Fund Administration Transition: Moving or Insourcing, and the Data Work to Staff on Contract

Last updated: November 5, 2026

By Khurram Tehseen, FSA, Founder and Principal at Operating Alpha

Staff the parallel run with two contract reconciliation accountants and one data engineer who loads the history, and plan at least one full quarter of shadow books. As of October 2026, at KORE1 Contract Rate Index 2026-Q4 national modeled medians, that team costs about $147,800 for three months.

Here is a break log. Invented numbers, real shapes.

Forty lines. The biggest is $62,100 and the smallest is $212. Nothing there is a typo. Each one is a rule applied one way in the outgoing administrator’s books and another way in the new ones, and the answer sits in the credit agreement or the fee schedule, not in whichever number happens to be larger.

Take the $62,100. The outgoing administrator stepped the management fee base down on the day the investment period ended. The new books stepped it down at the next quarter-end, because that is how the date got keyed into the fee table. Who is right? Probably the outgoing set. Unless one investor’s side letter says otherwise, in which case both are wrong for that investor. Somebody has to read the documents, write down the answer, and fix whichever side is off. Then thirty-nine more.

That work is the subject. When a credit fund changes administrators, or brings fund administration in-house, someone has to produce the second set of books, compare the two, and close every difference before the first set goes away. This piece covers who should do it, how long it runs, and what it should leave behind. KORE1, whose site you are reading, places contract accountants and data engineers for this work through its contract staffing practice, so weigh the staffing advice accordingly. The method works the same if you fill the seats from your own bench.

What Is a Shadow Period, and Why Run Two Books With One Record?

A shadow period is a stretch of month-ends in which the new books, whether a new administrator’s or your own team’s, are produced from the same source records as the outgoing administrator’s, while the outgoing set stays the official one. Nothing is signed from the new books until every difference has a reason and a name.

Some shops call it shadow accounting, others a parallel run. Same thing. Two books, one record. The record is the set of facts underneath both: every draw, repayment, rate reset, fee event, capital call, and distribution since the fund started. Two books built from one record should agree. When they don’t, one of three things happened. A rule differs. A fact is missing or wrong in one copy. Or someone made a judgment on one side that never got written down.

The third costs the most. At most funds I have seen, one person at the administrator is the database, and everybody knows the name. That person knows that one investor’s side letter caps the management fee. That a borrower elected to pay in kind two years ago and the term sheet on file still says cash. That the rate floor in the second amendment applies to the delayed-draw tranche and not to the term loan. Nobody wrote it down. None of it is in the data handed to the new team, and the new books won’t know any of it until a break drags it into the open. A break only appears in a month you are actually producing.

A loan carries more rules than a share does, which is part of why I expect a credit fund to throw up more of these than one holding listed equities. Day counts, reset dates, paid-in-kind toggles, unused fees, discount amortization. Each is a place where two competent teams can book the same loan differently. Your investors read the result, too. The ILPA reporting template asks for fees and expenses in a standard layout, and your books have to be able to fill it, which is one reason the fee tie-out earns its own person for the length of the shadow period. The pack that goes out is built from these books, and I cover that end in LP reporting data infrastructure.

Which Seats Does a Parallel Run Need?

Three. Teams tend to think of them in the wrong order.

Two reconciliation accountants. Not because the volume is heavy, but because whoever prepares a tie-out should not be the person who signs it off, and a three-month engagement leaves no time to find that out in month three. One takes the portfolio side: positions, accruals, cash, and the bridge from the outgoing NAV to the new one. The other takes the investor side: capital accounts, fee calculations, and allocations. They review each other. Senior level, since the questions go to the credit agreements and the partnership agreement, and an accountant who can’t read an interest section sends every break straight back to you. That is the profile KORE1 places through its contract accounting staffing.

One data engineer, and this is the seat I see left off. Somebody has to pull the history out of the outgoing administrator’s files, which usually means spreadsheets, PDF statements, and one system export in a format nobody documented. Then map the old chart of accounts and the old security and borrower identifiers onto the new ones, load all of it, and prove the load, with counts and sums by period, loan by loan and investor by investor. Borrowers are the awkward part. The same company turns up under three names across the administrator’s records and your own loan system, which is the problem I wrote about in the same borrower in five systems. The first thing this person builds is never a dashboard. It is the break log, kept as a table you can count by cause every Monday, so you can tell whether forty lines are heading for thirty or for sixty.

The third seat is not a contractor. It is yours. A named employee, usually the controller or the head of fund operations, decides which side is right when the documents are ambiguous, and signs the tie-out. Contractors bring the evidence. The decision is the fund’s. Give that person the administrator relationship too, meaning one name on their side and an agreed turnaround for break questions, because the outgoing team’s attention drifts once notice has been given.

Why contract at all? A parallel run ends. Three permanent hires for a three-month tie-out leaves three people in April with the books already matched. KORE1 partner Devin Hornick made the budget point in an email: “Companies may not approve three or four permanent hires, but they may absolutely approve the resources or project team necessary to complete a $300K or $500K initiative that has already been funded.” A funded transition is a project. It gets a project team.

If you are insourcing, the seats are the same with one addition to the plan. Your own hires need to run a quarter-end before the contractors leave, so their start date matters as much as the contractors’ does. If you are moving to another administrator, that firm does its own onboarding tie-out, and your contractors become the fund’s independent check on it. Different job. Same books.

How Long Should the Overlap Run?

At least a quarter. Three month-ends, and one of them a quarter-end, because that is where the valuation marks, the fee true-ups, and the investor statements land, and a shadow period that stops before the quarter closes has not seen the hard part.

The rule for stopping is separate from the calendar. Two consecutive month-ends where every remaining break has a reason and a name, the later of the two being a quarter-end. Miss that and you keep going, even if the plan said June. Plan three months. Contract for four.

Here is the usual quarter. My experience, not a benchmark.

The assignment data lines up with that. Across 1,887 ended KORE1 contract assignments, the median ran 116 days, a little under four months, with a quarter of them ending inside 44 days and a quarter running past 227. A three-month run sits comfortably in the middle of that range. The 44-day end is the one to watch, because a reconciliation seat that rolls off in week six has seen one month-end, and the second is where the pattern shows.

The date that decides it is a contract date. Your administrator agreement has a notice period and, most likely, a transition-assistance clause, and the day the old books stop being the record is written there, not in your project plan. Read it first. If notice runs out two weeks after the first month-end, you don’t have a parallel run. You have a rehearsal. And if the cutover lands in the quarter that closes your fiscal year, the auditors will want to know which books they are auditing, a conversation better had in September than in January.

What Breaks If the Parallel Run Ends Before the Books Tie?

You carry the break forward as an opening balance. Capital accounts roll from one month to the next, so a fee-basis difference like the one in the break log, left open in month two, sits in every investor’s balance in month three and in the statement that goes out with it. Then the contract ends. The person who knew about the side letter moves on to another client. Months later an investor asks why a balance moved in November. Nobody can say. Restating a statement is a letter nobody wants to write. The auditors, meanwhile, ask which set of books they are relying on.

The ones I would worry about hide in the months you did not shadow. A valuation mark loaded one notch off. A rate reset on the 29th that lands in the wrong period. A fee accrual that ties in a month with no capital activity and breaks the first time somebody calls capital.

Stopping early feels like progress. The old administrator stops invoicing, the project plan goes green, and a quarter later the explanation for one number costs more than the three months you saved.

What Do the Contractors Leave Behind?

A folder your own team can run from, or the new administrator can be held to.

• The break log, as data. Every difference, its cause code, who owned it, and how it closed.
• A register of special treatments. This is the person-who-is-the-database written down on paper: the side-letter fee cap, the payment-in-kind election, and the floor on one tranche. Read the record, not the recollection of whoever last touched the file.
• Mapping tables from the old chart of accounts and identifiers to the new, borrower matches included.
• Load scripts and checks that can rerun a closed month and land on the same pack, which matters the first time the new administrator’s system changes under you.
• The access record, meaning who saw which fund and which fields, and the day each account ended. The design for that is in contractor access to regulated financial data.
• One page of sign-off, drafted for the employee who owns the decision to edit and sign.

The rerun matters most. Proving a process on a month that has already closed, scored against the real pack, is how you earn trust before anyone is asked to rely on it. I use the same logic in a month-end close replay, and a shadow period is that exercise run forward, in real time, on live months. What a data contractor owes you at the end of a longer engagement is laid out in the 90-day deliverables. If you are insourcing, treat this folder as the training manual for the people you hire next.

Questions Before the Parallel Run Starts

Realistically, what does a contract senior accountant cost for the reconciliation seats?

About $77 to $94 an hour nationally, median $87, per KORE1’s modeled 2026-Q4 index cell. As of October 16, 2026, a 3-month, 40-hour-a-week contract at the median runs about $45,300, and six months about $90,600. Those are bill rates, so wages, payroll taxes, insurance, and the staffing firm’s own recruiting and account management all sit inside the hourly number. New York runs higher at about $56,700 for three months. Chicago and Dallas-Fort Worth run lower. Boston sits between.

Should the outgoing administrator help with the parallel run?

Yes, in writing, with a named contact and an agreed turnaround on every break question. The outgoing team holds the record and knows the special treatments, and its incentive to answer quickly fades once notice is given. Ask for a data dictionary, three closed months of source files, and one person who owns your questions. Then keep the reconciliation itself on your side of the table, using their output as an input. An administrator should not mark its own homework. Nobody does it well.

How many quarters of history should be loaded?

Two fiscal years of monthly ledger history, plus the full life of every open loan and every investor. The ledger history gives the auditors and the investor statements something to compare against. Since-inception detail is what the fee and carry calculations stand on, and a shortened investor history can misstate balances that look right until the first waterfall runs. Closed loans can come in as summary balances unless the waterfall needs them. That is my default, and your partnership agreement and your auditor can change it.

Who reconciles the books when a credit fund changes administrators?

The new administrator reconciles during onboarding, and the fund’s own controller owns the acceptance. Contract accountants add independent capacity on the fund’s side, which matters because an onboarding team has a go-live date it is measured against. When fund administration moves in-house there is no second party, so the contractors are the second set of eyes until your own hires have run a quarter-end.

Can contractors see investor data in the shadow books?

Only through a named account per contractor, scoped to the funds and fields the work needs. A tie-out needs amounts and dates far more than it needs identities, so the mapping work can run on investor codes instead of names and tax identifiers. The fuller design is in access design for regulated financial data.

Count the Lines You Cannot Map

Take the last month-end pack your administrator sent and the trial balance you would load into the new books. Lay them side by side and count, in ten minutes, the lines you cannot match to each other on sight. That count sets the number of seats and the length of the overlap far better than a vendor’s timeline does. Near zero? One accountant and a short run may do it. Near forty? Now you know.

I am happy to look at a break log if you have one. Connect with me on LinkedIn.