The Method · The Shortlist

The Method

How a founder gets qualified, reviewed, and routed.

Eighteen thousand outcome-labeled companies and growing. Eighty-five hand-researched anchors with three primary sources each, spanning seven sectors and five countries. Every introduction sits on this work.

The methodology

By the numbers.

18,038 Outcome-labeled companies in the validation set, as of today. Built from primary regulatory filings, public alumni status directories, and disclosed venture portfolios. No Wikipedia. No third-party data aggregators. Growing weekly.

85 Hand-researched anchor companies, 57 winners and 28 failures (18 blowups plus 10 plateaus). Three primary sources minimum per entry, one of them a primary document. Adding 5 to 10 per week.

47 Rubric questions every applicant is scored against, across nine sections. Built from diligence work at three venture funds, layered with patterns the research database has surfaced.

The dataset

Three tiers, each doing a different job.

The narrative comes from Tier 1. The rigor comes from Tier 2. The context comes from Tier 3. Anyone trying to skip directly to Tier 3 with AI judgments is producing content marketing, not research.

The Anchor Set

85 today, growing weekly · hypothesis generation

Hand-researched. Each company carries founder bios, funding history, problem signal, moat type, why-now signal, outcome timeline, and a minimum of three primary sources. Every candidate pattern starts here. New entries are added at a deliberate pace, five to ten per week, with sector balance across B2B SaaS, deep tech, fintech, climate, healthtech, and consumer.

The Validation Set

18,038 outcome-labeled companies and growing · pattern testing

Structured records built from primary regulatory filings, public alumni status directories, and disclosed venture portfolios. Each record carries an outcome label, a retrieval date, and a primary-source URL. Patterns are tested at this scale with reported effect sizes and Wilson 95% confidence intervals.

The Population

~500,000 companies · base rates and context

Firmographic data only. Used to compute base rates and contextualize findings against the broader market. We do not run AI judgments on the population tier. Only structured statistics.

Patterns from the rubric

The patterns run in the newsletter.

Every applicant is scored against 47 rubric questions across nine sections, with win-rate and fail-rate measured against the anchor set and Wilson 95% confidence intervals on both. We publish one pattern at a time in The Shortlist, each Thursday, with the effect sizes and the named companies behind it.

Read the newsletter

Sources of truth

Where the data actually comes from.

Primary regulatory filings. What companies say about themselves in documents they file under penalty of fraud. Used for outcomes, founder narratives, and disclosed financial context.
Public alumni and status directories. Outcome-labeled records where the status field is verifiable.
Disclosed venture portfolios. Public portfolio pages from established funds, cross-checked against status directories.
Federal court records. Bankruptcy and dissolution evidence for failures where no public exit exists.
Founder interviews and depositions. Required as one of three primary sources for every hand-researched anchor entry.

Wikipedia is not a primary source for The Shortlist's research database. Neither are press releases, sponsored blog posts, third-party data aggregators, or AI-generated summaries.

How outcomes are defined

Public records only. No opinions.

Winner, public. The company has completed an IPO and the listing is verifiable in the regulator's record.
Winner, acquired. The acquisition is completed and the transaction is on a regulatory filing or a verifiable public status directory. Undisclosed acquisitions are excluded.
Failure. Bankruptcy on the federal court record, or confirmed inactive in a public status directory with corroborating public dissolution evidence.
Active, unknown. Still operating with no exit event. Excluded from pattern analysis. We do not publish predictions on companies whose outcomes are not yet decided.

A pattern claim about "successful companies" is only as honest as the definition of success. Ours is public-record only.

What we deliberately exclude

Credibility through restraint.

Undisclosed acquisitions. If the transaction terms were not disclosed, the company is not counted as a winner. Quiet acqui-hires are not exits.
Active companies. Excluded from pattern analysis until an outcome is final. We do not publish predictions on companies whose stories are still being written.
Unattributable claims. If a pattern cannot be sourced to either an effect size in the validation set or a primary document in the anchor set, it does not appear in The Shortlist. The label on each pattern is published: Empirical, Documented, or Original.
Survivorship bias as evidence. A pattern that only appears across winners is reported, but it is not called predictive until the inverse pattern is checked in the failure set.

The method, step by step

How the research becomes an introduction.

The research underneath is the slow work. What follows is the operational flow for each application that reaches the editor's desk.

Structured intake

Founders apply through a long-form application covering nine sections: Problem, Solution, Market, Traction, Team, Moat, Why Now, the Ask, and Competition. Free text, no checkboxes. We do not pre-filter by stage, sector, or revenue. The intake itself is the first signal of quality.

A written rubric

Every section is scored against an explicit rubric. Each section defines what Must Read, Might Find Interesting, and Weak look like in concrete terms. The rubric draws on public frameworks from Sequoia, Y Combinator, Andreessen Horowitz, First Round, NFX, and Maveron, layered with the patterns the research database has surfaced. The rubric is updated as outcomes inform what works and what does not.

First-pass scoring

Every application is run through a structured feedback engine. The engine reads the application against the rubric, scores each section, proposes an overall tier, and writes a draft feedback with a specific test the founder should pass next. The engine never makes the final call. It exists to be fast and consistent so the human review is high-quality.

A human read

We read every application personally. We refine the draft into our voice, override anything the engine missed, and stamp the final tier. No application is auto-approved. No application is auto-rejected. The tier you see attached to a founder is ours.

Thesis matching

Each Must Read founder is mapped against the active partner funds' theses, captured when each fund joins. Match means a warm introduction. No match yet means the founder is held in queue until a fitting fund appears or until the founder evolves into another fund's lane. Matching is done by hand, not by keyword similarity.

The introduction

When a Must Read founder matches your thesis, you receive their deck, full application, and a short note from us on why they fit. The note is the kind of thing you can forward to a partner internally without rewriting. The introduction email comes from us, not from a queue or a dashboard.

Outcome tracking

Every introduction is logged. Outcomes are tracked: meeting taken, second meeting, term sheet, closed, passed. As outcomes accumulate, they feed back into the rubric and the criteria sharpen. Closed rounds are the strongest calibration signal when they arrive.

What "qualified" means

Three concrete things, every time.

The founder has been read in full against a rubric distilled from how Sequoia, a16z, Y Combinator, NFX, and First Round actually evaluate, layered with the patterns the research database has surfaced.
The company has been mapped against your specific thesis, not a generic sector tag.
The tier comes with a written rationale you can defend to your partners.

Economics

Who pays what.

Founders never pay. Applying to The Shortlist is free for every founder, in every cycle. There is no path where a founder's place in the queue depends on whether they paid for it. That is the whole point.

Partner funds are free for the next three months. No fees, no minimums, no commitment. After the initial window, partner participation moves to a subscription. Founders stay free either way.

If you would like to receive the founders the system surfaces, apply.

Partner spots are limited and reviewed monthly.

Apply to be a partner fund