When it comes to salary increase budgets, 2026 is looking very similar to 2025 — and perhaps even a little less.
An October 2025 survey of Canadian employers by Mercer projected average merit increases to be three per cent and total salary increases at 3.3 per cent in 2026, matching what organizations delivered in 2025. Four in five employers still plan to spread those modest dollars broadly, favouring near across-the-board increases rather than sharper differentiation for critical skills or top performers, according to the Mercer survey.
Reports by other organizations such as the Conference Board of Canada (now Signal49 Research), Gallagher, TELUS Health, and Normandin Beaudry all show 2026 salary increase outlooks slightly below 2025 levels.
At the same time, two-thirds of organizations in the Mercer survey say the economy will have a moderate to significant impact on compensation decisions, even as they prioritize skill development and market competitiveness. In addition, eight in 10 employers are still defaulting to equal distribution of salary increase budgets.
Using AI, analytics to strengthen performance data
That makes performance-based pay less about finding more money and more about using technology to make every dollar work harder, according to Yao Yao, assistant professor at McMaster University’s DeGroote School of Business. And Yao says that starts with treating analytics as a practical tool to improve performance-based pay schemes and then AI as a way to inform how organizations should strategize their talent management.
“The idea and philosophy of pay for performance (PFP) is very straightforward, and the link between pay, individual performance, and organizational objectives themselves aren’t too difficult to establish,” says Yao. “But in practice, one of the major pitfalls for PFP is an accurate and a comprehensive measure of performance appraisals — and I think this is one major area where AI and analytics can be really helpful.”
Yao’s first focus is on the quality of the performance information that’s feeding pay decisions. She says that many organizations don’t fully trust their own ratings, which makes it difficult to defend sharper pay differentiation when merit pools are tight.
“For instance, we all know that performance review forms are often too generic, and in the past, especially for large organizations, it would seem humanly impossible for HR departments to carefully curate each performance review form to customize for each single job,” she says. “But now with AI, these jobs can really be scaled very quickly.”
That, in turn, supports clearer expectations and more comparable ratings across jobs and business units, says Yao. “We know that generative AI is not a great decision-maker because sometimes it still hallucinates, but it’s really good at copying what you’re already doing,” she says.
More data points to inform pay decisions and predictions
From there, analytics can surface additional, ethically sourced data points — for example, collaboration patterns from chat tools or job-specific software usage — to give managers a fuller view of contribution, with the goal of not automating judgment but equipping leaders with consistent, defensible evidence when they carve a flat merit budget into differentiated performance-based increases, says Yao.
Even in sectors forecasting slightly higher increases, such as chemicals and high tech, Mercer’s data show overall salary projections clustered around the three per cent mark, with chemicals at four per cent and high tech at 3.8 per cent for total increases. For Yao, that makes it essential to use analytics and AI to look ahead at which skills will command pay premiums over the next decade, so scarce reward dollars are aimed at the right targets.
“So organizations, I think, really need to be thinking about how to incentivize the skill development and the competency development that are of high value to them,” she says.
Yao’s research on HR and AI points to two broad groupings of in-demand skills: those needed to build, understand, and work effectively with AI; and human and relational capabilities that are unlikely to be automated quickly but remain central to strategy, workforce planning, and client relationships. Knowing these groupings gives HR a basis for modelling which roles should see relatively more performance-based upside, even when average budgets are flat, she says.
Redesigning work, reward, and risk in the age of AI
While Yao focuses on measurement, Anna Filice, chief people officer at Ontario’s Workplace Safety and Insurance Board (WSIB), is looking at how AI will reshape jobs and what that means for pay structures and performance expectations.
“Just like predictive analytics, with AI there are endless opportunities,” she says. “But are we AI literate and are we equipped from a leadership change management perspective to manage this change? And, as HR leaders, we’re going to have to look at how we design the workforce.”
Filice says her team is concentrating on using AI to improve service delivery, building AI literacy and change capability, and rethinking how work is organized as AI agents take on more tasks in operations and support functions. That inevitably leads to questions about how traditional compensation architecture should evolve.
“You look at historical compensation structures, where you get extra points for having 10 to 20 people reporting to you — well, you might have 10 agents and no people now,” she says.
Measuring performance differently
For performance-based pay, that shift raises fundamental questions about how to value leaders who manage AI agents versus human teams, and who “owns” the transition from headcount-based to skills-focused planning, says Filice, who sees HR’s role as pivotal in that transition — not just technically, but culturally.
“HR is going to be critical to either the success or failure of the implementation of AI in any organization,” she says.
Despite the hype, Mercer’s survey suggests AI is only beginning to influence workforce plans: just one per cent of respondents cited AI and automation as a reason for reduced hiring in 2026, and only three per cent said they are proactively planning AI-related headcount changes. Filice says there’s a window to establish clear guardrails before AI-driven analytics take on a larger role in pay and talent decisions.
Boundaries and transparency
Filice emphasizes firm boundaries on sensitive data alongside “safe spaces and sandboxes” for lower-risk experimentation, including in HR’s own processes. That approach recognizes that enthusiastic early adopters will find ways to use AI regardless, so sanctioned pathways are also a risk-management strategy, she says.
For Yao, transparency is equally critical as analytics and AI begin to influence performance-based pay more directly — particularly as pay transparency rules, such as those that have come into force in Ontario, force employers to explain pay decisions more clearly.
“Performance rating training that may be only offered to managers can be made available to employees to help them understand how their performance rating was generated, and also how the technology-captured data is used,” says Yao. “Be very transparent about how they’re being used, and also open the floor to the employees to invite opinions and concerns, so that you listen to your workforce — transparency is really the key to build trust.”
Originally written by: Jeffrey R. Smith
Source: Human Resources Directior
Published on: 6 February 2026
Link to original article: Pay increases flat in 2026? Here’s how AI and analytics can move the needle
