Financial Wellness in America
Annual Report

In nearly two decades of doing this research, we’ve never seen anything quite like what happened in 2025. Two major employee segments moved in sharply opposite directions by almost exactly the same amount. Workers under 30 with household incomes below $60,000 saw high or overwhelming stress fall 2.4 percentage points. Workers aged 30 to 44 earning between $60,000 and $100,000 saw it rise 3.2 points. Because those two forces are so close in magnitude, they cancel each other out in aggregate data. An employer looking at their system-wide numbers this year would see modest, unremarkable movement and have no idea that one group is quietly improving and another is in a deepening crisis.

What strikes me about this year’s data is how clearly it reflects something we’ve been watching build for a long time. It’s always been shaped by where people are in their lives, what resources they have access to, and whether the institutions around them — employers, benefits systems, financial services — are actually designed for their reality or for someone else’s. What’s different now is that the divergence is sharp enough, and the data is specific enough, that it’s very hard to look away. We have one cohort of workers genuinely improving their financial lives, and another that is falling further behind despite earning more and being further along in their careers. That gap doesn’t close on its own. It closes when employers stop measuring success at the aggregate level and start asking harder questions about who they’re actually reaching.

I hear this in coaching sessions constantly. And it really does come down to what Liz often talks about: income doesn’t solve complexity. It’s not “I don’t know what to do.” It’s “I know what I should do in each of these areas, but every decision I make in one area affects three others, and I don’t know where to start.” Nearly half of this cohort (46%) are caregiving for children or aging parents, compared to 36% of the workforce overall. Only 28% feel on track for retirement. And a lot of them are carrying student loan debt on top of a mortgage on top of all of that. There’s no single benefit that was designed to untangle that kind of complexity. Which is kind of the whole point.

And what the data shows is that this isn’t a marginal problem. Among 30- to 44-year-olds earning $60,000 to $100,000, squarely middle class, stress jumped from 40.3% to 43.5% in a single year. That’s the steepest rise of any segment we tracked. These aren’t employees who are struggling to make ends meet in the traditional sense. They’re employees who are making ends meet and still drowning a little. And most of the standard benefits employers offer were built for a different set of problems than the ones this cohort is actually facing.

This is the part of the research I find most significant, because it moves the conversation from diagnosis to direction. For years, the dominant question in enterprise AI has been about efficiency: how do we do more with less, how do we automate what was manual. And most organizations are still in that mode. What this data shows is that a different question is now getting answered in financial wellness: can technology actually make people’s lives better in a meaningful, measurable way? And the answer, based on what we’re seeing across millions of employee interactions, is yes. But it happens in a specific way. The programs producing the strongest outcomes across both cohorts are the ones that pair AI with human expertise rather than treating them as alternatives. That combination is what makes personalization possible at scale. And that’s a fundamentally different model than what most of the benefits industry is still operating under.

The AI plus human finding is the one I keep coming back to, and what makes it more compelling is that it’s consistent with what consumers are telling researchers independently of our data. Corporate Insight recently surveyed over 2,000 U.S. adults on their attitudes toward AI in financial services and found that two-thirds already assume their provider is using AI — among Gen Z, that number jumps to 89%. So the question employees are asking isn’t whether their employer is using AI. They’ve already decided the answer is yes. The question is whether they trust it. And on that front, 45% believe AI is being used to replace human support, and 58% have abandoned a chat session the moment they realized there was no human available. That’s not apathy. That’s a trust problem. CI’s conclusion maps directly onto what we see in our outcome data: consumers aren’t rejecting AI’s assistance, they’re rejecting autonomy without accountability. Across 27 of the 28 milestones we tracked, employees who combined AI coaching with a human CFP® professional outperformed those using AI alone by wide margins. Emergency savings of $1,000 or more: 22% with AI only versus 36% with AI plus a human coach, a 64% improvement. On track for retirement: 38% versus 53%. Eliminating high-interest debt: 42% versus 58%. The employees who made real progress weren’t the ones who used the platform most. They were the ones who combined it with a human coach. That’s the trust gap showing up directly in the outcome data.

From the coaching floor, this makes complete sense. AI is genuinely great at a lot of things. It’s available at 11pm when someone is lying awake anxious about a financial decision, it surfaces information quickly, it tracks progress over time. What it’s not good at is holding the whole complicated picture of someone’s life and making a judgment call about which problem to tackle first when three of them are urgent and all of them are connected. The 40-year-old dealing with a parent’s care costs, a mortgage, two kids, and a retirement account that never feels like enough: that person doesn’t need a faster chatbot. They need someone who can sit with the whole situation and say, okay, here’s where we start. AI gets them in the door. The human is what makes it count.

So this one is really illustrative. Two employers, same retirement content, completely different approaches to delivery. One promoted their webcast through standard marketing channels and got 5.6% attendance. The other identified employees over 50 who were facing a voluntary separation decision and sent them a targeted invitation at the exact moment that decision was sitting in front of them. Their attendance rate was 72.4%. Same content, thirteen times the engagement. The only variable was context: whether the guidance arrived at a moment when it felt genuinely relevant to something the employee was already dealing with.

What I love about this finding is that it reframes the whole engagement question. The industry has spent years asking how to get employees to care about financial wellness. But I don’t think that’s actually the problem. Employees care plenty when the guidance shows up at a moment when it’s relevant to something they’re already thinking about. A benefit election, a dependent update, a major life decision. Those moments are already in every employer’s system. The opportunity is just connecting them to the delivery before the moment passes.

What this finding really confirms is that the delivery of financial wellness is as important as the design of it. Most employers have spent years investing in content — calculators, webinars, modules — and wondering why utilization stays flat. The 13x gap isn’t a content problem. It’s a timing problem. When guidance arrives at the moment an employee is already facing a financial decision, it stops being a benefit they have to remember to use and starts being something they actually need right now. That shift, from broadcasting to meeting people where they are, is where the real gains are. And most employers haven’t made it yet.

Start by looking at your own data differently. If you’re evaluating your financial wellness program by overall engagement rates or aggregate stress metrics, there’s a real chance you’re missing what’s actually happening underneath. Pull it by age band and income segment. See if you have two cohorts moving in different directions, because if what we found at the national level is present in your workforce, your headline number is hiding it. That analysis probably takes an afternoon, and what you find will tell you more about where to invest than anything else.

And then make sure there’s actually a human available, not as a premium add-on, but as a real part of the program. Among employees who came in with high or overwhelming stress and kept engaging, 61.8% reduced that stress to some or none. Over one in three eliminated high-interest debt entirely. Those are meaningful outcomes. And the employees who did best weren’t just using the platform. They were talking to a person. That’s not a coincidence.

What this research points toward is a model that works across every cohort — not because it delivers the same thing to everyone, but because it delivers the right thing to each person at the right moment. That’s what AI plus human coaching makes possible at scale. AI removes the friction: it’s always available, it meets employees wherever they are, it surfaces what’s relevant to their specific situation. The human coach provides the depth: the judgment, the relationship, the ability to hold someone’s whole financial picture and help them navigate it. Together, that combination is what mass personalization actually looks like in practice. Not a platform everyone has access to. A genuinely personalized experience that scales. For Gen Z, it looks like the right tools at the right moment. For Millennials, it looks like a human who can sit with the complexity and help them start somewhere. The benefit is the same. The experience is built for each of them. That’s the standard employers should be building toward.