Your Raise Might Already Be Decided, and a Person Hasn’t Looked at It Yet
Sign Up For Daily Job Market News
Your next raise might get decided by a computer before your manager ever weighs in. The Wall Street Journal’s Callum Borchers reports in his “On the Clock” column that more companies are using AI tools to scan job boards for what competitors pay in similar roles, then benchmarking every employee against that number. Pay-transparency laws are making the job easier for these tools, since more job postings now list a real salary range instead of staying silent on pay.
The raise pool itself isn’t moving much. A Marsh survey of 1,001 U.S. employers found the average planned merit increase for 2027 sitting at 3.2%, inside a total raise budget of 3.5%. The Conference Board’s separate survey landed on a nearly identical number. Both figures track closely with what employers paid out in 2024, 2025, and 2026, so 2027 is shaping up to be another year of modest, unremarkable raises on average. What won’t be modest is the spread between the largest and smallest individual raises, as more of that flat budget gets steered toward specific people instead of spread evenly across a team.
Stello AI is one of the companies selling this kind of tool. It pulls pay data from job postings, sites like Glassdoor, and payroll processors, then puts every employee at a new client through a comparison against the market. The tool flags underpaid high performers as the first people worth fixing, since they’re also the most likely to leave for a recruiter’s call. It flags the opposite group too: people who are overpaid, often because they changed jobs at exactly the right moment and landed a package today’s slower market wouldn’t support. Those workers should expect a small raise, or none.
Companies are also planning to pay up for a specific mix of skills going forward: AI fluency alongside leadership, judgment, and the kind of people skills that are hard to automate. Some employers are rethinking the review process itself, weighing pay for outcomes and impact instead of pay for hours logged.
With roughly 3.5% to go around and a manager deciding who gets what share of it, a bigger number for one person on the team means a smaller one for somebody else. The question worth asking isn’t just “Did I do good work?” it’s “Where do I sit in the pay range for my role?” because a strong year won’t buy much of a raise if you’re already near the top of that range.
I’d also push back a little on how much weight these AI benchmarks deserve. Job postings often list a wide salary band that covers several levels at once, so a tool reading that range can spit out a number that looks precise and isn’t. Treat the AI output as the opening bid in a pay conversation, not the final word, and let the manager who knows the employee’s work finish that conversation.
The retention logic in this story matches what shows up on the staffing side too. An underpaid high performer is the person most likely to take that recruiter’s call, and closing the gap on their pay almost always costs less than replacing them costs later. If your team has raise money to work with this cycle, that’s the group to fund first, not the group to spread evenly across. If your comp structure needs a closer look before your 2027 planning locks in, talk to us about hiring and we’ll help you figure out where the real gaps are.
Staffing your team doesn’t have to be hard.
Reach out and see how we can help.
Layoffs Just Hit Their Lowest Point Since Mid-July
The same week employers are debating how AI should shape a raise, the government’s read on job security came in about as good as it’s been all year. The Labor Department reported 197,000 initial unemployment claims for the week ending September 19, down 1,000 from the prior week and 22,000 below the same week last year, when claims stood at 219,000. It’s the lowest weekly total since the week ending July 18, when claims came in at 189,000. The 4-week moving average, which smooths out weekly noise, slipped to 202,250.
Continuing claims, the count of people still collecting benefits, came in at 1,719,000 for the week ending September 12, up a modest 2,000, after the prior week’s figure was revised down 13,000, from 1,730,000 to 1,717,000. The insured unemployment rate held steady at 1.1%. On the state level, Kentucky posted the sharpest increase, adding 1,041 claims tied to manufacturing layoffs, while California, Texas, New York, and Michigan posted the biggest declines, with New York and Michigan both pointing to fewer layoffs in transportation, warehousing, healthcare, and manufacturing. New Jersey continues to hold the highest insured unemployment rate in the country at 2.3%. Initial claims from former federal employees also fell, to 362 from 635 a year earlier.
Layoffs are about as low as they’ve been in months, and that’s worth sitting with for a second. Employers aren’t cutting people at any real scale right now, which tells you companies that trimmed headcount earlier this year have mostly found the staffing level they want. Manufacturing is the one thread I’d keep pulling on. Kentucky’s jump this week and Michigan’s improvement point in opposite directions inside the same industry, which reads less like a national trend and more like the pressure moving from one plant, and one state, to the next.
For employers, the people you want to hire are mostly employed already, and a low layoff rate means they have no urgent reason to leave. If you’re trying to pull someone away from a stable job, you need a specific, better reason for them to move, and you need to make that offer fast once you find the right person. If you’re the one job hunting in a market like this, see what’s open right now instead of waiting for layoffs to open up a bigger pool of competition-free openings, because that pool isn’t forming right now.
Holiday Hiring Keeps Shrinking, and the Big Names Have Gone Quiet
Retailers are about to post their second straight year of shrinking holiday hiring. Challenger, Gray & Christmas forecasts 450,000 seasonal retail jobs for the fourth quarter of 2026, down from 461,500 last year. Last year’s total was already the smallest holiday hiring season since 2008.
Shoppers haven’t pulled back. August retail sales rose 1.2%, and back-to-school spending held up fine. Retailers are leaning on automation, their existing staff, and on-demand worker pools before they post a single new seasonal opening, which explains why the announcements so far are thin. Spirit Halloween is hiring 52,000 people for its biggest push in company history. Michaels plans to bring on more than 10,000 seasonal workers. Bass Pro Shops and Cabela’s are holding a national hiring event on October 7 and 8 targeting 4,000 hires. Amazon, Target, Bath & Body Works, and Kohl’s, each of which has posted a consistent seasonal number for years, haven’t announced anything yet for 2026.
The warehouse and shipping side of the holiday season is following the same script. Transportation and warehousing employers added 266,500 jobs in the fourth quarter of 2025, down 12% from 2024 and the lowest total since 2018. Challenger expects roughly the same again this year, and UPS, FedEx, and USPS haven’t put out targets either. Overall retail employment sat at 15.47 million people in August, up just 63,000 from a year earlier.
Retailers are still staffing the holiday season, just differently than before. A company keeps a pool of on-demand workers ready and squeezes more hours out of current staff before it ever posts a seasonal opening, so the job seeker refreshing job boards for a “now hiring 100,000” headline may be waiting for an announcement that simply isn’t coming this year. The most useful number in this whole report is buried in the Michaels detail: a large share of last year’s seasonal hires there turned into permanent roles, which means a holiday job at a company like that is often a real audition, not a dead end. Anyone taking a seasonal role this year should treat it that way starting on day one.
For employers, the risk sits on the shipping and warehouse side of the business. When peak volume finally hits and the automation and on-demand pools come up short, every company in the region ends up chasing the same small set of available workers at the same time. Lining up backup staffing now, while people are still easy to reach, beats trying to find them in the middle of the rush. If your seasonal plan needs a partner who can move fast when volume spikes, explore our staffing services and get ahead of it instead of scrambling once the season is already underway.
The Skilled Trades Are Overworked and Undervalued at the Same Time
A new World Economic Forum white paper puts a name to the people who build, install, and operate the physical parts of the economy: “makers.” This group covers electricians, welders, machine operators, assemblers, farmworkers, cleaners, and food prep workers, among others, and the report argues this workforce is being stretched thin while employers quietly downgrade how much its skills matter.
Nearly 1 in 4 U.S. job postings, 24.1%, mention maker skills, according to Indeed data cited in the report. This share climbs to 72% in supply chain and transportation, 67% in manufacturing, 64% in infrastructure, and 60% in automotive and aerospace. Makers make up roughly 24.5% of all employment in North America. Demand looks set to keep climbing too: of 28 skills the report studied, constructing and installing ranks first for jobs projected to grow through 2030, and more than 40% of the roughly $7 trillion in global data-center investment expected by 2030 is projected to land in the U.S., a buildout that depends directly on this same workforce.
The strain is already visible in the payroll data. ADP figures cited in the report show hourly makers logging about 5 times as much overtime as other hourly workers, and at the peak, more than 6.5 times as much. Despite that, only 4.8% of North American employers describe the physical and sensory abilities behind this work as “core” to their business, and more expect those abilities to matter less by 2030 than more, a net reading of negative 24.6%, the most negative of any region in the report.
The pay data cuts the other way entirely. Among U.S. makers, every 10-point gain in physical and sensory ability is tied to about $2,500 more in annual pay. Digital skills pay even better, adding roughly $3,500 per 10 points, well above the under-$2,000 bump digital skills add for other workers. Makers with the strongest digital skills make roughly twice what those with the weakest digital skills make. Industrial machinery mechanics have a median pay of $63,760, compared with $34,220 for food prep workers.
The pipeline behind all of this looks thin. Worldwide, only 13% of boys and 4% of girls expect to work in a maker job. In North America, women hold about 21% of maker jobs against roughly 55% of other jobs, the widest gap the report found anywhere. Cost is part of the story: a four-year U.S. degree averages $79,200 in tuition and fees, against $15,200 for a trade school certificate.
If the numbers in this report hold up once it’s published in full, the contradiction at its center is hard to miss. North American employers are working their hands-on workforce nearly 5 times harder on overtime than everyone else, while also being the most negative employers in the world on whether that workforce’s skills will matter in a few years. Both of those things can’t stay true for long. If you manage a plant, a warehouse, or a construction crew and your overtime sits at 5 times normal, that’s usually a sign your headcount plan is short, and the people covering the gap are the ones most likely to burn out or leave for whoever’s hiring.
For job seekers weighing a trade program, the digital skills data is the clearest signal in the whole report. A maker who can run diagnostics, operate programmable equipment, or read a monitoring system makes close to double what someone at the bottom of the digital scale makes, so choosing a program that teaches the technology alongside the tools is worth the extra effort. For employers, a resume can’t show you whether a candidate can troubleshoot a machine or read a sensor readout. Build a practical task into the interview and watch someone do the work instead of guessing from a list of past titles. If your maker workforce is stretched as thin as this report suggests, schedule a free consultation with our team before the overtime line turns into a resignation line.
