In 2015, the math for a 2.5x return on a PE deal looked like this. Put up 50% equity, finance the rest at 6-7% interest, grow EBITDA by 5% annually, and sell in five years. Financial engineering did the heavy lifting. The leverage amplified modest operational improvement into strong returns.
That math is broken.
Bain’s 2026 Global Private Equity Report lays it out clearly. Today’s deals require 30-40% more equity as a share of the purchase price. Interest rates sit at 8-9%. And to hit the same 2.5x MOIC that a 5% EBITDA growth rate delivered a decade ago, you now need 10-12% annual EBITDA growth.
Twelve is the new five. And that changes everything about how PE firms need to operate their portfolios.
The leverage era is over
For two decades, financial engineering was the primary return driver in private equity. Buy a company, optimize the capital structure, ride multiple expansion, and sell. The operational improvements were real but secondary. The math worked even when operational gains were modest.
That model depended on three conditions that no longer exist. Cheap debt. Available leverage. And a rising tide of multiples across the market.
Today, the average cost of funding for PE middle market term loans has come down from its peak, but it is still roughly double 2021 levels. Syndicated loan activity hit a record $404B in Q3 2025, which sounds healthy until you realize that overall leverage contribution to purchase prices is down significantly. The debt is available, but the terms have shifted against the buyer.
Meanwhile, 80% of GPs surveyed by Bain and StepStone expect multiples to remain flat in 2026. You cannot buy your way to a return through multiple arbitrage when the exit multiple is likely the same as the entry multiple.
What remains is operational value creation. Not as a bonus on top of financial engineering. As the entire return.
The distribution crisis makes this urgent
This would be uncomfortable in any environment. In the current one, it is existential.
Distributions to LPs sit at 14% of NAV. That is the lowest since 2008-09, and it has been below 15% for four consecutive years. An industry record that nobody wanted.
The unrealized portfolio tells the story. 32,000 companies worth $3.8 trillion are sitting inside PE portfolios waiting to exit. The average holding period at exit has stretched to seven years, up from five to six during 2010-2021. 40% of portfolio companies have been held for five years or longer, up from 29% in 2019.
IRR declines after year seven. Every additional year of holding erodes the return. The math is unforgiving.
LPs are responding. Buyout fundraising dropped 16% in 2025 to $395B. Fund closes fell 23%. Management fees averaged 1.6%, down 20% from the traditional 2%. Over 50% of LPs report they have more leverage with GPs than they did a year ago. 53% say they are limited in new commitments because of undrawn commitments from prior vintages.
The message from LPs to GPs is straightforward. Show us operational value creation. Show us distributions. Or we are not re-upping.
What 10-12% EBITDA growth actually requires
Growing EBITDA by 5% annually is achievable through organic momentum and basic cost discipline. Most competent management teams can deliver it without fundamental changes to how the business operates.
Growing EBITDA by 10-12% annually for five to seven years is a different exercise entirely. It requires systematic improvement across every operational lever. Pricing optimization. Procurement efficiency. Go-to-market effectiveness. Customer retention. Revenue quality. Margin expansion.
Every one of those levers depends on data.
You cannot optimize pricing without accurate cost data by customer segment, by product, by channel. You cannot improve procurement without spend visibility across categories and suppliers. You cannot accelerate go-to-market without pipeline data that reconciles to revenue. You cannot reduce churn without cohort analysis that shows where and why customers leave.
And you cannot do any of this if three systems give three different answers to the same question.
Why data infrastructure is now a return driver
In the 5% growth era, data quality was a nice-to-have. The financial engineering carried the return regardless. Sloppy data was a nuisance, not a deal-breaker.
In the 10-12% growth era, data quality is the foundation every operational improvement sits on. If the data is wrong, the pricing model is wrong. If the reporting takes three weeks to reconcile, the operating partner cannot course-correct in time. If the AI initiative runs on inconsistent inputs, the outputs are useless regardless of how good the model is.
GF Data’s analysis of 360 mid-market transactions since Q3 2024 shows this directly. Sellers who paired a quality-of-earnings analysis with a data quality assessment achieved 7.4x EBITDA multiples. Sellers without the data quality component achieved 7.0x. That is a 0.4x difference. On a $50M business, it is $2M.
That premium exists because buyer confidence is directly correlated with data quality. When a buyer’s diligence team can validate the seller’s numbers quickly and cleanly, the investment committee gains confidence. Confidence compresses timelines, reduces diligence friction, and protects multiples.
When the numbers do not reconcile, confidence erodes. Timelines extend. Earnouts get introduced. Multiples get adjusted downward.
The compounding window
The firms that treat data infrastructure as a first-100-days investment are the ones seeing the compounding effect play out across the holding period.
Companies that achieve operational improvements in the first 100 days post-acquisition tend to sustain those gains throughout the holding period. Companies that stumble early rarely recover momentum.
This is not abstract advice. It is math. If you invest in data infrastructure in year one and it enables a 2% margin improvement by year two, that improvement compounds through every subsequent year of the hold. On a five-year hold, you get four years of compounded benefit. On a seven-year hold, you get six.
If you wait until year three to start the data work, you get two years of compounded benefit on a five-year hold. And on a seven-year hold, you are running data remediation while simultaneously trying to prepare for exit. That is the worst possible timing.
One mid-market platform company I studied failed to establish data integration standards across its add-on acquisitions. The result was a six-month delay in value creation initiatives and $8M in forgone EBITDA. That delay did not just cost $8M. It compressed the remaining window for improvement, reduced the exit narrative, and lowered the final multiple.
What the best firms are doing differently
The firms that are generating returns in this environment share a few characteristics.
They treat data infrastructure as part of the value creation plan from day one. Not as an IT project. As an operating initiative with a direct line to EBITDA.
They fund it upfront rather than deferring it to year three when the pain becomes unbearable. The first 100 days is the window.
They hire data talent, but only after they have decided the strategy. Which metrics matter. Who owns them. What the canonical source of truth is. The hire executes the strategy. The hire does not figure out the strategy.
They standardize reporting across the portfolio so board meetings are working sessions, not reconciliation exercises. Consistent reporting dramatically accelerates capital allocation decisions and enables fund-level analytics that were impossible when every portfolio company reported differently.
And they connect data quality to the exit thesis from the start. Because in a market where 12 is the new 5, the companies that can prove their EBITDA growth with clean, defensible, verifiable data are the ones that trade at premium multiples. The ones that cannot are the ones that sit.
The new deal math in one sentence
Financial engineering used to carry the return. Now the return is the operation. And the operation runs on data.
Every firm knows this intellectually. The firms pulling ahead are the ones acting on it in the first 100 days, not the last 100.