# Shorter referral brief experiment for founders | SourceX

Canonical: https://refer.sourcex.si/guides/does-a-shorter-referral-brief-improve-founder-response
Last updated: 2026-11-04

Nobody knows yet, and the answer turns on what "response" means. If it means more yeses, a shorter brief can win by leaving out the conditions that make founders say no. Count decided answers instead: yes, no or not now, given within 14 days, with every founder who was sent a brief in the denominator. Keep every condition in both versions, shorten only the explanation, and let a count of late surprises veto the short one. No results exist.

This page is for a platform team that sends portfolio and network founder-CEOs a written brief about SourceX before anyone asks for a conversation, and suspects the brief is too long to be read.

## What this test changes, and what earlier tests already cover

It varies one thing: the length of the first brief a founder receives, before any yes. Neighboring designs answer other questions. [The long and short introduction test](/guides/how-to-test-long-and-short-company-introductions) tests an email sent after a CEO has already agreed to hear more and judges it on qualified applications. [The context-level evaluation](/guides/how-much-context-makes-an-introduction-better) varies how much detail a partner's brief to SourceX carries. [The credibility protocol](/guides/what-makes-a-referral-brief-credible-to-an-executive) has executives score real briefs blind, and [the eligibility messaging test](/guides/how-to-test-eligibility-messaging-without-hiding-restrictions) changes framing while holding length steady.

Here the claims stay fixed and the word count moves. The new element is the measure: whether a founder reaches a decision, and whether a founder who decided quickly later finds a condition they missed.

## Why shorter might help, and why it might not

The case for shorter has a respectable source. The Behavioural Insights Team's EAST framework advises, in paraphrase, that clearer, simpler messages can raise response rates and that a complex goal is better broken into simple steps. Simpler is not the same as shorter, though. A short brief that sends the founder to three linked pages may be harder work than one longer page that answers their questions in order.

Research ethics adds a caution. The Belmont Report, writing about consent to research, holds that how and where information is presented matters as much as the information itself (our paraphrase). A founder's yes to a conversation is not a research consent, but the point carries: a brief that is read and understood beats a brief that is merely complete.

Three predictions, written down before the send:

1. **Short wins.** Busy founders decide faster on one screen, and late surprises do not rise.
2. **Long wins.** Founders wary of anything about their company's data want the detail first and stay silent without it.
3. **No difference.** The relationship owner's name matters more than the brief's length.

## The decision core neither version may drop

Seven lines, identical in both briefs. Each is what a founder needs in order to say no correctly.

| Line | What it says | Its basis |
|---|---|---|
| 1. What SourceX does | It takes a license to a company's own records, dealing with the company itself, for training and evaluating AI | SourceX's process page |
| 2. Who decides | The company's CEO, not the fund and not the board | SourceX's screening page |
| 3. Size and history | At least 50 people on full-time payroll at the company's largest, contractors not counted, and years of records the company itself created | Section 1 of the terms; the screening page |
| 4. Rights and place | Records the company may license, not material belonging to its clients. Companies in the U.S., Canada and the EU qualify, as do companies in the UK, Australia, New Zealand, India, Singapore, South Africa, Latin America, the UAE and Saudi Arabia when at least 70% of the data is in English. | The screening page |
| 5. What is asked now | Only a reply; if interested, the company applies itself and nobody at the fund sees its records | SourceX's process page |
| 6. The fund's interest | The fund could receive a reward from SourceX, which section 4 of the terms says "is paid separately by SourceX and is never deducted from the supplier's agreed compensation" | Section 4 of the terms |
| 7. What nobody knows yet | Fit, rights and any offer are settled only in SourceX's review; a one-word "no" ends the matter with no follow-up | SourceX's process page; the fund's own commitment |

## The brief-length experiment specification

| Version | Length | Contents |
|---|---|---|
| Short | About 150 words, one phone screen | The seven core lines, then links to [who qualifies](/who-qualifies), [how it works](/how-it-works) and the [terms](/terms) |
| Long | About 700 words, two pages | The seven core lines word for word, then what direct licensing means, what SourceX's review looks at (operating history, documentation, data rights, relevance to current demand), the order of steps, one Illustrative company that fits and one that does not, and three questions founders often ask |

| Specification | Entry |
|---|---|
| Unit | One founder-CEO of a portfolio or network company with a named relationship owner |
| Sampling frame | Every such founder on a list frozen on a date, minus companies with contact with SourceX you know of |
| Before assignment | Founders hear in a routine update that the team is trying two formats for optional program notes, and can step out before a set date |
| Assignment | Within each owner's founders, alternate versions in a random order fixed by coin toss and written into the log before the first send |
| Held constant | Sender (the owner), subject line, send within one week, one identical check-in on day 7 to anyone who has not replied |
| Not measured | Opens, read time, clicks; no read receipts or tracking pixels in either version |
| Primary count | Founders with a decided answer by day 14, per founder sent a brief |
| Secondary counts | Yes with no failed fit fact by day 14; verified applications on day 120, too few to decide anything and reported anyway |
| Guard counts | Core surprises, detail questions, misreadings, anything sent to the team from company records |
| Analysis | Each founder stays in the version they were sent, replied or not |

Fourteen days is our choice, long enough to cover a busy fortnight and short enough that the count reflects the brief rather than later events.

**Method note.** The seven core lines restate SourceX's terms, screening page and process page as checked on 2026-10-28; the measures come from the design's question, whether founders reach an informed decision. Nothing here measures demand for any company's records, record quality or payouts.

## A decided answer is the response

| Founder's position on day 14 | Counted as |
|---|---|
| Yes, wants a conversation | Decided |
| No | Decided |
| Not now, with a date or a condition ("after the audit", "if we pass 50 full-time") | Decided |
| "Tell me more" with no decision | Undecided |
| A colleague replied; the founder has not | Undecided, routed to the founder |
| No reply | Undecided |

**Core surprises** are the veto count. At the first conversation, or in any later message, the founder says they did not know something one of the seven core lines states, for example that the fund could be paid, that contractors are left out, that the CEO decides, that client-owned material is usually not a fit, or that nothing is promised. Both versions carried those lines, so a gap means the short format buried them.

**Detail questions** are the expected cost of brevity: questions the long version answered and the short one left to a link, such as what the review looks at. They are reported, never a veto.

**Misreadings** are logged in the founder's own words: "so we qualify," "the fund is selling our data," "the reward comes out of what we're paid." Any record, sample or export sent to the team stops the test until the wording is fixed.

## Rules for a small result, set before the first send

Illustrative sizing: 60 founders, 30 per version.

- **Adopt the short brief** if it shows at least five more decided answers than the long one and no more core surprises.
- **Keep the long brief** if it shows at least five more decided answers, or if the short brief logs two or more core surprises beyond the long one's count, whatever else it shows.
- **No verdict** otherwise. Keep the long brief, because it tells founders more, and record that the test did not settle the question. Why a narrow split is not a result is worked through in [avoiding a winner from five replies](/guides/how-to-avoid-declaring-a-winner-from-five-replies).
- **Stop early** if a founder says the brief or its follow-up felt like pressure, if company records reach the team, or if the short version reaches three core surprises before day 14.

Descriptive and causal claims stay apart. The assignment makes the comparison fair for these founders and these two texts. A founder's explanation, "I said no fast because it fit on one screen," describes that founder; it does not explain the gap.

**Publish what did not work.** Report both versions' counts, the losing text, the founders who stepped out before assignment and the core-surprise log. If founders' understanding is the bigger worry, measure it directly with [the founder comprehension pilot](/guides/a-pilot-that-measures-founder-understanding-before-applications).

## What the brief must never imply about fit

No brief, short or long, establishes that a company fits. The records are the company's own; its CEO decides whether to talk; demand, rights and scope are settled in SourceX's review. A brief that a founder reads as "the fund has checked us and we qualify" has failed whatever its length. Whoever submits an introduction after a yes is a registered partner who has signed the Channel Partner Agreement, and the company describes its own records when it applies.

## Where this specification can mislead

- **Founders compare notes.** Two founders who share a board member may show each other both versions; record it when you hear of it.
- **The owner matters more than the text.** Alternating within owners balances them; it does not remove them.
- **One pair of texts.** A better long brief or a worse short one would change the result. The test compares these two texts.
- **"Not now" can be a polite no.** It counts as decided either way, which is the point of the measure.
- **Day 120 is thin.** Too few applications to read is a finding about scale, not about length.

## Next step

> **Refer a company. No cap on rewards.** Default reward on a company's first qualifying transaction: the greater of $50,000 plus 5% and 15% of Actual Supplier Payout, with no cap; further qualifying transactions first agreed within 12 months of the first Closing pay 15%. Eligibility and payment conditions apply, including the USD 100,000 Minimum Transaction, and a signed Schedule A or addendum can differ. A lead, meeting or signature alone does not trigger payment.

Before either brief goes out, the person who will submit introductions should [register](/register), read both texts against [who qualifies](/who-qualifies) and confirm the reward line still matches the [program terms](/terms).

A brief-length test is about referral practice. A reader asking whether SourceX itself deserves investment is making another decision entirely, and can request materials through the [SourceX investor overview](https://sourcex.si/investors).

This article is for information only and is not an offer to sell or a solicitation of an offer to buy securities. Any offering by SourceX is made only to verified accredited investors through its offering documents.

## Sources

- https://refer.sourcex.si/terms
- https://refer.sourcex.si/who-qualifies
- https://refer.sourcex.si/how-it-works
- https://www.bi.team/wp-content/uploads/2015/07/BIT-Publication-EAST_FA_WEB.pdf
- https://www.hhs.gov/ohrp/regulations-and-policy/belmont-report/read-the-belmont-report/index.html
- https://sourcex.si/investors
