# The Missed-Round Audit Playbook: In-Thesis Rounds to Dated Channel Fixes

*Take one quarter of in-thesis rounds to a stated coverage percentage, a four-way miss log, and a dated, owned list of channel fixes in about two working days.*

- Canonical URL: https://www.refolk.ai/guides/missed-round-audit-playbook
- Pillar: Investing and deal sourcing
- Format: Playbook
- Published: 2026-11-08
- Last reviewed: 2026-11-08
- Reading time: 22 min
- Keywords: how to track missed deals venture capital, vc deal flow coverage audit, measure deal sourcing coverage fund, anti-portfolio review process, which rounds did we miss in our thesis

## Key takeaways

- Coverage in a missed-round audit is rounds seen with human contact before first sale divided by in-thesis rounds, and no VC fund has published a target, so your own trailing quarters are the only fair benchmark.
- Form D is due within 15 calendar days of first sale and gives a sworn first-sale date, but it names officers, directors, and promoters rather than investing funds, so you have to trace filed-only rounds through people.
- Early-stage, biotech, pharmaceutical, and high-tech companies are less likely to file a Form D, so a universe built only from filings overstates coverage for the rounds an early-stage thesis cares about most.
- In a private equity benchmark of 176 firms, median coverage was 38.8% on broadly marketed processes and 3.4% on limited ones, which is why never-seen misses need relationship fixes rather than more inbound.
- In Refolk's index, 887 US profiles match Scout roles at VC and PE firms, but only 2 of 25 sampled hold Scout as their current title and 11 of 25 have founder or CEO-type titles, so credit misses to people, not to the scout label.
- A miss with no written pass note dated before first sale is seen-late, not seen-and-passed.

A missed-round audit is a quarterly check of three things: which rounds inside your thesis closed without you, why each one never reached you or reached you too late, and which sourcing channel to change so the same kind of miss does not happen again. This playbook is for early-stage investors, platform and talent partners at funds, and angels who want a repeatable procedure rather than an occasional round of regret. It produces two things: a coverage percentage you can compare from quarter to quarter, and a dated list of channel changes with an owner on each. The whole run takes about two working days.

## What does a missed-round audit measure?

A missed-round audit measures sourcing coverage: the share of in-thesis rounds your fund had real human contact with before the round's first sale. It does not grade investment judgement. A deal you saw early and passed on counts as a sourcing success, whatever the company later became.

That distinction is why the audit is its own job. Published anti-portfolios are stories, not taxonomies:

- Bessemer's anti-portfolio records companies the firm saw and passed on.
- Version One's records who introduced each company. One example is the friend at K9 Ventures who introduced Henry Ward of eShares, the company that became Carta.

Both are useful, but neither separates a sourcing failure from a judgement pass. No fund has published a classification that does, so this playbook supplies one.

The terms used throughout:

- **In-thesis round**: a round that passes your written filter spec for stage, sector, geography, and round size, inside the quarter's date window.
- **First sale**: the SEC's Form D definition, "the date on which the first investor is irrevocably contractually committed to invest." This is the cutoff for "early enough".
- **Seen**: a human activity in your CRM, such as an email, a meeting, or a call note, dated before the first sale. A record merely existing does not count as a sighting.
- **Coverage**: the number of rounds seen before first sale divided by the number of in-thesis rounds.
- **Miss**: any in-thesis round the fund did not invest in. Every miss goes into exactly one of four classes: never-seen, seen-late, seen-and-passed, or competed-and-lost.

The coverage formula adapts a private equity definition: "Seen divided by closed-and-in-mandate is coverage." In a benchmark of 176 private equity firms, median market coverage was 17.6% and the top quartile reached 27.5%. No VC fund has published a seen-rate target, and one VC careers resource lists missed-deal rate as "Internal only". Treat the private equity figures as context, not as a target. Your own trailing quarters are the only fair benchmark.

Sources disagree on cadence. One practitioner process puts the missed-deal review at the end of a weekly loop. The lineage-tracing method treats it as a standalone exercise. I run the audit quarterly and as a standalone exercise, because Form D filings take up to 15 days to arrive and announcements lag further. The weekly loop is still useful for keeping pass notes current, so the quarterly classification has something to read.

## The quarterly audit procedure, step by step

The audit has seven steps, run in order: fix the universe, pull the rounds, match them to the CRM, classify the misses, compute coverage, trace lineage, and decide the fixes. The times are my own estimates. No source publishes timings, so log your actual hours for the first two quarters and adjust.

#### The missed-round audit, one quarter

1. **Fix the universe** - Write the thesis filters (stage, sector, geography, round size) and the quarter's date window into a versioned, dated filter spec. About 1 hour. Done means another analyst could apply the spec and pull the same list.
2. **Pull announced and filed rounds** - Pull announced rounds from databases and news and Form D filings from EDGAR, remove pooled investment funds, and deduplicate. About 3 hours. Done means one list with every row tagged announced, filed, or both.
3. **Match rounds against the CRM in layers** - Match by exact domain, then domain redirect, then founder or officer identity, then company name, and review every name-only match by hand. About 3 hours. Done means every round is marked matched or unmatched with the match layer recorded.
4. **Classify every miss four ways** - Use first human-activity dates, stage history, and pass notes to sort each miss into never-seen, seen-late, seen-and-passed, or competed-and-lost, judged against the first-sale date. About 2 hours. Done means 100% of misses are classified.
5. **Compute coverage** - Divide the number of rounds seen before first sale by the number of in-thesis rounds and record the filter spec version beside the result. About 30 minutes. Done means one percentage with its numerator, its denominator, and the count in each class.
6. **Trace lineage for never-seen misses** - List the investors, accelerator, angels, and founders' prior employers for each never-seen miss. For filed-only rounds, start from the related persons on the Form D. 4 to 6 hours. Done means named upstream nodes with your relationship to each marked.
7. **Decide dated channel fixes** - A partner and the platform lead pick named people or firms to build relationships with, starting with nodes that touched many misses and that the fund barely knows. About 2 hours. Done means a dated list with an owner on every change.

The estimates add up to roughly 15.5 to 17.5 hours, which is the two working days. Lineage is the step that varies most, because it grows with the number of never-seen misses rather than with the size of the universe. If your first run produces more never-seen misses than you can trace in six hours, trace the ones that sit most squarely in thesis first. Carry the rest into next quarter's log rather than tracing all of them badly.

## How do you define the in-thesis round universe?

Build the universe by combining announced rounds and Form D filings into one list, then filter it with a written, versioned spec. Each source on its own leaves gaps, and both sets of gaps make your coverage number look better than it is.

### What each source gives you

| Source | Gives you | Misses |
|---|---|---|
| Announced rounds | Investor names, size, stage | Rounds never announced; small seeds |
| Form D filings (EDGAR) | Sworn first-sale date; officers | Fund names; many early-stage rounds |
| Both, combined and deduplicated | The fullest list available | Rounds that neither file nor announce |

The Form D details that matter here:

- **Deadline**: Form D is due within 15 calendar days of first sale. If that falls on a weekend or holiday, the deadline moves to the next business day.
- **Lateness**: some filers miss the deadline. Superhuman Platform filed 14 days after its first sale, and Alchemy Cloud filed 21 days after. The SEC has brought enforcement actions against companies for filing Form D late, which tells you late filing happens often enough to be worth policing.
- **No investors**: the filing lists "officers/directors/promoters, not fund names". That makes Form D your source for the timestamp, not your source for lineage.
- **Pooled funds**: investment funds file Form D too. One vendor claims about two thirds of filings are funds raising money from their own investors. That figure is unverified, but remove pooled funds before matching either way.

The filing record also leans away from early-stage rounds. Hanley and Yu found that "the majority of venture-capital backed financing rounds are not accompanied by a Form D filing". They also found that early-stage firms, and companies in biotech, pharmaceutical, and high tech, are less likely to file, while larger rounds and rounds already covered by the media are more likely to. One fragment of the same paper gives an average filing rate of 75%, but its context is unclear and it conflicts with the "majority" finding, so do not rely on either number. Check your own slice instead: take 20 announced in-thesis rounds and count how many have a Form D.

The press has the opposite gap. As one tracker of unannounced rounds puts it, "The announcement comes weeks later, or never." Nobody has published the typical gap between first sale and announcement, or what share of rounds are announced versus only filed. When a round is announced but not filed, you have no sworn first-sale date. Use the announcement date as the cutoff and flag the row. That flag matters because a later cutoff makes it easier for a round to count as "seen" than it should be.

### How big the universe should be

| Measure (US) | Value | Per day (derived) |
|---|---|---|
| Deals in the quarter (Q3) | 5,012 | ~54 |
| Deals year to date through Q3 | 14,199 | ~52 |
| First financings year to date through Q3 | 8,625 | ~32 |
| Same quarter, one year earlier | ~4,208 | ~46 |

These figures come from the quarterly Venture Monitor published with the NVCA, for one Q3, and are estimates. Pull fresh figures each quarter. At roughly 54 US deals a day, a thesis that captures 2% of deal count produces about 100 rounds a quarter. That is an illustration, not a benchmark, but it shows the scale: about 100 rows is few enough to classify every one by hand in two days. If your spec produces several hundred rounds, it is too broad to lead to channel decisions. Tighten the spec rather than sampling.

**Universe filter spec**

```
Filter spec version: v1
Dated: YYYY-MM-DD
Owner: <name>
Quarter window: YYYY-MM-DD to YYYY-MM-DD
Window keyed on: first-sale date where filed; announcement date where not (flag these rows)
Stage: <e.g. pre-seed, seed>
Sectors: <list>
Geography: <countries or metros>
Round size: <min> to <max>
Sources: announced rounds (databases, news) + Form D filings (EDGAR)
Exclusions: pooled investment funds; <other>
Definition of seen: first human activity (email, meeting, call) dated before first sale
Match layers: exact domain > domain redirect > founder/officer identity > name (manual review)
```

*Fill every field before pulling any data. Bump the version and date whenever any field changes.*

## How do you match rounds against your pipeline records?

Match in layers, from the strongest key to the weakest: exact domain, then domain redirect, then founder or officer identity, then company name. Record which layer produced each match. Treat a name-only match as unconfirmed until a person has checked it.

#### CRM match layers, strongest first

1. **Exact domain** - Strongest key; safe to accept automatically
2. **Domain redirect** - The same company behind a new or old domain; review it
3. **Founder or officer identity** - Catches renamed companies and filed-only rows with no domain
4. **Company name** - Lowest confidence; confirm by domain or founder before accepting

*Each layer down catches more matches but with less certainty, so a person must confirm name-only matches before they count.*

This mirrors how CRMs handle duplicates. Affinity's duplicate tool sorts organisation matches into "Identical Domain, Domain Redirect, or Identical name". It warns that the lower-confidence types "require individual review because: Different companies can share the same name," and its "Merge all" action works only on identical-domain matches. For people, Affinity describes its email-based matching as about 97% accurate. One Attio integrator puts the rule plainly: "Company dedup uses domain as the strongest signal."

Two practical consequences follow:

- **Filed-only rows usually have no domain.** Form D gives a legal name and the related persons. Go straight to the officer layer and search the CRM for the named founders and officers before you try the company name.
- **Renames break the top layer without warning.** A company that rebranded or moved domain fails the domain match and looks unmatched. Before you mark any row unmatched, check its founders through the person layer.

Nobody has published how often name-only joins between a round list and a pipeline produce false matches or miss real ones. Measure it yourself: count how many name-only matches your reviewer overturns each quarter, and record that number in the audit log.

> **Watch out:** A record date is not a sighting
>
> CRM records created by bulk import or data enrichment carry early creation dates that look like early sightings.
> Use the date of the first human activity, such as an email, meeting, or call, and never the record's creation date.

## The four-way miss taxonomy

Every in-thesis round the fund did not invest in goes into exactly one of four classes. The class depends on when the fund first had human contact with the company and what it recorded afterwards. Only never-seen and arrived-late rounds count against coverage. Passes and competitive losses are questions about judgement and competitiveness, for a different meeting.

| Class | Test | What it proves | When it lies |
|---|---|---|---|
| Never-seen | No human contact before the round went public | A channel gap upstream | A rename hid an existing record |
| Seen-late | Contact after first sale, or before it with no decision | A timing or queue failure | Bulk-import dates make it look early |
| Seen-and-passed | Contact and a written pass, both before first sale | Sourcing worked; judgement decided | The pass note was written afterwards |
| Competed-and-lost | Seen before first sale; term sheet, IC memo, or allocation request | Sourcing worked; lost on access or price | A "would have" claim with no document |

Seen-late has two causes, and they need different fixes:

- **Arrived late**: the first contact came after the first sale but before the round went public. A channel delivered the deal, but too slowly. This counts against coverage.
- **Aged out**: there was contact before the first sale, but no decision was ever recorded. One practitioner write-up describes the misses that matter as "the ones that aged out of the queue before anyone looked at them." This is a process failure, not a channel failure. Count it as covered, because the channel delivered the deal in time, but report it on its own line so the failure stays visible.

Rounds the fund invested in stay in the universe and count as seen. They never enter the miss log.

> **Rule:** No pass note, no pass
>
> A miss counts as seen-and-passed only if a written pass note is dated before first sale.
> Without one, classify it as seen-late (aged out). Notes written after the round closed do not count.

## How do you trace a never-seen miss to its source?

For each never-seen miss, list every node that touched the round before it closed: the investors, the accelerator, the angels, and the founders' prior employers. Then mark which of those nodes the fund has no working relationship with. The unconnected nodes are where channel fixes belong.

The investor half of this is already documented. A quantitative-VC exercise calls it lineage tracing, with the goal to "identify where missed deals came from - who the investors in those deals were." Its third step is a spreadsheet of "all the investors in those companies - whether you know them or not." The published example covers investors only, uses Series B and C eCommerce rounds, and is several years old. No published example extends the trace to accelerators, angels, or prior-employer networks, even though at early stage those are often where a round really came from. Gompers, Gornall, Kaplan, and Strebulaev surveyed 885 respondents from 681 VC firms and found that "Over 30% of deals are generated through professional networks." A founder's former colleagues are exactly that kind of network, and tracing to prior employers is how you see it.

#### Lineage trace for a never-seen miss

1. **Investors** - The lead and every participant, known or not
2. **Accelerator** - The programme and cohort the company came through
3. **Angels** - Individuals named in the round or on the cap table
4. **Prior employers** - Where the founders worked, and with whom
5. **Named node** - One person to build a relationship with, strength marked

*The trace runs from the round's investors out to the people around the founders, and ends at a named person the fund can reach.*

### Filed-only rounds start from people

Form D gives a timestamp but no investors. For a round that was filed but never announced, start the trace from the related persons named on the filing: the founders, officers, and directors. Work outward from them to the people they worked with before, the accelerator they went through, and the angels who have backed them. The most common way lineage stalls is stopping at "lead investor unknown".

Most of the hours in this step go on turning a list of officers, or a sector and a city, into named angels and former colleagues. [Refolk](/) takes that question in plain English and returns the people. A node such as "angels active in this niche and city" becomes a list of names you can check against your CRM in minutes.

Ask me this: `Angels who invested in pre-seed developer-tools startups in New York in the last 12 months` - [run the search](https://www.refolk.ai/start?q=Angels%20who%20invested%20in%20pre-seed%20developer-tools%20startups%20in%20New%20York%20in%20the%20last%2012%20months).

*Returns named angels with their recent pre-seed investments, ready to mark against the fund's CRM as known, weak, or unknown nodes.*

### Credit people, not labels

**8% - of sampled US "Scout" profiles that hold Scout as their current title**

In Refolk's index, 887 US profiles match Scout or Venture Scout at VC and PE firms, but only 2 of a 25-profile sample still hold the title.

| Role group (US VC/PE) | Matches | Role is current title | Founder/CEO-type title |
|---|---|---|---|
| Scout / Venture Scout | 887 | 2 of 25 (8%) | 11 of 25 (44%) |
| Platform / Talent | 88 | 18+ of 25 | 3 of 25 |
| Ratio | 10.1x | n/a | n/a |

Source: Refolk's index. The percentages and the ratio are derived from the sample. In this sample, scouts are mostly founders, and scout programmes sit inside founder networks. If a miss traces to a "scout", log the individual and their own network, not "the scout channel". Otherwise you will invest in a programme when the deal really came through one founder's former team.

**Never-seen miss lineage row**

```
Round: <company> | <stage> | <size> | first sale <date> | source: announced / filed / both
Match result: unmatched | layers tried: domain, redirect, officer, name
Node: Investor | <firm> | <named partner> | relationship: none / weak / working
Node: Accelerator | <programme, cohort> | <named contact> | relationship: none / weak / working
Node: Angel | <name> | relationship: none / weak / working
Node: Prior employer | <company> | <founder's former colleague> | relationship: none / weak / working
Fix candidate: <node> | owner: <name> | by: <date>
```

*One block per never-seen round, one line per upstream node. Leave relationship blank rather than guessing.*

## How do you turn traced misses into channel fixes?

Count how many never-seen misses each upstream node touched, and compare that with how strong the fund's relationship with the node is. Every node that touched many misses but has a weak relationship gets a dated, owned action. A fix names a person or firm, an owner, and a date. "More networking" is not a fix.

#### Where a channel fix pays back first

Horizontal axis runs from Few never-seen misses touched to Many never-seen misses touched. Vertical axis runs from Weak or no relationship to Working relationship.

| Quadrant | What it means |
| --- | --- |
| Known, quiet | Maintain; no action this quarter |
| Known, leaking | Ask why their deals skip you; fix the ask, not the relationship |
| Unknown, quiet | Watch; act if the node comes up again next quarter |
| Unknown, leaking | Priority fix; name a person, an owner, and a date |

*Nodes that touched many never-seen misses and that the fund barely knows are the first places to spend relationship time.*

The lineage exercise ends with a firm number: "Identify 5 investors who did deals in companies you'd not seen that you are going to reach out to." Keep that discipline but widen the types of node. Five named nodes a quarter, drawn from investors, accelerators, angels, and prior-employer alumni, is a list a partner can actually work through.

Why relationships rather than inbound? In the private equity benchmark, median coverage was 38.8% on broadly marketed processes and 3.4% on limited ones, a gap of about 11x. Early-stage rounds, especially insider-led ones, behave like limited processes, where relationships decide who gets access. A low never-seen coverage number therefore calls for relationship fixes, not more inbound volume. Seen-late misses are different. A pattern of arrived-late misses points to a slow channel you already have. A pattern of aged-out misses points to your own queue.

Expect most of the universe to be misses whatever you do. In the same survey of VC firms, "The median firm considered about 200 companies and closed 4 deals a year." The audit's job is to move misses into better classes, not to make the log short.

> **Tip:** Name the partner, not the firm
>
> When a co-investor firm is the node, find the specific partner who led or joined the round.
> Asking Refolk for the partners at that firm who led seed rounds in your sector turns a firm name into a person you can actually call.

### Who owns the fix

The right owner for a channel fix depends on how the fund is staffed, and that varies by country.

| Country | Platform/talent roles | Top title in sample | Top-title share |
|---|---|---|---|
| United States | 88 | Talent Partner (12 of 25) | 48% |
| United Kingdom | 10 | Head of Platform (9 of 10) | 90% |
| US ÷ UK | 8.8x | n/a | n/a |

Source: Refolk's index of Platform Partner, Head of Platform, and Talent Partner titles at VC and PE firms. The shares are derived from the sample. Andreessen Horowitz is the top US employer, with 3 people. No UK profile holds a Talent Partner title. In a UK fund, the Head of Platform is usually the natural owner of channel changes. In the US the work is often split across roles, so put a person's name on each fix rather than assigning it to "platform".

## How does a missed-round audit go wrong?

Most failures in this audit inflate coverage, and that is the dangerous direction: a flattering number stops anyone from fixing a channel. Each failure below comes with a check you can run inside the audit itself.

1. **Universe built only from Form D filings.** Early-stage and high-tech rounds file less often, so the rounds you are most likely to miss never enter the denominator, and coverage comes out too high. Check: take 20 announced in-thesis rounds and count how many have a Form D.
2. **Universe built only from press.** Small seed rounds often go unannounced, so the denominator again leaves out the hardest-to-see deals. Check: compare against EDGAR for the same quarter.
3. **Name-only match.** False positive: two different companies share a name, and you record a sighting you never had. False negative: a company rebranded or changed domain, and an existing relationship shows up as never-seen. Check: confirm every name-only match by domain or founder.
4. **First-logged date taken at face value.** A record created by bulk import or enrichment looks like an early "seen". Check: require a human activity, such as an email or meeting, before the first sale.
5. **Seen-late misfiled as seen-and-passed.** A deal that aged out in the queue gets remembered as a deliberate pass, which hides a process failure behind a judgement call. Check: a pass needs a written pass note dated before first sale. Without one, classify the miss as seen-late.
6. **Lineage stops at the lead investor.** A Form D-only round names no investors, so a trace that begins with investors goes nowhere. Check: start from the related persons on the Form D.
7. **Scout attribution.** In Refolk's index, 44% of sampled US scout profiles have founder or CEO-type titles. Crediting "the scout channel" hides which founder network actually produced the deal. Check: attribute the miss to the person, not their label.
8. **Denominator drift.** If the thesis filters change between quarters, the coverage numbers cannot be compared. Check: version the filter spec and date it.

> A coverage number with an unversioned denominator is a mood, not a measurement.

## How do you keep coverage comparable quarter to quarter?

Version and date the filter spec, keep the match rules and class tests fixed, and score last quarter's fixes before writing new ones. A coverage series is only useful if every point in it was computed the same way.

- **When the thesis changes**, re-run the previous quarter on the new spec so the two versions share one overlapping data point.
- **Before each run**, re-check the Form D rules on the SEC's Form D FAQ rather than relying on this page.
- **Each quarter**, re-pull universe volumes from the quarterly Venture Monitor, because the per-day sizing will move.
- **For each node on last quarter's fix list**, record whether its next in-thesis round reached you before first sale. That is the only real test of whether a fix worked.
- **Keep the miss log cumulative.** A node that shows up in never-seen traces for three quarters is a structural gap, not bad luck.

#### Before calling the quarter's audit done

- [ ] The filter spec is versioned and dated, and was applied unchanged to the whole quarter.
- [ ] Every row is tagged announced, filed, or both, and pooled investment funds are removed.
- [ ] Every round is marked matched or unmatched, with the match layer recorded.
- [ ] Every name-only match has been confirmed by domain or founder.
- [ ] Every "seen" rests on a human activity dated before first sale, not on a record's creation date.
- [ ] Every seen-and-passed miss has a written pass note dated before first sale.
- [ ] Every miss sits in exactly one of the four classes.
- [ ] Coverage is stated as a percentage with its numerator, denominator, and spec version.
- [ ] Every never-seen miss lists named upstream nodes beyond the lead investor.
- [ ] Every channel fix names a person or firm, an owner, and a date.

## Frequently asked questions

### How do you track missed deals in venture capital?

Audit one quarter at a time. Define the in-thesis universe in a written spec and combine announced rounds with Form D filings. Match each round to your CRM by domain and founder identity. Then classify every round you did not invest in as never-seen, seen-late, seen-and-passed, or competed-and-lost. Coverage is the number of rounds seen before first sale divided by the number of in-thesis rounds. Trace each never-seen miss back to its investors, accelerator, angels, and the founders' prior employers.

### What is a good deal sourcing coverage rate for a VC fund?

There is no established good number, because no VC fund has published a coverage target. The closest benchmark comes from private equity. Across 176 firms, median coverage was 17.6% and the top quartile 27.5%. It was 38.8% on broadly marketed processes and 3.4% on limited ones. Early-stage rounds behave more like limited processes. Use your own trailing quarters, computed on a versioned filter spec, as the benchmark.

### Can I build the round universe from Form D filings alone?

No. Hanley and Yu found that most venture-backed rounds have no Form D filing, and that early-stage, biotech, pharmaceutical, and high-tech companies are less likely to file. A filings-only universe drops the rounds an early-stage thesis cares about most, and that inflates coverage. Combine filings with announced rounds. To test your own slice, check how many of 20 announced in-thesis rounds have a Form D.

### How is a missed-round audit different from an anti-portfolio?

An anti-portfolio is a set of stories about companies a fund passed on or missed. A missed-round audit is a measurement. Published anti-portfolios such as Bessemer's and Version One's record what the fund saw or who made the introduction, but none separates a sourcing failure from a judgement pass. The audit does that with four classes and a coverage percentage. Only never-seen and arrived-late rounds count against your coverage.

### Who should own the quarterly missed-round audit?

An analyst runs the data steps, and a partner plus the platform lead own the channel fixes. In Refolk's index, 10 people in the UK hold platform or talent titles at VC and PE firms, and 9 of them are Head of Platform, so that role is the natural owner there. The US has 88 such people, often split between Talent Partner and platform titles, so name the owner on each fix explicitly.

### How do I trace a missed round that has no named investors?

Start from the people named on the Form D. The filing lists officers, directors, and promoters, not fund names, so you cannot trace a filed-only round through its lead investor. Work outward from the named founders and officers: their prior employers, the accelerator they attended, and the angels who have backed them. Record each one as a named node and mark your relationship to it. Then pick fixes from the weakest links.

---

*From the Refolk guide library. I revise these guides rather than replacing them, so the current version is always at https://www.refolk.ai/guides/missed-round-audit-playbook*