: the actual offers your competitors are making (gathered through recruiter networking and interview feedback), your own historical offer acceptance rates by band, and the current posting data from LinkedIn or Indeed. None of those are perfect, but they are closer to reality than interpolating from a guide that has a hollow center for your specific situation. The guides are useful. They just require actual work to make them useful.
