IFAC published research in March 2026 with a number worth pausing on. More than 1,000 accountancy firms worldwide have taken private equity investment in the past decade, and fewer than 200 initial investments facilitated nearly 900 subsequent transactions.
The Shape of the Money
That ratio is the whole model. A sponsor buys a platform, then the platform buys everyone else.
The US picture matches. The Cornerstone PE deal tracker, reported by CPA Trendlines in February 2026, counted 22 private equity backed accounting transactions in 2023, 65 in 2024 and 104 in 2025, with more than 250 recorded since 2019 and 52 distinct sponsors active since 2024. The top ten sponsors account for just under half of all deals.
So it is not a broad market of many buyers making one acquisition each. It is a small number of platforms making a lot of them, quickly, and the pace has been rising every year.
For a firm owner reading this, one of two things is true. Either you will get a call, or you are already competing with a firm that took one.
What the Model Assumes
The investment case is not complicated and it is not sinister. Buy at one multiple, build something bigger, sell at a higher one. Between those two events, the return depends on margin, and in professional services margin comes from doing the same work with fewer hours.
Which means standardising delivery across the group. One methodology, one tech stack, one way of running an engagement, so that the fortieth acquisition is cheaper to absorb than the fourth.
That plan is written before the first deal closes. It survives contact with practice management software, which does migrate. It survives contact with a chart of accounts.
What Actually Arrives With Each Firm
Then there is the part nobody underwrites.
Every acquired firm arrives with a document estate. Twenty years of engagement letters, workpapers, correspondence, prior returns and client files, distributed across a portal the firm bought in 2019, a shared drive organised by a bookkeeper who left, an email archive, and a room with actual cabinets in it. The naming conventions are local. The folder logic is local. The reason a particular client's file sits where it sits is usually a person, not a rule.
You can migrate that in the sense of copying the bytes. What does not migrate is the ability to find anything in it.
And underneath the files sits the part that was never filed at all: why this client's entity structure looks the way it does, what was agreed verbally in 2021, which position the firm has taken consistently for a decade and why. That knowledge belonged to the partner who signed the deal, and the deal usually starts that partner's countdown.
Why This Is the Binding Constraint
Watch what the platform wants to do next. Cross sell advisory into the acquired client base. Move clients onto the group's service model. Reprice. Every one of those moves requires knowing the client well enough to have the conversation, and the acquiring firm's knowledge of that client is exactly zero on day one.
So the group does the only thing available. It routes the question back to the acquired firm's partners, who are the search index. That works, at a cost, until it does not, and the moment it stops working is roughly the moment the earn out ends.
The same logic runs in reverse for firms that stay independent. A platform can outspend you on software and out-recruit you on salary. What it cannot buy quickly is a partner who remembers what happened with a client three years ago. That is a real and defensible advantage, and it has an expiry date attached to a specific human being unless the firm makes it something the firm holds rather than something a person holds.
The Shape of the Fix
Treat the document estate as an asset with a stated condition, the way you would treat a receivables ledger. If the answer to what did we do for this client in 2022 requires a phone call, the asset is not in the condition anyone assumed during diligence.
MetaWurks is built for that half of the problem. It ingests each client's returns, statements, contracts, workpapers and correspondence, whatever conventions they arrived under, and lets someone query the whole set in plain English, so a question about a client acquired last quarter is a question rather than an archaeology project. Role based access controls decide who can open which client's records, which matters more when several firms are becoming one, audit logs record who opened what and when, and documents ingested into the platform are not used to train models or exposed to other users.
It does not merge two firms. It removes the assumption that the only working index of a client relationship is a partner who is now three years from the door.
The deals are the visible part. Nearly 900 of them have already happened, and each one handed somebody a filing cabinet they did not design.
Join the Conversation
If your firm were acquired tomorrow, how much of what you know about your largest client exists anywhere other than in your own head?