How to Choose a Fund Administrator: Fund I vs Fund II vs Fund III

Almost every guide to choosing a fund administrator is written as though there were one answer. There are three, and they are not refinements of each other. A first fund is buying a price it can compute and a setup that can issue a capital call within weeks. A second fund is buying answers to an LP diligence questionnaire, an audit that closes on time, and a structure that can hold a co-invest vehicle without a second set of books. A third fund is buying consolidation across entities, controls evidence an institutional operational due diligence team will accept, and the ability to run more than one administrator without losing the thread.

What moves between those three is not the feature list. It is who is responsible for the numbers, how many separate legal entities have to tie together at quarter end, and who is asking. The single most expensive mistake available to a first-time manager is buying the third-fund answer at the first fund: an enterprise engagement with a multi-month implementation, priced for a structure that will not exist for six years. The most common mistake is the opposite one, carrying a first-fund setup into a third fund because nobody wanted to run the migration, and discovering the limit during an audit rather than before one.

Read the row for your stage, then read the row above it. The point of the table is that the criteria you can safely ignore today are the ones that decide the next selection, which is why the switching section at the end of this page matters more than any feature comparison.^1,4

The people asking this question are not the funds the large administrators are built to serve, and the filings say so plainly. In the second quarter of 2026 there were 5,844 new Form D filings from pooled investment funds, of which 2,457 identified themselves as venture funds. That is up 6.5% on the prior quarter and up from 1,667 in the same quarter of 2025. Fund formation is not slowing; it is fragmenting.

The size distribution is the part that matters for this decision. Of those 2,457 venture filings, 2,033 stated a defined offering amount, and the median of those was $560,000. Not $56M. Five hundred and sixty thousand dollars. Of the filings with a defined offering, 95.7% were under $100M. The modal new venture vehicle in the United States is a single-deal SPV or a sub-$5M first fund, filed by one or two people with no finance staff, and the count of them has risen every quarter for two years.

Set that against how fund administration is actually sold. Almost nothing in the category publishes a price, which means a manager raising a $3M first vehicle cannot tell in advance which providers will even take the call, let alone what the answer costs. Outsourced administration is generally quoted as basis points on committed capital with an annual minimum, and at these sizes the minimum is the only number that binds: a basis-point rate applied to a $5M fund produces a fee no provider will accept, so the floor is the price. The observed band for a $10M to $50M fund is $25,000 to $75,000 a year, which is a real number and, for most of the 2,457, a larger number than the fund's entire annual management fee.

This is why the stage framing is not a stylistic choice. A guide that answers the question with a shortlist of institutional administrators is answering a question almost none of this cohort is in a position to ask. The honest Fund I answer starts from what can be priced and operated by one person, and moves up from there as vehicles accumulate.^3,4

Advisers registered with the SEC disclose their fund administrators. Question 26 of Form ADV Part 1A, Schedule D Section 7.B.(1) asks, for every private fund, whether the fund uses an administrator other than the adviser, and if so, names it. We read those blocks for the 100 firms in the VC Tech 100 panel on 21 September 2026. Eighty-one of them report private funds at all, covering 2,506 funds between them. Two of the 81 panel entries resolve to a single adviser filing, so the honest count is 80 distinct advisers.

Across the 81, 56 name at least one outsourced administrator and 25 file no administrator record on any of their funds, which the form renders as 'No Information Filed' on every block. That second group is not a data gap; it is the disclosure a firm makes when it administers its own funds. Those 25 filers hold 872 of the 2,506 funds, so roughly a third of the venture funds in this panel are administered by the manager rather than by an outside firm.

Bucketing the 81 filers by how many funds they report turns that headline into something a manager can use. The distribution is not a smooth gradient. Outsourcing is near-universal at the bottom of the panel, dips in the middle where the oldest single-strategy firms sit, and returns at the top with a different shape.^1,2

A first fund has one real constraint, and it is not cost. It is that nobody is going to do this work except the general partner, a fractional controller a few days a month, and a CPA once a year. Every criterion that matters follows from that, and every criterion that does not matter fails the same test: it describes a problem the fund does not have yet.

Start with a price you can actually compute. This is the single most underrated selection criterion at this stage, because a number you can put in a model is worth more than a lower number you cannot. Two providers in this market publish a rate. Archstone, our own platform, publishes $297 a month, which is $3,564 a year regardless of fund size, and covers fund accounting and NAV, capital calls, whole-fund waterfall math, K-1 distribution and an LP portal; we build it, we sell it, and we say so on every page where it appears. AngelList publishes a formula rather than a price: 0.15% of fund size plus $20,000 a year on its Full Service tier, which on a $10M fund works out to $35,000 and on a $50M fund to $95,000, and which behaves nothing like a flat fee as the fund grows. Everyone else quotes, which means your budget line is a guess until a sales conversation ends.

Second, time to the first capital call. A first fund usually has a signed LPA, an anchor commitment and a deal it wants to do, in that order and within the same eight weeks. An implementation measured in months is not a feature trade-off at that point; it is a reason the deal does not close on your paper. Ask for the date the first call can be issued, in writing, counted from the signed engagement letter rather than from an internal kickoff.

Third, one operator. Whatever you buy, someone has to run quarter-end, and that person has a name. If the answer is 'the GP, between board meetings', the requirement is a system one person can operate without a professional-services engagement, and the honest failure mode of this stage is a platform bought, half-implemented and quietly abandoned in favour of a spreadsheet. If the answer is 'an outside administrator', the requirement is a named point of contact rather than a shared service pool, and you should ask what happens when that person leaves.

Fourth, K-1 timing, as a date rather than an intention. Late K-1s are the most reliable way to damage LP trust in a first fund, because they are the one deliverable an investor's own accountant is waiting on. Ask for the target delivery date, ask what happened last year, and ask who chases the portfolio-company information the return depends on.

Fifth, the LP portal, judged narrowly. At this stage it needs to hold statements, capital account balances, call notices, tax documents and the fund's documents, and it needs to look like a fund rather than a file share. It does not need a CRM, a data warehouse or custom report building. Investors in a first fund are forming a view about whether you are organised, and a clean portal is disproportionate evidence that you are.

Sixth, whether you can leave. Ask for the export format and the transition-assistance obligation before you sign, while you still have leverage, and get both in the engagement letter. An outgoing provider is most needed at precisely the moment it has least reason to help you, and the shape of the data you get back is what determines whether the next migration takes six weeks or six months.^4,5

The second fund is where the selection criteria change most sharply, and where most managers discover that the first decision was made against the wrong rubric. Nothing about the first fund's operations has broken. What has changed is that someone is now checking, and that there are two of everything.

The first new criterion is the LP diligence questionnaire. A second fund is usually raised from a partly new investor base, which is the first time a manager encounters operational due diligence as a written exercise rather than a conversation. The questions are specific and they are about your administrator: who produces the NAV, who has authority to move money, what segregation of duties exists between the person who approves a wire and the person who records it, who runs anti-money-laundering and know-your-customer checks on each LP and what evidence is retained, and whether there is a controls report. A manager self-administering a first fund can answer all of those honestly and still lose the allocation, because the answer an institutional allocator wants is that somebody other than the GP produces the numbers the LPs rely on. That is the independence argument, and it is not about your arithmetic.

The second is audit coordination, which is the first genuine test of an administrator rather than a platform. An auditor's first question is not what your NAV is; it is how you can prove it. That means an immutable, timestamped trail behind every capital account entry, every valuation change, every call and every distribution, plus a party who assembles the confirmation package and answers the auditor's follow-ups directly rather than forwarding them to you. Ask any candidate how many audits they supported last year, which firms, and whether they deal with the auditor directly. A provider that positions audit support as a special project rather than a standing service is telling you where your quarter-end will go.

The third is the first SPV or co-invest vehicle, and it arrives earlier than most managers plan for. A deal too large for the fund's concentration limit, an anchor LP who wants more of one company, a follow-on with no reserve left: each produces a separate legal entity with its own capital accounts, its own carry and its own tax return. This is the point where per-entity pricing stops being a footnote and becomes the largest variable in the quote. Ask what the marginal cost of the fifth vehicle is, not the first, because an SPV programme is where administration costs actually surface.

The fourth is carry and waterfall correctness, which quietly becomes a different problem when there are two funds and a vehicle. The first fund's waterfall was whole-fund and straightforward. Now there is per-deal carry in the SPVs, a management fee offset to track, and the possibility of distributions from Fund I while Fund II is still calling capital. Waterfall errors are the expensive category of error because they are errors about who is owed money, and they tend to surface years later, in front of the people they shortchanged.

The fifth is the migration cost of leaving the first fund's setup, which is the thing nobody prices. If Fund II is going on a new system, the honest options are to migrate Fund I as well, which is a real data-migration project, or to run two systems in parallel for the remaining life of Fund I, which means two quarter-end processes, two report formats and two reconciliations for six or seven years. Most managers discover this after choosing. The cleanest version of the Fund II decision is made during Fund II formation, deliberately, with the Fund I migration priced as part of it.¹

By the third fund the question has inverted. The first fund asked what it could afford; the third asks what its LP base and its structure require, and then finds a way to pay for it. Three things drive the selection at this stage, and all three are visible in the Form ADV data.

The first is institutional operational due diligence, conducted properly. Endowments, foundations, funds-of-funds and the larger family offices run administration as a pass or fail item, and the requirements are concrete: an independent administrator producing the books, a service-organisation controls report the allocator's team will actually read, role-based access and a maker-checker step on anything that moves money, a documented valuation policy applied consistently under fair-value standards, and the ability to produce LP-specific reporting formats rather than one statement for everyone. The ILPA reporting template is the usual reference point for capital-account and fee disclosure. A provider that cannot supply a controls report, or that treats bespoke LP formats as a professional-services engagement each quarter, becomes the reason an allocation does not happen.

The second is multi-vehicle administration, which is a different job from administering three funds. A fund family at this stage typically holds a main fund, an opportunity or growth vehicle, a set of SPVs, sometimes a feeder for non-US investors and a co-invest structure, each accounted for separately and reported consolidated. What breaks first is not the accounting; it is the reconciliation step that lives in one person's head. Ask candidates to describe how they consolidate across entities in one ledger, how equalisation is handled across multiple closes, and how side letters with bespoke fee terms are applied without a manual adjustment each quarter.

The third is the split-administrator option, which the filings show is what firms at this scale actually do. Of the 18 panel firms reporting 31 or more funds and naming an administrator, 10 name more than one. Insight Partners reports 186 funds and names Gen II Fund Services on 39, Standish Management on four and Morgan Stanley Fund Services on two, with the remainder naming none. Sequoia Capital reports 61 funds and splits them between Standish on 48 and Linnovate Partners on 17. General Catalyst reports 49 and names Standish on 23 and HedgeServ on five. Index Ventures reports 32 and names EFG Wealth Solutions on 25, Standish on four and Intertrust on three. NEA reports 32 and names Standish on six and Aduro on one, with 25 funds naming no administrator at all.

Read those rows carefully and the pattern is not a best-of-breed strategy. It is history plus geography: a legacy provider on older vintages, a specialist on the offshore or Asia-domiciled vehicles, a third name on something acquired or inherited, and a tail of entities small enough that nobody migrated them. Splitting has a real cost, which is that no single party can produce a consolidated view of the family without a manual step, and firms that avoided it generally avoided the structural complexity that causes it rather than managing it better.

The fourth option at this stage is in-house administration, and the data is the argument against attempting it early. The 25 panel entries that name no administrator on any fund cover 872 funds between them, and twenty-one of the 25 report more than ten funds. Counting the two Andreessen Horowitz entries once, the firms are Andreessen Horowitz, Accel, Battery, Lightspeed, Founders Fund, Kleiner Perkins, F-Prime, Redpoint, Benchmark, Flagship, Canaan, Spark, IDG, CRV, DCM, Meritech, Highland, Greylock, ARCH, Partech, Matrix, Menlo, Third Rock and Viola. These are the oldest brand names in the asset class, running finance functions built over decades. In-house administration in venture is what a multi-decade fund family grows into, not a cost decision a third fund makes.^1,2

Switching is a project, not a purchase, and the cost is measured in weeks and risk rather than in fees. What moves is the whole historical record: per-LP capital account balances since inception, the full contribution and distribution ledger, the state of the waterfall, valuation history with its support, and the reporting history your investors will compare the first new statement against. If the first statement from the new provider does not tie to the last statement from the old one, the LP who notices will not care whose fault it was.

Switch for the right reason. The defensible triggers are structural: the provider cannot serve the fund family you now run, an anchor LP requires independence you cannot currently provide, K-1s or quarterly statements keep arriving late enough to damage investor trust, or the service model changed underneath you because the provider was acquired. Switching on price alone almost never pays inside the first year, because implementation, parallel running and internal time routinely exceed the fee saving that motivated the move.

Time it to the calendar rather than to your patience. The window is after the annual audit closes, when the books are reconciled and the valuation support is already assembled, and before the next close. Never move during K-1 season, never mid-audit, and never in the weeks around a close, because those are the three periods when an error is both most likely and most visible. For most venture funds that leaves a usable window between late spring and the end of summer.

Run the migration as a project with a named owner, and plan a parallel quarter. Two to four months is realistic, with at least one full quarter produced in both the outgoing and incoming systems and reconciled line by line. The parallel run is the step that gets cut when the timeline slips, and it is the single most common reason a migration produces an LP-visible error. Get your auditor comfortable with the incoming provider's reporting output before you commit rather than after the first statement goes out; their comfort with the audit trail is a gating item, not a formality.

Negotiate the exit before you need it, which in practice means negotiating it at the start. Transition assistance and data-export obligations belong in the original engagement letter, when you still have leverage, and the export format is the specific leverage point: ask what you get back, in what shape, and how long the outgoing provider is obliged to answer questions about it. Then tell your LPs before they notice. A short note naming the new provider, the reason, the date and the portal they will use prevents the conversation where an investor discovers the change from a statement that looks different.⁴

The same provider will answer all of these, and the answers should change depending on which list you read from. Take the list for the fund you are raising, and ask for everything in writing. The pattern worth watching for is not a wrong answer; it is a provider who will answer in conversation and not on paper.^1,4