Your Next Raise May Be Whatever the Model Says You’re Worth
The average raise next year is still about 3.5%. The range around that midpoint is getting wider, and a model is helping draw the line.
By Sean Anees Saifi · Capital Wealth · Published Thursday, October 1, 2026 · Source: The Wall Street Journal, Monday, September 28, 2026 edition, whose market figures are the Friday, September 25 close (Personal Journal)
Key Points
Compensation surveys suggest the average pay increase next year will be about 3.5%, in line with recent norms, Callum Borchers writes, but the range of raises around that midpoint will be wider than usual.
A Marsh survey of more than 1,000 U.S. employers found the average planned merit increase for 2027 is 3.2%, the lion’s share of a total raise budget of 3.5%. A Conference Board survey was nearly identical.
More firms are using AI tools to comb job boards for competitors’ salary ranges, an exercise made easier by pay-transparency laws. Third-party companies now offer AI salary benchmarking as a service.
Stello AI uses public listings, forums such as Glassdoor, and data purchased from payroll companies. CEO Amee Parekh told the Journal the firm tells clients how much to pay — and flags high performers who are underpaid as the first raises to fix. It also flags workers who are overpaid after hopping jobs into a hotter market.
Lesley Uren of Korn Ferry Consulting says firms are wrestling with whether billable-hour pay should shrink if AI cuts the time a task takes, and whether sales pay should rise if AI finds more leads. One vision: tie pay more closely to impact and outcome, and de-emphasize hours as a proxy for productivity.
3.5%
average planned total raise budget for next year (surveys)
3.2%
average planned merit increase for 2027 (Marsh, 1,000+ employers)
1,000+
U.S. employers in the Marsh raise survey
Wider range
around the 3.5% midpoint — the model helps pick who sits where
The benchmarking file used to be a drawer. Now it answers in milliseconds — and it decides the merit line.
In one line: Raises are still clustering near 3.5%, and AI benchmarking is widening who gets the top of that range — and who gets told they’re already paid enough.
A lot of us worry AI will take the job. On the Clock’s Callum Borchers says it is already in the raise. Compensation surveys put next year’s average increase around 3.5%, in line with recent norms. The range around that midpoint is getting wider. Whether you land high or low could come down to what a machine-learning model thinks you’re worth. Pay-transparency laws made scraping the going rate easier. Third-party firms now sell the scrape as a service.
Stello AI is one of them: public listings, Glassdoor-style forums, payroll-company data. When it takes a new client, CEO Amee Parekh told the Journal, one of the first jobs is to assess every employee and recommend pay. High performers who are underpaid get first claim on the raise budget. Another category: people who job-hopped into a hotter market and are now overpaid relative to today’s tepid one. Parekh says she’s never seen someone fired just for making too much — replacing them is expensive — but they should brace for a modest raise, if they get one at all.
Hours were the proxy. Impact is harder to meter
A Marsh survey of more than 1,000 U.S. employers found a 3.2% average planned merit increase for 2027 inside that 3.5% total budget; the Conference Board came in nearly the same. Firms will pick and choose. Professional-services shops that still live on billable hours are stuck on a nastier question: if AI cuts the time a task takes, does the paycheck shrink? If it finds more sales leads, does the commission grow? Lesley Uren of Korn Ferry Consulting told the paper clients are wrestling with what good looks like now — tying pay to outcomes instead of hours, which is messier because impact can be intangible. Tauseef Rahman at Marsh says a decent manager will tell a well-paid worker they’re already at the top of the range, and talk promotion path instead of a fat merit bump.
Our read
Cash Flow (M5): a 3.5% midpoint is not a plan. If the household budget assumes a raise that a model can withhold, the budget is the model’s. Keep proving the work, yes — and keep a cash buffer that doesn’t need next year’s merit line to clear. If you’re the underpaid high performer, the paper’s point is that some firms now have a list with your name on it. If you’re the overpaid hopper, the list exists too.
Don’t build a lifestyle on a range you haven’t seen. Ask how pay is set this cycle, and size the emergency fund as if the midpoint is what you get. Fifteen minutes with a statement beats arguing with a bot about what you’re worth.
What It Means For Your Portfolio
Hold — budget off the midpoint, not off the model’s favorite
The average raise is still about 3.5%, and AI benchmarking is deciding who sits above that line and who is told they’re already paid enough.
General planning principles, not advice for anyone in particular. A household budget that needs a top-of-range raise to work is a budget the employer can rewrite. Size recurring costs off the midpoint, and keep cash that doesn’t depend on next year’s merit letter.
If pay is being set by a benchmark you can’t see, ask how the cycle works and what ‘top of range’ means in your shop. Then fund the gap yourself — skills, a promotion path, or a smaller lifestyle — rather than waiting for the model to notice.
