The Fund I LP Playbook: Commitment Records to a Ranked Prospect List

This playbook is for emerging managers raising a first or second venture fund, and for the platform partners and angels who help them. It turns a vague target of "family offices, foundations, fund-of-funds" into 200 to 300 named LP prospects. Each row cites a public commitment record and states a check size, an emerging-manager stance, a named decision-maker, and a tier set by written rules. The result is a list you can defend line by line to a co-GP or placement adviser. A purchased spreadsheet of names that have never backed a Fund I gives you nothing to defend.

What a defensible Fund I LP list contains

A defensible Fund I LP list has 200 to 300 rows, and every prospect on it has earned its place with a dated, public commitment to a manager like you. Each row names the allocator, the person who decides, the record that proves appetite, the typical check, the emerging-manager stance, and a tier.

A few terms, defined once:

• LP (limited partner): the investor who commits capital to your fund.
• Emerging manager: definitions vary by allocator. Hamilton Lane's covers institutional Funds I to III below $2B, which is far larger than most venture Fund Is.
• Commitment record: a public document that ties a named LP to a named fund. Examples are a pension's board minutes, a fund-closing announcement, or a foundation's tax schedule.
• Delegated program: a program where a pension hands capital to a fund-of-funds or fund-of-one manager, and that manager picks the underlying funds.
• Tier: one of four labels: anchor candidate, core, long shot, or drop. Every row gets one, set by the same written rules.

Capture every row in the same shape. Copy the schema below into your sheet's header row before you open a single filing. Fields you leave empty then show up as gaps, not as memory.

lp_name:
entity_type: single-family office / multi-family office / RIA / trust company / foundation / pension / fund-of-funds
entity_check: adviser registration lookup result; client language on the firm's own site
record_type: board minutes / board packet / closing announcement / 990-PF line 13 / Form ADV
record_citation: document title, page, date
record_date:
comparable_manager: fund name and fund number (I, II, III)
record_status: approval / commitment / deployment / year-end exposure
amount_or_value:
typical_check_range:
em_stance: dedicated program / delegated via fund-of-funds / ad hoc / unknown
route: direct / via named delegated manager
decision_maker: name, title, date confirmed
tier: anchor candidate / core / long shot / drop
tier_rule_fired:
second_source:

Which public records name LP commitments

Pension board minutes and packets are the richest public record of LP commitments, because they name the LP, the fund, the amount and the meeting date in one place. Foundation tax filings reach smaller LPs but show only year-end value. Adviser filings never name LPs at all.

Pension board minutes and packets

These records give you the strongest evidence. One set of Philadelphia Investment Committee minutes records a recommended $15M commitment to Crescent Credit Solutions Fund IX. Connecticut's meeting materials record a recommended commitment of up to $250 million to a fund-of-one managed by GCM, to be invested over a 3-year term. CalPERS publishes every GP and vehicle in its private equity portfolio, with vintage years, performance and unfunded commitments.

What a pension record proves: this allocator committed this amount to this fund at this meeting. When it lies: a recommendation or approval may never close. Hold approval, commitment and deployment as separate statuses.

Foundation Form 990-PF, Part II line 13

Line 13 ("Investments - other") needs a schedule listing each investment held at year end. The schedule shows book value, whether the holding is carried at cost or end-of-year market value, and fair market value in column (c). It does not record the commitment amount or date. Many foundations collapse everything into one line. The Kellogg trust's schedule for one tax year shows a single "PRIVATE EQUITY FUNDS" line at $1,179,378,735.

What it proves: the foundation held this exposure at the end of that tax year. When it lies: a lumped line looks like venture appetite but names no manager.

Form ADV Schedule D 7.B.(1)

The adviser files this about its own funds. It lists the fund name, the GP's name, and a flag that is true when the private fund is a fund of funds. That flag makes it a way to find fund-of-funds managers. It never names LPs, and advisers may list a fund by code instead of by name.

Records I have not verified

I have not verified the related-person fields on Form D, and I have not covered university endowment reports here. If you want endowments on the list, read each endowment's published annual report and apply the same rule: a named fund or nothing.

Which LPs back first-time VC funds, and at what check size

Family offices, wealthy individuals and fund-of-funds are the LP types emerging managers name most often, and institutional capital trails behind. Pension emerging-manager programs exist, but their gates assume much larger funds, and much of their capital flows through delegated managers.

In Citrin Cooperman's survey of emerging managers, 70% cited family offices as a source, 64% cited wealthy individuals and 43% cited fund-of-funds. Those numbers count how often a type was named, not how much capital it supplied. No survey with a published method gives a capital-weighted split by LP type. Some secondary sources say roughly 70% of Fund I capital comes from family offices and high-net-worth individuals, but they give no method, so do not build a plan on that number. The same survey found only 35% of private equity and venture respondents got allocations from emerging-manager programs.

One family-office database reports that 16% of its more than 2,700 offices are active in VC funds and 42% invest directly in startups. A family office that backs startups directly is not necessarily a fund LP. Ask for a fund record.

Emerging-manager program gates

These gates were built for buyout-sized funds, not for sub-$50M venture Fund Is. A gate that says "1st to 3rd fund" tells you nothing about fit until you also check the size gate, whether there is a venture sleeve, and the minimum check.

The route matters more than the gate. CalPERS commits to GCM-managed fund-of-funds vehicles (DEM I, II and III). It has also committed about $2B to emerging managers directly, outside fund-of-funds, in a single fiscal year. Connecticut used a GCM fund-of-one. In both cases a GP pitching pension staff for the delegated portion is knocking on the wrong door.

Decision timelines by LP type have not been established publicly. Log the date of your first meeting and the date of each decision on every row, so that by Fund II you have your own numbers.

How to find family offices that invest in emerging managers

Single-family offices are invisible to adviser filings, so you find them through their people and through co-LP mentions in comparable closings, not through entity databases. Under the SEC family office rule, an office is excluded from adviser registration if it has no clients other than "family clients", is wholly owned and controlled by family clients, and does not hold itself out to the public as an investment adviser. Such an office files no Form ADV.

The SEC states that the exclusion applies only to single-family offices. A multi-family office is a different entity type and may be registered. The difference changes how you approach it, so the entity check in step 6 is not optional.

Refolk's index shows how the decision-maker population differs by market:

The query covered profiles with a CIO, Director of Investments, Head of Private Investments or Investment Director title whose headline mentions "family office". The shares come from samples of 25, so read them as direction, not precise proportions. The US total may sit at the query ceiling, which is why the ratio is a floor.

Two practical consequences follow:

• Title filters are market-specific. In the US sample, Chief Investment Officer dominates. In the UK, the sample splits evenly between CIO and Investment Director. If you reuse a US-tuned filter in London, you lose about half the people who decide.
• Employer names hide. Four of the top employer entries in each sample are generic: "Family Office" in the US and "Single Family Office" in the UK. You cannot match those rows to an entity until you cross-check location and co-investment mentions.

This is the slow part of the job: working from people back to a confirmed entity and a record. Refolk lets you ask for the person in plain English across LinkedIn and the open web, so the search step takes minutes. The verification still belongs to you.

The step-by-step procedure

The procedure runs in eight steps in this order: settle the offering exemption, gate the programs, harvest the three record types, confirm the people and entities, then tier and sequence. Steps 3 to 5 can run in parallel if you have more than one researcher. Steps 6 and 7 cannot start until those harvests are done.

Who usually owns each step:

• Steps 1 and 2: the GP, with fund counsel on step 1.
• Steps 3 to 5: an analyst or a platform partner.
• Step 6: the GP or a researcher, because judging the entity type takes judgement.
• Steps 7 and 8: the GP, ideally checked by a co-GP who did not build the rows.

A note on time. No source I trust gives an hour count for each stage, so I won't make one up. Log your own hours per step on the first 50 rows and project the rest from that. For step 6, a people search in Refolk cuts the time spent finding names, but the time spent checking registration and client language stays the same, and it should.

How to tier each prospect

Tier on two questions: how strong the evidence is, and how well the allocator fits a Fund I of your size. A row with strong evidence but poor fit gets rerouted or dropped. A row with good fit but weak evidence gets a second-source search before anyone contacts it.

Apply the rubric below exactly as written. Change the wording only when you write down your own thresholds, such as how far back a commitment can be and still count, and then hold every row to them.

ANCHOR CANDIDATE (all must be true)
- Named, dated commitment to a Fund I or Fund II manager with a comparable strategy
- Typical check range covers the anchor-sized check you need
- Route is direct: the record shows this entity decides, not a delegated manager
- Decision-maker named and confirmed at the entity

CORE (all must be true)
- At least one named commitment record to an emerging or venture manager
- Typical check range covers your minimum check
- Entity type confirmed and decision-maker named

LONG SHOT (any is true)
- Evidence is exposure only: a 990-PF value or a lumped line
- Single co-LP mention with no second source
- Employer anonymised and not yet matched to an entity

DROP (any is true)
- No named record after a second-source search
- Program gate excludes your fund size or fund number
- Entity is not an allocator (law firm, accounting practice) or its type is wrong and nothing supports the corrected type
- Capital is delegated: drop the pension as a direct target and add the delegated manager as its own row

How this goes wrong: failure modes and false positives

Most bad LP lists fail in one of eight predictable ways: they mistake exposure for commitment, gates for fit, keywords for entities, or approvals for capital. Each failure has a cheap check. Run them before tiering, not after a wasted meeting.

1. Lumped 990-PF schedules. What it looks like: a "PRIVATE EQUITY FUNDS" line worth $1.2B that reads as venture appetite but names no manager. Check: drop the row unless a named fund appears or a second source confirms it.
2. Treating 990-PF values as commitments. What it looks like: year-end fair market value entered as a check size or vintage. Check: record the tax year and label the value as exposure, never as a commitment.
3. Reading program gates as Fund I fit. What it looks like: a "1st to 3rd fund" gate paired with a $500M fund size, filed as a match for a $25M venture Fund I. Check: confirm that a venture sleeve exists and that the minimum check fits your fund size.
4. Listing delegated programs as direct. What it looks like: a row pitching CalPERS staff when the capital flows through GCM. Check: does the board packet name a fund-of-funds or fund-of-one manager? If it does, that manager is the row.
5. Family-office false positives. What it looks like: Refolk's index surfaces "Mostyn Law Firm/Family Office", "Edmunds Family Office & Accounting Services" and "The 1911 Trust Company" under family-office keywords. Multi-family offices also slip in, because the exclusion covers single-family offices only. Check: run an SEC adviser registration lookup (a registered firm is not a single-family office) and read the client language on the firm's own website.
6. Keyword drift. What it looks like: an "emerging manager" search that returns emerging-markets portfolio managers and off-topic titles such as "Emerging Wal-Mart Suppliers". In Refolk's index, a US search on "emerging manager" with investment titles returned 54 profiles. The top title was Portfolio Manager (13 of 25 sampled), and the top employers were asset managers (Allspring, Ashmore, Nuveen), not pensions. Check: require a pension, endowment or fund-of-funds employer type.
7. Anonymised employers. What it looks like: a strong-looking CIO profile whose employer is just "Family Office". Check: cross-reference location and co-investment mentions before you assign an entity, and keep the row at long shot until you have.
8. Counting approvals as closings. What it looks like: a board approval recorded as capital in a fund. Check: keep approval, commitment and deployment as separate fields, and give tier credit only for commitment or deployment.

How many LPs you need, and how to sequence outreach

You need 200 to 300 named prospects, because published conversion from first conversation to commitment runs at a few percent at most. Sequence the work so that relationships with your strongest anchor candidates start well before you go to market.

The figures conflict, and only the Citrin Cooperman number comes from a survey. The others are practitioner estimates. One practitioner's own record was 847 LPs met for 14 commitments, a 1.7% rate. One illustration has 300 targeted LPs producing 30 to 45 responses. Plan for the low end. At 1.5% to 2.4%, 200 conversations yield roughly 3 to 5 commitments, and you won't get 200 conversations from 200 rows. That is why the list floor is 200 even for a small fund.

Order of contact

Practitioner sources say to build relationships 12 to 18 months before going to market and to pre-market 3 to 6 months before launch. One source says to start with 5 to 10 anchor LPs rather than broad outreach. Others imply building the full list first. My view: build the full list first so that your anchors are chosen by the rubric rather than by familiarity, then open conversations with the anchors while the core tier is verified.

On 506(c): SEC staff guidance in a no-action letter allows accredited-investor verification to rest on minimum investment amounts of $200,000 for natural persons and $1 million for entities, and binding capital commitments count towards the minimum. Commissioner Hester Peirce cited $169 billion raised annually under 506(c) against $2.7 trillion under 506(b), so most managers still choose 506(b). Ask counsel what the current position is. Don't rely on a summary.

Keeping the list current

An LP list goes stale as each new record is published. Pensions publish new packets for every meeting, foundations file a new 990-PF every tax year, and programs rewrite their policies. Tie each refresh to the publishing cycle of its source, not to a calendar reminder.

• Pension rows: re-read the board packet for each meeting of every eligible program. Move a row's status from approval to commitment to deployment only when a record says so.
• Foundation rows: when a new 990-PF appears, update the tax year and value, and check whether a lumped line has started naming funds.
• Program rows: re-check gates whenever a policy document is revised. The NYC Comptroller publishes an emerging-manager report, and CalPERS publishes portfolio detail, so both can be re-read on their own cycles.
• People rows: re-confirm the decision-maker before every first contact. Rerunning the same plain-English query in Refolk shows title and employer changes faster than rechecking each profile by hand.