# The First-Screen Clearance Standard for Sourced Companies

*You will grade any sourced company advance-or-pass at first screen against fixed gates, so two analysts on the same deal reach the same verdict.*

- Canonical URL: https://www.refolk.ai/guides/first-screen-clearance-standard
- Pillar: Investing and deal sourcing
- Format: Standard
- Published: 2026-09-07
- Last reviewed: 2026-09-07
- Reading time: 16 min

Deciding whether a sourced or inbound company earns a partner's first meeting is the highest-volume judgement a fund makes, and it is usually the least consistent. This guide is for early-stage investors, platform and talent partners, and angels who run or feed the first screen. It gives you a gradeable definition of done: fixed gates that turn "does this deserve a meeting" into a verdict two analysts on the same deal reach the same way.

Existing public material describes the screening funnel narratively and hands you a generic due-diligence checklist. Neither fixes the advance-or-pass line. What follows does: hard gates for thesis fit, minimum evidence, and structure, plus the auto-pass red flags and the champion requirement, written so you can adopt them as team policy.

## Why the first screen needs a fixed standard

The first screen kills on portfolio-list fit, not company quality, which is exactly why fixed gates make two analysts agree. Surveying 885 VCs at 681 firms, the median firm closes about four deals a year and considers roughly 100 opportunities for each one it closes. Most of those opportunities never reach a human conversation.

A practitioner write-up citing Strebulaev puts it plainly: roughly 70 of 100 opportunities die in the initial screen, usually within minutes, without any meeting. SheetVenture reports 80 to 90% of inbound rejected at screen. The consequence is counterintuitive but load-bearing for this standard: most "no"s carry no signal about the business. They are stage, sector, geography, or cheque mismatches. If most passes are mechanical, then the screen should be mechanical too, and mechanical decisions are the ones two people can grade identically.

**~70% - Of 100 sourced opportunities, share that die in the initial screen with no meeting**

Usually within minutes of the deck landing, on portfolio-list fit rather than company quality.

The alternative is what most funds actually do: screen by feel. When the screen is a vibe, two analysts on the same deal disagree, partner time gets spent on deals that never fit, and genuinely on-thesis companies leak out because nobody wrote down why they passed. A standard fixes the line so the disagreement moves to where it belongs - the actual judgement calls - and stops happening on the mechanical gates.

> Most passes carry no signal about the business, which is exactly why the gates that produce them should be fixed and gradeable.

## The four hard thesis-fit gates

Thesis fit is four gates, and each is binary: stage, sector, geography, and cheque or ownership. A company that misses any one is a pass at the gate, regardless of how good it looks. This is the core of what makes the screen reproducible.

Screening assesses whether the company's stage fits the fund's preferred stages, whether it sits in a targeted sector, and whether it is based in a region where the fund invests. The fourth gate is the one people skip when a deal looks hot: cheque and ownership math. Every fund operates within defined parameters for check sizes and ownership targets. Some firms write $1 to 2.5M first cheques for 10 to 20% ownership. If a company is raising a $15M round, that fund's cheque will not achieve the ownership it needs, and the deal fails the gate before anyone reads the traction slide.

Ulu Ventures publishes a concrete version worth copying: four fit factors, and a startup needs at least three of the four to clear a first meeting. That threshold is what turns four soft preferences into a hard rule.

> **Rule:** The cheque gate is not optional
>
> Confirm the round size against your ownership target before any meeting is scheduled. A company raising more than your cheque can meaningfully own is a pass at the gate, no matter how on-thesis the rest looks.

Each gate says what it proves and what it looks like when it lies. Stage proves the company needs the kind of capital and involvement you offer; it lies when a founder labels a $6M raise "seed" to fit your mandate. Sector proves you can add value and hold conviction; it lies when a company describes itself with your buzzword but sells into an adjacent market. Geography proves you can support and govern the company; it rarely lies. Cheque and ownership prove the deal can return your fund; it lies when the round size in the ask contradicts the ownership target in your model.

## What documents you screen on, and what you do not

At first screen you work from a thin, founder-prepared stack: the pitch deck, a one-pager or executive summary for very early companies, and a cap table in PDF or spreadsheet when available. That is the whole surface. You do not have, and should not wait for, a CIM or formal data-room documents - those are a private-equity artifact reserved for later, typically after a mutual NDA.

This matters for the standard because it bounds what "minimum evidence" can mean. You are grading a deck of 8 to 40 slides covering team, market, product, traction and the fundraising ask, plus whatever cap-table file exists. The gates must be answerable from that stack alone. If a gate can only be judged after diligence, it is not a first-screen gate.

| Document | Available at first screen | Used for |
|---|---|---|
| Pitch deck (8-40 slides) | Yes | Thesis-fit gates, minimum evidence |
| One-pager / exec summary | Yes, for very early companies | Fast thesis-fit read |
| Cap table (PDF or sheet) | When provided | Structural pre-screen |
| SAFE notes | Sometimes | Structural pre-screen |
| Founder financial model | Sometimes, quality varies | Later diligence, not screen |
| CIM / data room | No, follows an NDA | Deep-dive only |

The practical rule: if you find yourself wanting a document you would only get after an NDA, you are trying to do diligence at the screen. Stop, and decide advance-or-pass on what you have.

## The structural pre-screen gate

Cap-table structure is a meeting-efficiency filter, not diligence, and it runs in under ten minutes on inbound materials. Institutional investors run this fast structural pre-screen before agreeing to meet. It is its own gate for a reason: a stacked SAFE, dead equity, secondary sales, a depleted option pool, or governance fragmentation each triggers a specific investor inference, and each can auto-pass a company before its story is ever heard.

Because the check is so fast and so consequential, it must sit at step three, ahead of the meeting, not be discovered afterward. A clean-looking deck routinely hides a cap table that is already un-financeable at the next round. Running this late is one of the most common ways funds waste partner time.

> **Watch out:** A clean deck hides a broken cap table
>
> Stacked SAFEs and dead equity do not show up in a pitch. If you run the structural pre-screen only after the partner meeting, you spend an hour on a deal that failed a ten-minute gate. Run it before scheduling.

Treat this gate as pass or auto-kill. The pre-screen is not the place to negotiate structure or model a cleanup - it is a filter that answers one question: is the current structure financeable enough to justify a partner's time. If the answer is no and the founder is unwilling to restructure, that is a documented pass without a meeting.

## The minimum-evidence bar: evidence, not assertion

Every required field must carry evidence, not assertion. A deck claims traction; the standard asks whether the claim is backed by something you could show a partner. The four required fields at the claimed stage are team, market size, traction curve, and differentiation.

The sharpest false positive here is reading a peak as a trend. Consistent month-over-month growth signals traction; a single spike does not. Read curves, not peak numbers. A founder who shows one strong month is showing you a number, not a curve, and the standard should record that field as unmet.

Team is the field most often under-weighted at screen, and the survey data says that is a mistake. VCs rate the management team above product and technology, and attribute portfolio success more to team than to the business itself. A screen that gates only on market and traction is mis-weighted against how deciders actually vote. Weight team at least as heavily as the other fields in your rubric.

#### What each evidence field must show

1. **Team** - Relevant operating or domain history, founder-market fit, evidence they can execute
2. **Market** - A sized, reachable market with a credible path in, not a top-down TAM slide
3. **Traction** - A curve over time, consistent month-over-month, not a single revenue spike
4. **Differentiation** - A clear, defensible reason this team wins in a contested space

*Each field is graded present only when backed by evidence a partner could see, not an assertion in a slide.*

Finding real evidence for the team field is where a public-graph search pays for itself. When a deck asserts a "senior team" without specifics, [Refolk](/) lets me confirm it in plain English against public records before the field is graded present.

I ran this search: `Seed-stage B2B SaaS founders in the US who previously worked as senior engineers or PMs at a company that reached Series C or an exit.` - [see the full result list](https://www.refolk.ai/s/p3xxs73gd7).

*Returns founders whose prior roles are verifiable in the public graph, so the team evidence field is graded on record, not on the deck's claim.*

## The screening procedure, step by step

The screen is an ordered eight-step procedure, and the gates fire in sequence so a deal fails cheaply and early rather than after a partner has spent an hour. Each step has an owner, a rough time cost, and a definition of done.

#### First-screen clearance procedure

1. **Intake and log** - Enter the company in the CRM with source, deck, one-pager and cap table if present, and assign a named owner. Done means a record exists with a named owner.
2. **Hard thesis-fit gate** - In a 2 to 5 minute scan, check stage, sector, geography, round size and ownership math against the fund mandate. Done means a pass on any single gate missed, or an advance.
3. **Structural pre-screen** - In under ten minutes, review the cap table, SAFE stack, option pool and obvious legal or IP flags. Done means no auto-kill structural defect remains.
4. **Minimum-evidence check** - Confirm team, market size, traction curve and differentiation are present as evidence at the claimed stage. Done means each required field is backed by something you could show a partner.
5. **Score against fixed rubric** - Rate each criterion on a set scale, weight team at least as heavily as market and traction, and apply the threshold. Done means a numeric score plus advance, pass, or needs-human-review.
6. **Champion route** - A named partner must own conviction before scheduling. Done means a specific partner agrees to take the first meeting.
7. **First meeting** - The champion partner meets the founder over video or coffee for 30 to 60 minutes. Done means the meeting happened and notes are in the record.
8. **Debrief and verdict** - The partner and team debrief and decide to advance, hold or pass, typically within 3 to 7 days. Done means a reason code is logged.

Two structural notes. First, sources disagree on the order of scoring and champion routing. Most funds route junior-first: an associate triages and advances the most promising fit-checked deals to a partner. Others, following Suster, have the founder meet a partner directly who then rallies internal support. Pick one and write it down; either works, but a fund that leaves it ambiguous produces inconsistent verdicts.

Second, the decision forum for what advances past the meeting is usually the weekly Monday partners' meeting, where most funds require unanimous concurrence - except that in large funds a single partner can sometimes decide unilaterally on small cheques between $500K and $1M. Your first-screen standard governs everything up to the meeting; the partner-meeting decision rule is a separate policy.

#### Where the sourced pipeline narrows

| Stage | Figure | Note |
| --- | --- | --- |
| Opportunities considered | 100 | Per closed deal, median firm |
| Survive the screen | 30 | Roughly 70 die in the screen with no meeting |
| Reach a management meeting | 25 | About a quarter of opportunities |
| Reviewed at a partner meeting | 8 | About a third of those met |

*Volumes from the Gompers and Strebulaev survey and practitioner reports, widest first.*

## The champion requirement

Absence of a champion is the real pass mechanism, so the standard requires a named partner owner before anything is scheduled. Interest from one partner does not mean the firm is interested. When a VC goes quiet, it often means no one inside stepped up to own conviction, and the deal dies in silence rather than by decision.

This is the gate that separates a screen policy that works from one that leaks. A rubric can score a company well and a deal can still evaporate because no partner will put their name on it. So the definition of done for advancing to a first meeting is not "the score cleared the threshold" - it is "a specific partner has agreed to take the meeting." Log the partner's name in the record. If no partner will, that is a pass, and it should be recorded as one with a reason code, not left as an open thread.

This matters more where deciders are scarce. In Refolk's index there are 86 US junior screeners against 36 in the UK, a 2.4x ratio, with UK supply about 88% concentrated in London. Where few partners must absorb concentrated inbound, an undisciplined screen that leaks good deals into silence is the more expensive failure.

| Market | Junior screeners | Top title | Concentration |
|---|---|---|---|
| United States | 86 | Investment Associate | Top hub New York |
| United Kingdom | 36 | Investment Associate | ~88% in London |
| Ratio | 2.4x US:UK | - | - |

## Auto-pass red flags

Some signals are auto-pass at the screen, independent of the rubric score. These are documented kills, and putting them on the standard means an analyst does not have to re-litigate them deal by deal.

- **Structural cap-table defects.** Stacked SAFEs, dead equity, secondary sales, a depleted option pool, or governance fragmentation. Caught in the under-ten-minute structural pre-screen; each triggers a specific investor inference.
- **Anyone but the CEO fundraising.** Per Lemkin, anyone but the CEO reaching out is close to a red flag - the fundraising has to be run by the CEO.
- **No clear reason they win.** In a competitive space, not being clear on why you win is close to an immediate pass.
- **Sector overexposure in the portfolio.** A fourth AI health-tech deal faces a much higher bar if the fund already backed three. This one is portfolio-relative, so check current holdings before advancing.

> **Note:** Sector overexposure is a portfolio check, not a company check
>
> A company can clear every generic gate and still be an auto-pass because your fund already holds three in its category. This flag depends on the portfolio, so build the current-holdings check into the standard rather than assuming a clean rubric result is final.

## How this goes wrong

The failure modes below are where a screen produces the wrong verdict or an inconsistent one. Give these the most attention: a standard is only as good as its resistance to these traps.

- **A gate skipped for a hot deal.** An on-thesis-looking company advanced without confirming ownership math. The fix is procedural: run the cheque gate against the ownership target before any meeting, every time, even when the deal feels obvious.
- **Assertion mistaken for evidence.** A single revenue spike reads as growth. Read curves, not peak numbers; a field backed by one strong month is unmet.
- **Reading the pass email, not the verdict.** Every rejection looks identical - a screen-kill and a considered verdict send the same three sentences. Grade the internal reason code, not the outbound email.
- **No champion equals a silent pass.** One partner's interest is not the firm's interest. Require a named partner owner before scheduling, and record a pass with a code when none steps up.
- **Structural defect surfaced too late.** A clean deck hides a stacked SAFE cap table. Run the structural pre-screen at step three, before the meeting, not after.
- **Automation over-trusts the deck.** Nothing a tool extracts tells you whether the founder can execute or how they respond to pushback. Use a Pass, Fail, or Needs-Human-Review lane, never a binary advance-or-pass.
- **Sector overexposure ignored.** A fourth deal in a saturated category clears the generic rubric. Check current holdings before advancing.

#### Routing a screened company

Horizontal axis runs from Weak evidence to Strong evidence. Vertical axis runs from Fails a hard gate to Clears all hard gates.

| Quadrant | What it means |
| --- | --- |
| Clears gates, weak evidence | Needs-human-review lane; confirm team and traction before scheduling |
| Clears gates, strong evidence | Route to a named partner champion for a first meeting |
| Fails a gate, weak evidence | Pass and log a reason code; no meeting |
| Fails a gate, strong evidence | Pass on the gate regardless of quality; log why, it may reopen later |

*Two axes - thesis and structure fit, and evidence strength - decide the lane a company enters.*

## The clearance checklist and keeping it current

A company has cleared the first screen only when every item below is true. Use this as the gradeable definition of done - if two analysts run it against the same deal, they should reach the same verdict.

#### First-screen clearance checklist

- [ ] A CRM record exists with source, deck, and a named owner
- [ ] Stage, sector, geography and cheque/ownership all pass the hard thesis-fit gate
- [ ] Round size confirmed against the fund's ownership target
- [ ] Structural pre-screen run in under ten minutes with no auto-kill defect
- [ ] Team, market, traction curve and differentiation each backed by evidence, not assertion
- [ ] Traction read as a curve over time, not a single spike
- [ ] No auto-pass red flag present (non-CEO fundraiser, unclear why-they-win, sector overexposure)
- [ ] Numeric rubric score recorded with team weighted at least as heavily as market and traction
- [ ] A named partner has agreed to own conviction and take the first meeting
- [ ] On any pass, a dated reason code is logged, not just an email sent

The last item is the one that compounds. The reusable asset of a pass is the reason code, not the email. When a fund passes, a second record goes into the system: a reason code, a short note, and a date, written so a colleague who never met the founder can open the file in eighteen months and know what to do with it. That is what converts an ad-hoc pass into team memory and lets you reopen a deal when the reason changes.

Keep the standard current by re-checking two things on a cadence. First, the gates: your fund's stage, sector, geography and cheque parameters drift as the fund matures and the portfolio fills, so review the thesis-fit gates and the sector-overexposure list each time you deploy into a new category. Second, the routing order: if verdicts are diverging between analysts, the usual cause is an unwritten choice between junior-first triage and partner-first rallying. Write down which one your fund uses, and the divergence closes. A standard that names its own limits and gets re-checked is the one a team can actually keep open while it works.

## Frequently asked questions

### What is the difference between a first-screen pass and a real verdict?

A first-screen pass kills a deal on portfolio-list fit - stage, sector, geography or cheque - usually within minutes and without a meeting. A real verdict comes after a partner meeting and debrief. The problem is that both send the founder the same three-sentence rejection email. The distinction lives in your internal reason code, not the outbound message, so grade and log the code rather than reading the email as signal.

### How many companies actually clear the first screen?

Roughly 80 to 90% of inbound is rejected at screen, and about 70 of every 100 sourced opportunities die there without a meeting. Funds screening 500 to 1,000 companies a year convert fewer than 3% to serious diligence. The first screen is designed to protect partner time, so a high pass rate is the system working, not failing.

### Which thesis-fit gates should be hard gates rather than judgement calls?

Four gates are consistently named as hard: stage, sector, geography, and cheque or ownership. The cheque gate is often missed - if a company is raising a $15M round and your fund writes $1 to 2.5M first checks for 10 to 20% ownership, the math will not clear your ownership target, and that is a pass regardless of quality. Ulu Ventures publishes a version requiring three of four fit factors to advance.

### Why require a named partner before scheduling a first meeting?

Because absence of a champion is the real pass mechanism. One partner's interest does not mean the firm is interested, and deals leak into silence when no one inside steps up to own conviction. Requiring a named partner owner before scheduling converts a vague nod into an accountable decision and stops good deals from going quiet.

### Should cap-table review happen before or after the first meeting?

Before. The structural pre-screen - stacked SAFEs, dead equity, secondary sales, depleted option pools, governance fragmentation - runs in under ten minutes on inbound materials and is a meeting-efficiency filter, not diligence. A clean-looking deck can hide a defect that auto-passes the company, so run it as its own gate at step three, not as a discovery after the partner has spent an hour.

### How do you keep automated screening from over-trusting the deck?

Use a Pass, Fail, or Needs-Human-Review lane rather than a binary advance-or-pass. Nothing a tool extracts from a deck tells you whether the founder can execute or how they respond to pushback. Route anything that clears the mechanical gates but carries execution uncertainty into the human-review lane, so the screen stays fast without pretending a slide answers a question only a conversation can.

---

*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/first-screen-clearance-standard*