Cracking the Retirement Profitability Puzzle

A Tale of Two Plans

Consider two 401(k) plans sitting on the same recordkeeping platform. Both hold roughly $250 million in assets. Both have about 2,000 active participants. Both generate similar revenue on paper.

By every metric that appears on a quarterly business review, they look identical.

But one of them is quietly destroying margin.

Plan A runs a straightforward safe harbor design with a single payroll feed from ADP, standard investment lineup, and minimal customization. Contribution processing runs clean. Compliance testing is automated. The plan sponsor calls once a quarter.

Plan B is a different animal entirely. It was inherited through an acquisition three years ago, still running on a separate system configuration that was never fully consolidated onto the primary platform. It has four excluded employee classes, a discretionary match formula that changes annually, two separate payroll providers feeding data in different formats, and a custom vesting schedule that doesn't map cleanly to the platform's standard rules. Every nondiscrimination test requires manual intervention. The plan sponsor's HR team calls weekly with questions the self-service portal can't answer. And every time a regulatory change hits, the development team has to implement it twice — once for each platform.

Same AUM. Same participant count. Same line on the revenue report. But Plan B costs three to four times more to administer — and nobody in the organization can quantify that with precision, because the systems weren't built to track it.

This is the retirement profitability puzzle. And most firms are solving it with intuition instead of data.

The Numbers That Should Worry You

Research from CEM Benchmarking, drawing on a database covering $1.3 trillion in DC assets, found that per-participant recordkeeping costs range from $20 to $80 per year. That's a four-to-one ratio for what is ostensibly similar service delivery.

But here's the finding that should concern every retirement technology leader: only about 25% of that cost variance can be explained by measurable plan characteristics. Scale, plan size, account balances — the variables most organizations use to price and segment their books — account for a fraction of the actual cost difference between plans.

Metric	Value
Range in per-participant recordkeeping costs	4:1
Cost variance not explained by standard plan characteristics	75%
Statistical relationship between account balances and servicing costs	0

That last number deserves emphasis. Account balances — the single most common basis for recordkeeping fee structures — have no statistically significant relationship to the actual cost of servicing a plan. The industry's dominant pricing model is, by the data, disconnected from its cost structure.

The remaining 75% of cost variance comes from factors that most recordkeepers do not systematically track at the plan level: administrative complexity, payroll integration challenges, customization depth, exception handling volume, and the operational overhead of maintaining multiple system configurations that were never fully consolidated.

The plans that look profitable on a revenue report and the plans that are actually profitable are often two completely different lists. Until you can see cost-to-serve at the plan level, you're flying blind.

Why Profitability Is So Hard to See

The retirement industry has undergone dramatic consolidation. The number of major DC recordkeepers declined from roughly 80 in 2004 to fewer than 30 by 2023. Half of the top 20 recordkeepers from 2011 have been acquired. The top five now control nearly 70% of DC assets and 56% of participants. Accenture projects that within the next decade, the top five could manage more than 75% of total market assets.

Each of those acquisitions came with a promise of scale efficiencies. And each came with a reality: inherited platforms, inherited configurations, inherited complexity.

The consolidation trap

When a recordkeeper acquires another firm's book of business, they inherit not just the plans and participants, but the entire technology stack those plans run on — often a custom-built, heavily modified system with years of undocumented configuration baked into it. McKinsey found that when a homegrown platform is involved, complexity increases materially: proprietary systems carry significant data-architecture challenges and accumulate large technical debt, especially because sellers often stop investing in systems once a sale is in motion.

The scale of these integrations is staggering. The Principal–Wells Fargo Institutional Retirement integration required standing up 140 separate IT applications and running four full dry-run migrations over six months before the final cutover. The Empower–Prudential integration took roughly two years from close to completion. The Empower–MassMutual integration required eight separate migration waves over 18 months.

McKinsey's research on retirement recordkeeping M&A found that integration teams frequently settle for workarounds in the near term — building overlays to give service representatives a single workbench, for example — while deferring an increasingly theoretical commitment to true platform consolidation, or abandoning the aspiration altogether. Harvard Business Review estimates that 70–90% of mergers fail to deliver their expected value.

The result: firms running two, three, sometimes four distinct recordkeeping platforms simultaneously. Each with its own data architecture. Each with its own batch processing logic. Each with its own compliance testing workflows. Each requiring its own specialized support staff.

Eight of the top ten US DC recordkeepers — holding 73% of total DC assets and serving 66% of participants — operate on custom-built, predominantly mainframe-based systems. And 43% of these institutions still rely on COBOL — a language introduced in 1959. (Sources: Everest Group, Profound Logic)

The hidden tax of running multiple platforms

Organizations across financial services spend 60–80% of their IT budgets maintaining existing legacy systems. Globally, financial institutions spent $36 billion on legacy technology costs in 2022 — a number projected to reach $57 billion by 2028. That's not a technology investment. It's a tax. Every dollar spent keeping a redundant platform running is a dollar not spent on automation, participant experience, or the analytics infrastructure that would let you see which plans are actually making money.

McKinsey estimates that technical debt accounts for roughly 40% of IT balance sheets across financial services. Protiviti's research found that 78% of financial services firms flagged technical debt as a barrier to innovation — higher than any other industry.

For retirement recordkeepers, this plays out in a very specific way. The operational cost of a plan is distributed across systems, teams, and processes that were never designed to be measured together. Contribution processing costs sit in one system. Call center interactions in another. Compliance testing in a third. Exception handling in someone's email. Manual reconciliation in a spreadsheet that only one person knows how to maintain.

You can see the forest, but you can't see the individual trees that contribute to that forest. And in retirement, the trees are where the margin lives or dies.

Where the Money Actually Is

McKinsey's analysis of the US retirement industry revealed a striking concentration: more than 75% of industry profits in 2023 came from just three segments — small and micro 401(k) plans, 403(b) plans, and 457(b) plans.

This means the segments that many recordkeepers pursue most aggressively — large corporate 401(k) plans with high AUM — contribute a disproportionately small share of actual profits despite holding significant assets.

The paradox is real and well-documented: as plan size grows, margins often shrink. Smaller plans tend to deliver gross margins in the mid-to-high 50% range. The largest mandates struggle to clear the low 40s. Larger plans appear attractive but often come with tighter pricing, higher service expectations, and more operational complexity.

Metric	Value
Industry profits from just three plan segments	>75%
Decline in recordkeeping fees (2014–2024)	26%
Total plan fee decline (2012–2023)	44%

Meanwhile, fees continue to compress from every direction. The revenue side of the equation is shrinking relentlessly, which means the cost side has to get smarter, not just smaller.

Even the most aggressive acquirers are finding that scale alone doesn't solve the equation. Empower — the industry's most prolific consolidator — achieved a 6% reduction in cost per participant over four years. That's meaningful, but it required completing full platform consolidation across every acquired book. Most firms never get that far.

The market is sending a clear signal. When Voya acquired OneAmerica's retirement business in 2024, industry observers noted the price was estimated at roughly one-fifth of what Empower paid for MassMutual — a sign that mid-tier recordkeepers without clear scale advantages or operational efficiency are seeing their valuations compress in step with their margins.

Everyone in this industry can tell you their revenue per plan. Almost nobody can tell you their cost per plan with any confidence. That's the gap where margin disappears.

From Chaos to Clarity

A national retirement services provider came to Convergent with a problem that had been compounding for years. Through a series of acquisitions and organic growth, the firm was operating multiple retirement platforms simultaneously — each with its own configuration standards, batch processing workflows, and compliance testing approaches.

There were no standardized production environments. No dedicated operations staff overseeing the full platform landscape. Changes were being made directly in production by individuals working in isolation, without coordinated change management. The firm's leadership knew something was wrong — service levels were inconsistent, compliance risk was growing, and costs seemed to rise with every new plan added — but they couldn't pinpoint exactly where the problems lived because no single view of operations existed.

Convergent deployed a team of five senior practitioners — people who had personally operated the same platforms this firm was running. Within weeks, not months, the team stabilized operations, established standardized change management processes, and began building a platform rationalization roadmap.

The engagement revealed what the client had suspected but couldn't prove: a significant portion of their operational cost was driven not by plan volume or participant count, but by the accumulated complexity of running unconsolidated systems. Redundant batch processing jobs. Duplicate compliance workflows. Manual reconciliation steps that existed only because two systems didn't share a common data format.

With visibility into where complexity actually lived, the firm could make informed decisions about which platforms to consolidate, which processes to automate, and where their operational investment was generating returns versus simply managing entropy.

They didn't have a technology problem. They had a consolidation problem wearing a technology mask. Every acquisition added another layer of configuration that made the next one harder to absorb.

What Getting It Right Looks Like

Solving the profitability puzzle doesn't require ripping out every system and starting over. It requires building the visibility layer that most organizations are missing — the ability to see cost-to-serve at the plan level, across platforms, and use that data to drive decisions.

Map your true operational topology. Before you can optimize, you need to see what you're actually running. How many distinct platforms? How many unique batch processing workflows? Where are the manual steps, the reconciliation bridges, the exception queues? One global financial services firm that undertook this exercise discovered it could reduce its application portfolio by 35% — saving $18 million annually.

Build plan-level cost attribution. The retirement industry has invested heavily in plan-level revenue visibility. It's time to build the cost side with equal rigor. What does contribution processing actually cost for a plan with four payroll feeds versus one? What's the compliance testing cost for a plan with seven excluded groups versus zero?

Stop treating platform consolidation as optional. McKinsey found that integration teams regularly defer true consolidation in favor of workarounds. That's understandable in the short term. But every year of deferred consolidation compounds the operational cost and makes the eventual migration harder. Everest Group found that a cloud-first core platform approach can halve the technology and operations cost base, yielding approximately 20% annual savings.

Invest in operational observability before AI. The industry is racing to adopt AI. But AI requires clean, connected, well-understood data. If you can't see your operational cost structure today, adding machine learning on top won't help. Build the observability infrastructure first — then use AI to accelerate what you can already measure.

Bring domain expertise into the room. Platform rationalization and cost-to-serve analysis are not exercises that general-purpose technology consultants can lead effectively. The difference between a vesting schedule that maps to standard configuration and one that requires custom code is the difference between a profitable plan and a margin drain.

The Cultural Shift

For most retirement organizations, profitability has historically been measured at the enterprise level: total revenue, total cost, total margin. Growth was measured in plans won and participants added. Success was AUM and headcount.

That worked when fees were higher, competition was lower, and the technology landscape was simpler. It doesn't work anymore.

Fee compression means there is less room for operational inefficiency. Consolidation means every acquired book brings complexity that must be actively managed. Regulatory expansion — particularly SECURE 2.0 — means compliance costs are rising, not falling. And the talent market means the people who understand legacy platforms are harder to find every year.

The cultural shift is a move from "How many plans do we administer?" to "Which plans are we making money on, and what drives the difference?"

Organizations that make this shift gain the ability to price with precision, allocate resources to their highest-return plans, identify which operational investments actually reduce cost-to-serve, and make acquisition decisions with clear-eyed understanding of what integration will actually require.

Organizations that don't will continue to grow the top line while watching margins erode — never quite able to explain why scale isn't producing the efficiencies it should.

The firms that win the next decade won't necessarily be the biggest. They'll be the ones that know, with precision, what it costs to serve every plan on their platform — and have the operational discipline to act on that data.

The Path Forward

The retirement profitability puzzle is not a mystery. The data exists. The operational complexity is identifiable. The path to plan-level cost visibility is well understood by organizations that have walked it.

What's missing, in most cases, is the combination of domain expertise and technical execution required to see it clearly — and the organizational will to act on what the data reveals.

The industry has spent the last decade consolidating. The next decade will be defined by which firms turn that consolidation into genuine operational advantage — and which ones continue to carry the hidden cost of complexity they never fully resolved.

The puzzle is solvable. But it requires looking at the pieces that most organizations prefer not to examine.

---

Sources & References

CEM Benchmarking via NAPA Net (2024) · McKinsey & Company — "The US Retirement Industry at a Crossroads" (2024) · McKinsey — "From Diligence to Delivery: Capturing Value from Retirement Record-Keeping M&A" (2019) · McKinsey — "Tech Debt: Reclaiming Tech Equity" (2020) · Accenture — "Reinventing Retirement Recordkeeping" (2023) · Everest Group (2020) · Evalueserve (2024) · CAPTRUST (2024) · PLANSPONSOR / NEPC (2025) · PLANADVISER (2024) · Principal Financial Group (2021) · Protiviti (2023–2024) · Harvard Business Review / McKinsey M&A Failure Rate Analysis (2016–2024) · WealthManagement.com (2025) · Profound Logic (2024) · Mechanical Orchard (2024) · SPARK Institute — Voluntary API Framework (2025) · Brady Martz — SECURE 2.0 Implementation (2024).