The GP Deal-Attribution Standard: Grading Every Claimed Deal
Grade any claimed deal on a GP's track record into one of five attribution grades, so two reviewers agree and you know which deals to strip before computing the multiple.
Key takeaways
- Every claimed deal on a GP's track record starts at Unsupported and moves up only on evidence the GP did not supply or select.
- A missing Form D or Delaware listing never justifies a Contradicted grade, because Form D needs no amendment for a relationship change or any change after the offering ends, and Delaware lists directors only as of the March 1 filing date.
- In Refolk's index of 5,280 US and UK profiles carrying a Venture Partner title, only 3 of a 25-profile sample list it as their top current title, so the title alone says nothing about decision rights.
- If LPs call only 3-5 of the 8-10 references a GP prepares, every on-list call can be friendly, so attribution has to rest on deal-team and off-list calls.
- A prior-firm attribution letter that says the GP was a member of the deal team supports a Supporting Role grade at most, never a lead.
- Strip Unsupported and Contradicted deals before computing the attributed multiple, and still show their performance separately so exclusions stay fact-based.
This standard is for LPs, fund-of-funds analysts, angels backing an emerging manager, and funds hiring a partner. Use it when you need to grade a track-record slide one deal at a time before you commit capital or make an offer. It defines five attribution grades, the public records and reference calls that earn each one, and the rule for which deals come off the table before anyone computes an attributed multiple. Two reviewers who apply it to the same slide should reach the same grades.
What do the five attribution grades mean?
Every claimed deal gets exactly one grade: Sourced and Led, Led on Platform Flow, Supporting Role, Unsupported, or Contradicted. The strongest evidence you hold sets the grade. The claim on the slide does not, and only Unsupported and Contradicted deals come off the attributed table.
Four terms, defined once:
- Sourced means the person originated the deal. The founder's first real contact with the firm came through them, and that relationship existed before the firm's check.
- Led means the person championed the deal inside the firm. They took it through the investment decision, ran terms, and took or directed the board or observer seat.
- Platform flow is deal flow that arrived through the prior firm's brand, its inbound, or another partner's network, not through the individual's own relationships.
- Attributed multiple is the return multiple computed only on deals the person can be credited with.
A written standard is needed because the most widely used template doesn't define these words. ILPA's due diligence questionnaire has a track record section. It asks "Are any investments in the Firm's track record excluded from the provided appendices?" and includes a performance and attribution appendix. But I can find no public ILPA definition of "sourced", "led", "board seat" or "managed" at the deal level. One practitioner says DDQ 2.0 "asks almost nothing about deal sourcing." ILPA itself says the questionnaire fits established private equity managers best and that it is developing modules for small and emerging managers. I could not confirm that such a module has been published, so check ILPA's DDQ page before assuming one exists. ILPA is also a trade association, not a regulator, and nothing compels a GP to use its format.
| Grade | Minimum evidence | Attributed table |
|---|---|---|
| Sourced and Led | Tenure covers check; dated origin evidence; independent call names them lead | Kept; in lead multiple |
| Led on Platform Flow | Tenure covers check; independent call names them lead; origin was the firm's | Kept; in lead multiple |
| Supporting Role | Tenure covers check; on deal team; lead not shown | Kept; role disclosed |
| Unsupported | Only GP materials or on-list calls | Stripped; shown separately |
| Contradicted | Dated record or named source says otherwise | Stripped; noted in memo |
An independent source is one the GP neither supplied nor selected. That covers a public filing, an off-list reference, or a founder account published without the GP asking for it.
The kept deals produce two numbers. The attributed multiple covers Sourced and Led, Led on Platform Flow and Supporting Role. The lead multiple covers only the first two. Abbott Capital's guidance supports showing both. It says a GP should disclose every attributable transaction and describe their role accurately, "be it as a deal lead, board member, or general deal team member."
Which public records place a person on a deal, and where do they go blind?
A filing proves that a person held a position on a date. It never proves who sourced or led the deal. Use filings to pin down presence and timing, and treat a missing filing as a gap in the record, not as a contradiction.
| Record | What it shows | Timing | Blind spot |
|---|---|---|---|
| Form D Item 3 | Director, officer or promoter when filed | 15 days after first sale | No amendment after offering ends |
| Form D amendment | Newly added related persons | As soon as practicable | Relationship change alone exempt |
| Delaware annual report | All directors on filing date | Yearly, by March 1 | Terms between filings vanish |
| Delaware LLC | Nothing | No report required | Members never disclosed |
| Companies House AP01 | Appointment date | 14 days | UK entities only |
Form D
Item 3 lists related persons, each ticked as Executive Officer, Director or Promoter. Form D's own instructions say "title alone is not determinative." Promoters must be listed if they acted within the past five years of the later of the first sale or the date the filing was due. One commentary puts it bluntly: "Who counts as a promoter is the filer's call." Form D also no longer requires owners of 10% or more to be listed as related persons, so a fund's large stake does not put its partner on Item 3.
The gap that matters most is amendments. None is required for a change after the offering terminates, or for a change only in a related person's relationship to the issuer. An annual amendment is due only if the offering is still running a year after the last filing. A board seat taken after a round closes may therefore never appear on that round's Form D.
Form D's real value for attribution is the date of first sale. It is the best public anchor for when the round happened, and you test it against the GP's tenure.
Delaware
The annual report lists "the names and addresses of all the directors as of the filing date of the report" and is due March 1. It is a once-a-year snapshot. A board term that starts after one March 1 filing and ends before the next leaves no public trace, so claims of board service lasting under a year can't be checked there. Delaware LLCs file no annual report at all.
United Kingdom
Director appointments must be filed at Companies House within 14 days of appointment, under section 167 of the Companies Act 2006. Since the Economic Crime and Corporate Transparency Act 2023, acting as a director before the company has notified Companies House within that window is an offence. That makes the UK record the strongest of the three. If a claimed UK directorship is missing, ask the GP about it directly, and first ask whether the seat was a director seat or an observer seat, because observers do not appear as officers.
A missing filing is a gap in the record, not a contradiction of the claim.
How do you tell sourcing from leading from platform flow?
Sourcing is proven by a relationship that existed before the firm's check and ran through the person, not the firm. Leading is proven when people who were in the room name the person as the deal's champion without being prompted. No public source ranks this evidence, so this standard sets the ranking.
ValueAddVC frames the test as "Did you find this deal yourself, or did it come through your prior firm's platform?" GoingVC describes founder pull shown "ideally in your DM or referral logs." Neither source ranks bylines, dated writing or earlier angel checks against each other. The figure below shows the layers this standard uses. Each inner layer is needed for a higher grade.
Attribution evidence layers
- ClaimThe GP's table, profile, deck and on-list references; required, but proves nothing alone
- TenureDated start and end at the prior firm from announcements or the IAPD record
- PresenceCheck date inside tenure; filing or deal-team call places them on the deal
- RoleIndependent founder, co-investor or deal-team call names them as lead
- OriginDated evidence the relationship ran through them before the firm's check
Origin evidence, strongest first
- A founder-published account naming the person as first contact. This proves origin when it predates the GP's fundraise. It lies when it was written after the GP started raising, or when it thanks the whole firm.
- A personal angel check that predates the fund's check. This proves the person was in before the firm. It lies when the "angel check" was a courtesy allocation in the same round, made after the firm had committed. Compare the dates.
- Dated writing on the company or its thesis before the round. This proves interest and attention, not contact. It lies when the writing is a generic sector piece that never names the company.
- DM or referral logs supplied by the GP. These can prove first contact. They lie when they are curated screenshots. Ask for the full thread with timestamps, and treat logs as GP-supplied, so they count only alongside an independent source.
Lead evidence
- A co-investor or former deal-team member who names the person, unprompted, as the one who ran terms or presented to the investment committee.
- An attribution letter that assigns a named role to each named deal.
- A founder who names the person as their main contact from first meeting to term sheet.
If a deal has lead evidence and no origin evidence, it grades Led on Platform Flow. That is a real grade and not a penalty. It shows the person can run a deal, but not that they can find one. Fund I underwriting cares about both.
Not every manager comes up through a firm. One blended estimate puts the share of emerging managers with prior institutional VC experience at about half, without showing its method. For managers from outside institutional VC, the track record is often personal angel checks. Attribution there is mostly a question of dates and cap-table position, and platform flow rarely applies.
What does a "Venture Partner" title tell you about decision rights?
On its own, very little. In Refolk's index, most people with the title hold it alongside an operating role, so it shows an affiliation with a firm, not authority to lead a deal.
| Refolk index, "Venture Partner" title, US + UK | Value |
|---|---|
| Total profiles | 5,280 |
| Sample size | 25 |
| Top current title is Venture Partner | 3 (12%) |
| Top current title is CEO or founder | 7 (28%) |
| Top current title is Board Member or Scout | 3 (12%) |
In the sample, a CEO or founder title outranks the venture partner title more than twice as often as the reverse. The plain reading is that many venture partners run companies and invest part-time. Firms vary in whether such a person can lead a deal or holds an investment committee vote, and the title doesn't tell you which arrangement applied.
Another result from the same index shows how unreliable self-applied labels are. Queries for "Partner" or "General Partner" filtered by a VC or PE industry label returned zero results in both the US and the UK. People who plainly invest often don't carry the industry label you'd expect. When building a reference list or a tenure timeline, search by firm name and title rather than industry tag.
Limit: the proportions come from one combined US and UK grouping and a 25-profile sample. Rerun the count before you build a policy threshold on it. The direction is useful. The exact percentages are not yet.
What to do with the title: for any deal claimed during a venture partner stint, ask a former-firm reference whether the role could lead a deal or vote. Until someone independent says it could, cap that deal at Supporting Role.
How should the reference calls be sequenced?
Run calls in a fixed order: founder, then co-investor, then former firm, then off-list. Each call tests a different part of the claim. Published call counts vary widely and none gives a mix by type, so set minimums per deal, not per slide.
| Source | Context | Calls |
|---|---|---|
| Altss | First-time GP | 3-5 direct, plus back-channel |
| LPBacked | Fund-of-funds | 10-20 |
| Investor Ready Capital | Real estate institutional LP | 15-25 |
| Angel Investors Network | Attribution, same deal team | 2-3 or more |
The real estate figure is for real estate managers, not venture. It shows how far norms stretch, not a venture benchmark. LPBacked's list for fund-of-funds diligence is the most useful mix: "founders you backed, founders you passed on, co-investors, former colleagues, and off-list references." No source gives a percentage split between founder, co-investor and former-firm calls.
Attribution reference sequence
- FounderWho made first contact, who ran the process, who took the seat
- Co-investorWho set terms and who the firm sent to the board
- Former firmWho brought the deal in, who presented it, what the role allowed
- Off-listA founder or deal-team member the GP did not name
- ReconcileCompare the names each source gave unprompted
Per-deal minimums
- Sourced and Led: a founder call confirming first contact through the person, plus at least one co-investor or former deal-team call naming them as lead. At least one of these must be off-list.
- Led on Platform Flow: at least one independent call naming the person as lead.
- Top deals by value: follow Angel Investors Network's floor of two or three calls with people who sat on the same deal team. The deals that drive the multiple deserve the most checking.
Finding off-list people is the hard part. You need former colleagues whose time at the firm overlapped the deal, and founders the GP backed or passed on, and the GP has no reason to point you to either. Refolk handles this as a plain-English search by firm, title and tenure window across LinkedIn, GitHub and the open web, so the back-channel list doesn't depend on the GP's own suggestions.
FOUNDER 1. How did your company first come into contact with the firm, and who was the first person you spoke with? 2. Did you know that person before the round? Since when, and how? 3. Who ran the process on the firm's side from first meeting to term sheet? 4. Who negotiated the terms? 5. Who took the board or observer seat, and did that change later? 6. If you needed one person at the firm on this company today, who would you call? CO-INVESTOR 1. Who did you deal with at the firm on this round? 2. Who set or negotiated terms on their side? 3. Who sat on the board for them, and how active were they? 4. Did anyone else at the firm play a larger part than the person you just named? FORMER FIRM 1. Who brought this company into the firm? 2. Who presented it to the investment committee? 3. Did this person's role carry a vote or the right to lead a deal alone? 4. Would you describe their role on this deal as lead, board member, or deal team member? CLOSE (every call) Who else worked on this deal that I should speak with? RECORD: the name the source gave unprompted, the question that drew it, and whether the source was on the GP's list.
Ask the open question first and never state the GP's claimed role before the source names someone.
Budget the time realistically. Fund-of-funds diligence overall runs 6-12 months from first substantive meeting to commitment. Attribution calls fit inside that window, and they slip when they are left until the end.
What is the deal-by-deal grading procedure?
Run eight steps in order, recording a dated source at each one. Sources disagree on where the attribution letter belongs. Angel Investors Network says to get it before going further, while fund-of-funds practice puts it alongside references. This standard requests it at step 2 and doesn't wait for it.
Grading a claimed track record
- Get the claimed deal listAsk for a deal-by-deal table with company, round, first-check date, claimed role and board seat, including excluded deals. Done when every row names a company, a round and a claimed role.
- Request the prior-firm attribution letterAsk the GP to obtain a letter from each prior firm with written consent to share deal data. Done when it is received, refused, or the GP declines to ask, and that outcome is recorded.
- Build the tenure timelineFix start and end dates at each prior firm from firm announcements and the Form ADV record on IAPD, not only the profile. Done when every date is tied to a named source.
- Pull first-check datesDate each deal from the Form D date of first sale or a round announcement, and flag any check before the start date or after the exit. Done when every deal is dated or marked undated.
- Check board recordsMatch each claimed seat to Form D Item 3, the Delaware annual report or Companies House. Done when each seat is matched or marked unmatched with the reason a match may be impossible.
- Gather sourcing evidenceCollect founder-published accounts, dated writing before the round, and angel checks that predate the fund's check. Done when each deal claimed as sourced has origin evidence logged or is marked as having none.
- Run reference callsCall founder, co-investor, former firm, then at least one off-list source, using the same script. Done when every deal claimed as led has at least one independent call naming the lead.
- Grade and stripAssign each deal one grade, resolving disagreement to the lower grade. Remove Unsupported and Contradicted deals, show their performance separately, then compute the attributed multiple.
Notes on the steps
- Step 1. Ask for excluded deals too. ILPA's questionnaire asks whether any investments were left out of the appendices, and Abbott Capital says exclusions should be fact-based and the excluded deals' performance still shown.
- Step 2. LPBacked recommends including the prior employer's written consent to share the data. Pipeline Road notes that where a prior firm restricts disclosure, the track record can be described qualitatively. How often prior firms refuse or confirm these requests is not publicly established, so don't treat a refusal as evidence of inflation.
- Step 3. I know of no published rate at which profile dates diverge from filed or announced dates. Treat profile dates as claims until a firm announcement or the IAPD record fixes them.
- Step 4. A check dated before the GP's start or after their exit, using a dated public record, is one of the few routes to Contradicted. A check dated only by the GP's own table is undated.
Please provide one row per investment in your track record, including any you exclude from headline figures. 1. Company legal name and state or country of incorporation 2. Round, and date of the firm's first check 3. Firm and fund that invested 4. Your title at the firm on the check date 5. Your role: sourced, led, board seat, observer seat, or deal team member 6. Who first introduced the company to you, and on what date 7. Others at the firm who worked on the deal 8. Board or observer seat holder, with start and end dates 9. Co-investors in the round, with one contact name each 10. A founder contact for a reference call 11. Current status, and realized or marked value 12. If excluded from headline figures, the reason Please also confirm whether each prior firm has consented in writing to sharing this data, or which deals you will describe qualitatively.
Send before the first diligence call; every column is required for every deal, including excluded ones.
At step 8, use the matrix below for any deal where records and references point in different directions.
Records versus references
How do deal-attribution checks go wrong?
They go wrong in two directions. The common one is crediting the person a filing or letter names, when the record only shows they were present. The other is grading a deal Contradicted because a record is missing, when the record was never going to exist. The seven failure modes below cause most mis-grades.
Inflation does happen. A secondary summary reports that the SEC fined a private fund manager $1 million for reclassifying an investment to inflate a track record. Locke Capital claimed an 11-year track record and over $1 billion in assets under management while holding under $165 million. Grading by this standard won't catch everything, but it removes the easy wins.
1. Crediting whoever Form D names
What it looks like: the GP appears in Item 3 as a director, so the deal gets graded as led. Why it lies: the board seat may be a firm seat, filled by whichever partner was available, while a colleague did the work. Form D's own instructions say title alone is not determinative. Check: the co-investor call and a per-deal role in the attribution letter.
2. Treating no filing as no seat
What it looks like: you find no Form D listing or Delaware entry, so you grade the deal Contradicted. Why it lies: seats taken after the offering ends need no amendment, a relationship change alone is exempt, LLCs disclose nothing, and Delaware terms between March 1 filings never appear. Check: ask the founder. Absence can support Unsupported, never Contradicted.
3. Trusting profile tenure dates
What it looks like: the profile says the GP joined in spring, the round closed in summer, so the check is inside tenure. Why it lies: profile dates are self-reported, and how often they diverge from filings is unpublished. Check: firm press releases and the IAPD record for Form ADV.
4. Reading the attribution letter as a lead confirmation
What it looks like: a letter on the prior firm's letterhead confirms the GP's involvement in the listed deals. Why it lies: it is drafted for the GP and often uses group language such as "was a member of the deal team." Check: does it name a role for each deal? If not, it supports Supporting Role at most.
5. Equating "Venture Partner" with decision rights
What it looks like: the GP was a Venture Partner when the check went in, so it gets read as a partner-level lead. Why it lies: in Refolk's sample, most holders list an operating title above it, so the role is often part-time. Check: a former-firm reference on whether the role could lead or vote.
6. Leaning on hand-picked references
What it looks like: five enthusiastic calls, all from the GP's list. Why it lies: Altss reports that LPs call only 3-5 of the 8-10 names a GP prepares, so every call can be friendly. Check: off-list calls, including founders the GP passed on.
7. Letting losses drop off as "off-strategy"
What it looks like: the slide covers a subset of deals, and the missing ones turn out to be write-offs. Why it lies: exclusion turns a mediocre book into a strong one without changing a single grade. Check: Abbott Capital's rule that exclusions are fact-based and excluded deals' performance is still shown.
What should be verified before computing the attributed multiple?
Before the multiple goes into a memo, every deal should have one grade, at least one cited source the GP did not supply, and a recorded reason for any exclusion. Use the checklist below as the sign-off gate, and give it to a second reviewer who grades the slide independently.
Attribution sign-off
- Every claimed deal has a company, round, claimed role and a single recorded grade.
- Every grade above Unsupported cites at least one source the GP neither supplied nor selected.
- The GP's start and end dates at each prior firm are tied to a source other than their own profile.
- Every first-check date is tied to a Form D date of first sale or a round announcement, or marked undated.
- No deal is graded Contradicted on the absence of a filing alone.
- Every claimed board seat is matched to a filing or marked unmatched, with the reason a match may be impossible.
- The attribution letter was read for per-deal roles, and group language was not treated as a lead confirmation.
- Every deal graded Sourced and Led has at least one completed off-list call.
- Every deal claimed during a Venture Partner title has a former-firm statement on whether the role could lead.
- Unsupported and Contradicted deals are removed from the attributed table, and their performance is shown separately.
- Every excluded deal has a fact-based reason recorded.
- Two reviewers graded independently, and every disagreement was resolved to the lower grade or by new evidence.
Keeping the standard current
- Re-grade when new evidence arrives. A later founder interview or a new filing can move a deal up or down. Record the date of each grade, not just the grade.
- Recheck the registries on their cycle. Pull the Delaware director list again after each March 1, and look for Form D amendments on any offering still running a year after its last filing.
- Watch the template. If ILPA publishes an emerging-manager module with its own definitions of sourced and led, map these five grades onto it, so a GP's DDQ answers and your grades use the same words.
- Refresh the title data. The Venture Partner figures come from a small combined sample. Rerun the count in your own market before setting a policy threshold on it. Refolk can return venture partners who also list a current founder or CEO role in a given city, which makes it quick to check how common dual-role holders are in your market.
Adopt the standard as policy by putting the grade table and the checklist in your diligence template. Then require that every track record presented to an investment committee or hiring panel shows both the attributed multiple and the lead multiple, alongside the stripped deals.
Questions practitioners ask
How do I verify which deals a VC partner actually led at their prior firm?
Ask the people who were in the room, not the partner. Call the founder and a co-investor with open questions such as who ran terms and who took the board seat, and let them name the lead before you mention the claim. Fix the partner's tenure dates to a dated source and confirm the first-check date falls inside them. A filing that names the partner proves presence, not leadership, because the seat may belong to the firm.
Is a prior-firm attribution letter enough to confirm a GP's track record?
No. The GP drafts or requests the letter, and it often uses group language such as was a member of the deal team. That wording supports a Supporting Role grade at most. A letter counts toward a lead grade only when it names a specific role for each specific deal, and even then it should be checked against at least one independent reference call.
Can I grade a deal Contradicted if the GP is not on the Form D?
No. Form D needs no amendment for a change in a person's relationship to the issuer or for any change after the offering ends, so a board seat taken after the round closes may never appear. Delaware LLCs file no annual report at all. Contradicted needs affirmative evidence, such as a check dated outside the GP's tenure or a founder naming someone else as lead.
How many reference calls should an LP make on an emerging manager's track record?
Published figures run from 3-5 direct calls plus back-channel checks for a first-time GP to 10-20 for fund-of-funds diligence, and no source gives a split by type. For attribution, set the minimum per deal rather than per slide: a founder call and at least one co-investor or former deal-team call for any deal graded as led, with at least one call off the GP's list.
What should I do if the GP's former firm refuses to confirm attribution?
Record the refusal and keep grading on other evidence. Prior firms sometimes restrict disclosure, and the track record can then be described qualitatively. How often firms refuse is not publicly established, so a refusal is not itself evidence of inflation. It does remove one source, so the affected deals need stronger founder and co-investor evidence to rise above Supporting Role.
Try it on the search you came here for
Stop building boolean strings. Just describe the person.
Type one sentence. I plan the search, read GitHub, public LinkedIn and Crunchbase records, and the open web as it is right now, and hand back a ranked list with the reason next to every name.
01Describe them
One plain sentence. Role, city, stack, stage, whatever matters to you.
02I read the web live
GitHub, public LinkedIn and Crunchbase records, the open web. Not a database that went stale last quarter.
03You read the shortlist
Ranked, with the reasoning under every name. Open a profile, ask a follow-up, narrow it down.
- Staff backend engineers in NYC who shipped Rust in production
- Series A fintechs in SF under 50 people, growing headcount this year
- Maintainers of fast-growing Rust web frameworks on GitHub
- No boolean, no filters, no seat to buy. One box.
- Read at search time, so a profile updated yesterday counts today.
- Every step visible as it runs, every name with its reason.
500 free credits on sign-up. No card, no demo call. See real searches.