Somewhere in your company right now, an offer is being built on numbers from a market that no longer exists. Maybe the salary survey came out this year but collected its data last year. Maybe the benchmark is a title match against companies twice your size. Maybe the range is what you paid the last person in the seat, who was hired in 2022, at the top of the hottest talent market in a generation. Any of those, and the offer is wrong before the candidate ever reads it.
The numbers tell a clean story. In the 12 months ending June 2022, private sector wages and salaries grew 5.7 percent, per the Bureau of Labor Statistics Employment Cost Index, up from 3.5 percent the year before. Salary increase budgets followed: WorldatWork’s long-running survey found U.S. budgets hit 4.4 percent in 2023, the highest in two decades. Then the correction. By June 2026, private sector wage growth had cooled to 3.1 percent, and civilian wage growth to 3.2 percent, down from 3.6 percent just a year earlier. The Atlanta Fed’s Wage Growth Tracker puts median individual wage growth at 3.6 percent as of June 2026.
So the market went up fast, came down slowly, and has now settled near its pre-pandemic pace. That sounds like a return to normal. It is not. Three years of correction did not undo the boom, it froze the boom’s distortions in place and layered new ones on top. People hired at 2022 peaks still hold those salaries. Colleagues hired before or after them do not. Remote work scrambled what geography means for pay. And whole categories of companies moved differently from the averages the whole time.
Which is why compensation benchmarking, a task most organizations treat as routine, is quietly one of the easiest places to lose a great candidate or wound your internal pay equity in 2026. The data is not hard to find. Data that describes your actual market, this quarter, is.
What actually happened to pay, 2021 to 2026
The boom was real and it was uneven. Wage growth ran fastest for people who changed jobs, in lower-wage service roles, and in whatever specialty a given industry was desperate for that quarter. The correction was just as uneven. ADP’s research group reported in early 2026 that the pay premium for switching jobs had shrunk to the smallest gap in its data going back to 2020, with job changers seeing 6.4 percent growth against 4.5 percent for stayers in January. The Atlanta Fed data shows the same pattern: job switchers at 4.1 percent versus stayers at 3.4 percent in June 2026.
Salary budgets tell the employer side. After the 4.4 percent peak in 2023, budgets stepped down each year. WTW reports 2026 budgets landed at 3.5 percent, with 3.4 percent projected for 2027. Mercer’s planning survey found merit budgets at 3.2 percent for 2026, flat to 2025. Payscale’s salary budget survey forecast 3.5 percent for 2026, with 68 percent of U.S. organizations expecting no change in their budgets at all.
Here is the benchmarking problem in one sentence: any data point collected during that ride describes the moment it was collected, not the moment you make your offer. A number from 2022 is inflated. A number from the deepest cutting of 2024 and 2025 may be too low. And an average of the whole period describes nothing at all.
Why stale and generic data breaks offers
The survey lag problem. Most published surveys collect data months before release and get refreshed once a year. In a stable market that lag is tolerable. In this one it produces whiplash. Life sciences shows how fast the picture can flip: BioSpace’s salary reports found average pay in the sector rose 9 percent in 2024, then fell 6 percent in 2025, from $159,419 to $149,945. A team benchmarking against the 2024 figure in mid-2026 would build offers around a market high that lasted about a year.
The compression trap. When merit budgets run at 3.2 percent while job changers command more, every external hire lands above the incumbents around them. The math compounds quietly. An employee hired at $150,000 in 2023 who received three 3.2 percent merit increases sits near $165,000 today. The candidate you recruit for the identical role has been riding the job-changer curve and expects more, and WTW found 36 percent of employers are hiring at elevated salary ranges while 32 percent raise starting salary ranges outright. The distortion cuts the other way too: ADP found new-hire pay frozen near $18 an hour for a year and a half while existing employees kept collecting increases. Benchmark a new offer without checking it against the people already in the seats and you either buy a retention problem or insult the market, depending on which way your data is stale.
The geography mess. Remote and hybrid work broke the old assumption that a role’s market is its metro area, and employers never agreed on a replacement. Some pay one national rate, some run formal geographic differentials, and your competitors for the same candidate may be benchmarking against different cities than you are, or against no city at all. The spreads are material. BioSpace’s hub data put average life sciences pay at $180,564 in the San Francisco Bay Area against $156,700 in greater Boston, a 15 percent gap between the two most famous clusters in the industry, and one that a single national line erases completely.
The company-type blind spot. This is the error we see most in executive and senior searches. A survey percentile that pools every company in an industry hides the structural differences that actually set pay. In life sciences, a large pharmaceutical company competes on cash and bonus while a mid-sized biotech competes on equity, and equity has been the most volatile piece of the package: BioSpace found participation fell from 36 to 30 percent in a year while average equity value dropped from $86,376 to $60,776. A contract manufacturer competes on scope and stability and rarely matches either cash structure. Pool them all and the 50th percentile matches nobody.
The same title, four different markets
Life sciences. A head of quality at a big pharma site, at a 200-person biotech, and at a contract development and manufacturing organization are three different jobs with three different pay structures, whatever the titles say. The biotech role carries broader scope and more risk, priced partly in equity that may be worth far less than the grant letter implied. The CDMO role is a margin business that pays closer to manufacturing norms than pharma norms. Benchmarking any of them against a pooled industry survey misprices all three, and in a sector where average pay swung 9 percent up and then 6 percent down in consecutive years, the vintage of the data matters as much as the cut.
Manufacturing. Manufacturing wage growth came in at 3.4 percent per the ECI, just above the economy-wide pace, but averages hide the fight for specific skills. Plants compete locally for maintenance technicians, machinists, and welders, where the relevant benchmark is the employer 20 minutes down the road and the premium over the survey median is set by who has an opening this month. They compete nationally for automation engineers and plant leadership, where local data misleads in the other direction. A benchmark built at the metro level for a role recruited nationally, or the reverse, misses by design.
Defense. Cleared talent is its own labor market with its own inflation rate. ClearanceJobs’ 2026 report put average total compensation for security-cleared professionals at $126,125, an all-time high after growing nearly 6 percent in 2025, a period when the general market grew barely half that fast. Clearance level moves pay the way a degree once did: polygraph holders average $149,875, roughly $30,000 above other cleared professionals. And this market is restless, with 78 percent of cleared professionals saying they are at least somewhat likely to change jobs within a year. A prime contractor benchmarking against general engineering surveys, or a supplier benchmarking against primes it cannot match, will be wrong in opposite directions.
Service industries. Here the pressure comes from the floor. On January 1, 2026, nineteen states and 47 cities and counties raised their minimum wages, lifting pay for 8.3 million workers by a combined $5 billion, per the Economic Policy Institute. Washington State’s floor reached $17.13, and for the first time more Americans live in states with a $15 minimum than in states at the federal $7.25. Meanwhile the ECI shows retail wage growth at 2.8 percent and leisure and hospitality at 2.5 percent, the softest of any major group. So the middle of the pay structure is quiet while the bottom climbs by law, which squeezes the differentials that make a shift lead or assistant manager job worth taking. A national chain can set one band for general managers and absorb the error in expensive markets. A regional operator competing across three states cannot, and its real benchmark is often the posted range of the national chain recruiting in the same zip code. Which points to the newest data source of all.
Fourteen states now require pay ranges in job postings, per HR Dive’s tracker, and that shift, covered in depth in our earlier article on pay transparency, has a benchmarking side effect: your competitors’ ranges are now public, current, and specific to real openings. Posted ranges are noisy, and some are drawn wide enough to be useless. But as a live signal of what companies your size are offering for roles like yours, they beat a survey percentile that is a year old.
What careful benchmarking looks like now
Triangulate at least three sources, and include a live one. A published survey, aggregated posting data or posted ranges, and real recent offers from your own searches or your recruiting partners. Recruiters see accepted and declined offers in your niche every week, which makes their read the freshest data available. No single source is reliable this cycle. The overlap between three is.
Match on scope, not title. Before pulling any percentile, write down what the role actually owns: budget, headcount, regulatory exposure, and the size and type of company it sits in. Then match against roles with the same scope. A director at a 300-person company can out-scope a VP at a 30,000-person one. Title-matching is how a CDMO ends up pricing against pharma, or a defense supplier against a prime.
Weight for recency and note the vintage. Treat every data point like produce with a date on it. Ask when survey data was collected, not when it was published. Discount anything gathered more than a year ago, and anything from 2021 through 2023 should be treated as history, useful for understanding an incumbent’s salary, useless for setting an offer.
Build a range with architecture, not a single number. A defensible range has a floor tied to internal equity, a midpoint tied to your triangulated market read, and headroom above it with named conditions for use, such as scarce skills, a clearance, or a critical start date. That structure is what lets a hiring manager move fast at offer stage without triggering a compression problem the following quarter.
Write the rationale down. One page: sources used, their dates, the scope match, the internal comparisons checked, and why the range is what it is. It speeds approvals, it keeps the next hire for the same role consistent, and in a transparency-law state it is the document that lets you post a range you can defend.
What this means for boards and CEOs
Compensation data quality is not an HR detail. It is embedded in every hiring decision, every retention save, and every posted range that now doubles as public positioning.
First, audit the vintage. Ask what data actually feeds offers in your organization and when it was collected. Most leadership teams have never asked, and the answer is often a single survey, purchased annually, already aging when it arrives.
Second, fund targeted range resets instead of even spreading. Mercer found 83 percent of employers still distribute salary budgets evenly across the organization rather than directing money at high-demand roles. In a 3.5 percent budget world, even spreading means your critical roles drift below market while everything else drifts slightly above it.
Third, require a written rationale for any offer outside range. Not to slow hiring down, but to surface how often the range itself is the problem. If exceptions cluster in one function, the benchmark is stale, and that is a data purchase, not a discipline problem.
The talent market of 2022 is gone, and the market of 2019 is not coming back either. What replaced them is quieter but more fragmented, with pay moving at different speeds by industry, by company type, by clearance status, and by zip code. The organizations that win offers in this market are not the ones paying the most. They are the ones whose numbers describe the market as it is, this quarter, for this role, at a company like theirs.
RX2 Solutions is a workforce solutions firm specializing in HR outsourcing, executive search, and strategic staffing. We partner with organizations to build high-performing teams through customized talent strategies, leadership placement, and scalable workforce solutions.
📞 Phone: 610.340.3490
📧 Email: info@rx2solutions.com
🌐 Website: www.rx2solutions.com
RX2 Solutions
A Respectfully Professional People Company
Get the next one of these.
We publish one piece a week on hiring, HR, and leadership in regulated industries. Written by the people running the searches.
One email a week. Unsubscribe in one click. Your address stays with us.
Working on this problem right now?
Talk to a Managing Partner. You will hear back within one business day.
Start a Conversation