More than half of PE-backed MSPs lost over 10% of their workforce last year. Not through layoffs. Through people leaving.

That’s the number buried in ConnectWise’s Service Leadership annual compensation report, which Channel Dive surfaced this week. The finding: 54.6% of PE-owned managed service providers experienced workforce churn exceeding 10% in 2025. Compare that to 22.8% for bottom-half MSPs without PE backing. The spread is not a rounding error.

This is the tax the industry is paying for a decade of roll-up mania. And it’s about to get worse before it gets better.

The Standard Response Is Wrong

When churn data like this comes out, the conversation defaults to “how do we retain people.” Retention programs. Counter-offers. Flexible work policies. Maybe a sabbatical benefit.

That framing treats a structural problem as a people problem. It isn’t.

The churn is a symptom of what PE-backed MSPs actually are: businesses optimized for exit multiples, not for delivering service. When you buy 12 MSPs and bolt them together, you get one entity with 12 different cultures, 12 different toolsets, 12 different compensation philosophies, and 12 sets of employees who joined a company that no longer exists. The people who had options left. The people who stay are the ones who couldn’t find something better — which is not the talent base you want running your service delivery. The MSP talent myth is that you can pay less and keep the same quality of work; the reality is more nuanced.

Service Leadership’s data makes the underlying economics explicit: the most profitable MSPs pay less than their peers and experience more churn. They’ve decided that’s acceptable. They’ve built hiring and training pipelines to backfill constantly, rather than paying to retain.

That’s a choice. And for a while, it works. Right until it doesn’t.

AI Is Accelerating the Bet

Here’s what’s new: AI is making the churn-and-replace model look smarter than it should.

The Service Leadership report notes that more MSPs cut headcount (43.5%) than grew it (36.4%) in 2025 — a reversal from 2024, when growth was more common. The interpretation offered is that firms are “leaning on AI to backfill open roles rather than rehiring.” That’s almost certainly true for a subset of the market.

If you can automate L1 ticket resolution, alert triage, and patch management — and the reality of AI in PSA and RMM tools is getting closer to that — you need fewer technicians. You can absorb churn without replacing headcount. The model holds together.

But let’s run the math the other way.

The same report shows that MSPs in the top profitability quartile lost more than 10% of their workforce at a rate of 38.6%. Not just PE-backed ones. The highest-margin MSPs are also high-churn. What that tells you is that the firms extracting the most margin are doing it partly by not paying people what the market says they’re worth. The AI automation story is real, but it’s also a cover story for compensation arbitrage.

That works until your clients notice. And they’re starting to notice.

What’s Breaking

Three things are converging that make this model unstable in ways it wasn’t two years ago.

First, the clients have learned to ask the question. Sophisticated buyers — enterprises, healthcare organizations, financial services firms — are now asking about employee turnover in MSP contract negotiations. They’ve seen what happens when the engineer who knew their environment leaves and gets replaced by someone reading from a runbook. That knowledge doesn’t transfer cleanly.

Second, the AI tools are not yet good enough to replace mid-level talent. L1 automation is real. L3 troubleshooting is not replaceable by any tool currently in production. The firms betting on AI to flatten their org charts are making that bet on a 12-month timeline when the underlying capability might be 36 months out.

Third, the exit math is getting harder. PE sponsors need to sell these businesses, and buyers are running their own diligence. A platform MSP with 54% annual churn across its PE-backed cohort raises questions that weren’t being asked when valuations were at 10x-12x EBITDA. With multiples compressing, quality of operations matters more.

The Firms That Will Win

They’re not the ones with the slickest automation stack. They’re the ones that figured out which roles are genuinely automatable and paid to retain the ones that aren’t.

The Service Leadership report notes that small firms — those with $1M to $2M in revenue — gave top-level raises to more than 30% of their employees on average. They’re protecting against manager attrition specifically, because they know they can’t afford to lose the people who carry institutional knowledge.

That’s the right instinct, even if it’s expensive. The human knowledge of how a specific client’s environment works, why a specific configuration decision was made in 2023, what the client’s CTO will accept and what they won’t — that’s not in any ticket system. That knowledge lives in people. When the person walks out, the knowledge walks out.

PE-backed platforms are betting they can systematize enough of that knowledge to survive the churn. Some will be right. Most won’t.

The firms watching this data and staying independent — or selling carefully, to buyers who actually care about service delivery — are sitting on a competitive advantage they may not fully recognize yet. The counterargument, of course, is that the channel needed PE’s money and many founders had no better option.

Client retention follows employee retention. That math hasn’t changed. And as the MSP profitability paradox shows, the firms that look best on paper aren’t always the ones best positioned for the long term.