Skip to content
← Writing

Ledger

Issue No. 023 · Execution

What really happens when a fund changes administrators?

The risk isn't the decision to switch. It's the parallel-run window where two systems of record both claim to be authoritative.

By Owen E. H. Meyer · December 19, 2025 · 5 min read

Switching fund administrators gets treated as a vendor decision. There's an RFP, a bake-off, reference calls, a signature. Most of the attention goes into that process. None of it is where the risk actually lives. The risk shows up after the decision, in the handoff itself, when the fund's books exist in two places at once and nobody has fully worked out what happens if they disagree.

What a parallel run actually requires

Once a fund selects a new administrator, the two firms don't perform a clean handoff on a single date. They run in parallel for a stretch of time — anywhere from a few weeks to several months, depending on the fund's complexity and how much historical data needs rebuilding. The outgoing administrator keeps producing NAVs and investor reporting exactly as before, while the incoming administrator independently builds its own books: loading historical transactions, setting up the chart of accounts, reconciling its output against what the outgoing administrator is producing.

LPs are meant to notice nothing — statements arrive on schedule, NAVs look continuous. That continuity is the point of running two systems side by side instead of cutting over in one step. But it obscures what's happening underneath: two independent sets of books, on two different platforms, both claiming to represent the same fund.

TWO BOOKS. ONE FUND. ONE QUESTION OF AUTHORITYThe parallel-run window between administrators.Outgoing administratorcontinues producing NAVs and reportingIncoming administratorbuilds books independently, reconciles against outgoingoverlap window — which system is authoritative?LEDGER
Simplified example. Overlap duration varies by fund complexity.

Where it breaks

The failure mode isn't the new administrator making an error — errors get caught in review, that's what reconciliation is for. The failure mode is structural: at some point during the overlap, a number from the outgoing system and the corresponding number from the incoming system won't match, and there's no default answer for which one is right. Both systems were built in good faith from the same underlying records. Neither becomes authoritative simply because it produced a number first.

For a few weeks, the fund has two books and one truth. Nobody's sure yet which is which.

The question isn't whether the numbers will disagree. It's who has authority when they do. Name the tiebreaker before the parallel run starts, not after the first discrepancy surfaces. Decide in advance whether the outgoing administrator's numbers govern until final cutover, or the incoming administrator takes precedence once its books pass a defined reconciliation threshold — in writing, with a shared escalation path for breaks that can't resolve at the working level.

The contract changes administrators. The parallel run changes the system of record.