# Test eligibility messaging, hide no restrictions | SourceX

Canonical: https://refer.sourcex.si/guides/how-to-test-eligibility-messaging-without-hiding-restrictions
Last updated: 2026-11-04

Test framing, never content. Write two versions of one introduction that state every published SourceX condition in the same order, using SourceX's exact wording wherever it publishes one, and change a single thing: whether the conditions read as what a company needs or as what rules a company out. Assign CEOs at random, judge the versions on qualified next steps and on founders who later felt misled, and stop the moment either version drops or softens a condition. This page is the design; no results are published here.

It is written for platform growth leads at venture firms who send SourceX introductions across a network and want to know which wording helps the right CEOs say yes without blurring who can take part. Earlier tests on this site compared [two opening angles](/guides/how-to-compare-two-outreach-angles-without-changing-everything) and [a fit checklist against a plain message](/guides/does-a-company-fit-checklist-improve-application-quality). This one holds the conditions fixed and asks whether their framing changes who comes forward.

## One variable: how the same conditions are framed

**Question.** When both versions carry identical conditions, does framing them as limits change how many CEOs whose companies fail a stated condition ask to proceed, without changing how many qualified CEOs do?

**Hypotheses, two-sided and written down before launch.** H0: framing makes no difference to misfit asks or to qualified next steps. H1: limits-first framing changes misfit asks in either direction, and does not lower qualified next steps beyond the margin set in advance.

**Unit and cohort.** One company, reached through one CEO whom someone on the platform team knows personally. Draw the full list before writing either version, and split it at random within each sender's contacts, so every sender sends both versions.

**What counts as hiding.** Dropping a condition, moving it behind a link, shrinking it into a footnote, saving it for a second email or softening it ("generally", "ideally") all hide it. FTC staff guidance gives a useful bar for this, written for advertising but apt as a test standard: "Clear and conspicuous" is "a performance standard, not a font size," judged by whether people notice, read and understand the disclosure (our paraphrase of the post). Its digital guidance adds a channel rule worth adopting outright: "If the disclosure cannot be made clearly and conspicuously on a device or platform, then that device or platform should not be used." Neither text states what law requires of a referral message; they set the standard this test holds itself to.

## The accurate-message variant pair

Both versions share a subject line, sender, greeting, description of SourceX, disclosure, links and sign-off. Only the condition block differs, and only in framing.

| # | Published condition | Version A: what a company needs | Version B: what rules a company out |
|---|---|---|---|
| 1 | Size | At its peak, the company had 50 or more full-time employees; contractors are not counted | A company whose full-time headcount never reached 50, contractors left out, is not a fit |
| 2 | History | Several years in operation, with records the company created itself | A company without several years of its own records is not a fit |
| 3 | Rights | It owns the records, or can license them without breaching client confidentiality, employee privacy or third-party contracts | Records it cannot license without breaching client confidentiality, employee privacy or third-party contracts are out |
| 4 | Earlier AI licensing | The data has never been licensed for AI training | Data already licensed for AI training is out |
| 5 | Who decides | The CEO decides whether to talk, and signs for the company | Nothing moves without the CEO; an introduction the CEO has not agreed to is not a fit |
| 6 | Geography | 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 same sentence, word for word |
| 7 | Program fit | SourceX checks in its own review whether the records match a current program | A company that clears every check above can still be declined if it matches no current program |
| 8 | Usual poor fits | Personal consumer data, scraped data, client-owned material and one-off files are usually not a fit | The same sentence, word for word |

The disclosure line is identical in both and never framed: "If you decide to apply, SourceX may pay me a reward; your company would receive the same amount either way." That last clause rests on section 4 of the terms, which says the reward is paid separately and never deducted from the company's agreed compensation.

**The parity audit.** Before the first send, a colleague who did not write either version checks five things and signs the list: every row appears in both versions; rows appear in the same order; the geography sentence is identical to SourceX's published one; the two blocks differ in length by no more than a tenth; and nothing outside the block differs. Rows 1 to 8 come from SourceX's screening and how-it-works pages, checked on the audit date. Repeat the audit if either page changes during the test. [The geography explainer](/guides/how-to-explain-sourcex-geography-without-turning-companies-away) shows what row 6 must never be stretched into.

## What to measure: qualified next steps and founder trust

The test records only what CEOs volunteer. Nobody asks a CEO for headcount, systems or records to score it, and nothing beyond the introduction itself reaches SourceX.

| Measure | Definition | Denominator |
|---|---|---|
| Qualified next step (primary) | The CEO personally asks in writing for the introduction, or says the company will apply, with no statement in the same exchange of a fact the message lists as ruling a company out | CEOs messaged in that version |
| Misfit ask (primary) | The CEO asks to proceed while volunteering a fact the message already lists as ruling a company out, such as a 30-person team | CEOs messaged in that version |
| Late surprise | A CEO who proceeded reports learning, later, a condition the message already stated | CEOs who proceeded |
| Felt misled | A "no" to "Did the message describe who can take part fairly?" or a "yes" to "Did anything about who can take part surprise you later?" | CEOs who answered the two follow-up questions |
| Harm signal | A complaint or a request to stop receiving messages | CEOs messaged in that version |

Send the two follow-up questions to every CEO who replied, yes or no, after their decision, and only with their agreement. Opens and clicks are not measured: a click decides nothing, which is also why the terms never let a link click alone settle credit.

## Stop rules and the decision rule, fixed before the first send

**Stop rules.**

1. **Parity breach.** Any sent message found missing a condition, softening one or altering the geography sentence: stop both versions, set aside every send so far and restart after a fresh audit.
2. **A founder felt misled.** Pause both versions, have someone outside the test reread both blocks, and resume only once the cause is found and fixed in both.
3. **Implied acceptance.** Any reply showing a CEO read either version as saying the company qualifies, that SourceX endorses it or that the network can decide for it: stop.
4. **SourceX changes its wording.** If the screening page or the terms change mid-test, the conditions changed and so did the test: stop and start again.
5. **A channel cannot carry the block.** Never send either version where all eight rows cannot appear in full.
6. **Ceiling.** Stop at the CEO count and end date written down before launch.

**Decision rule.** Write the margins as numbers before launch. As an Illustrative example of the form: adopt version B only if it shows at least three fewer misfit asks per 40 CEOs, qualified next steps no more than two lower per 40, and no more felt-misled answers than version A. Adopt A on the mirror condition. Anything else is inconclusive: keep the version already in use and report the test as inconclusive. With a few dozen CEOs, most honest outcomes will be inconclusive, and [the five-replies arithmetic](/guides/how-to-avoid-declaring-a-winner-from-five-replies) shows why a small split proves little.

## Reporting what happened, including what did not work

Report counts with their denominators, never a percentage on its own, and report every stop rule that fired.

| Line | What to report |
|---|---|
| 1 | CEOs drawn, messaged and excluded, per version, with reasons for exclusions |
| 2 | Qualified next steps and misfit asks, per version |
| 3 | Late surprises and felt-misled answers, with the count who answered |
| 4 | Harm signals, per version |
| 5 | Parity audit dates and results |
| 6 | Stop rules that fired, and what was changed |
| 7 | The decision, the pre-set margins and whether they were met |
| 8 | What did not work: wording that confused, CEOs who complained, rows nobody understood |

A lower count of applications in one version is not a failure if qualified next steps held; [the screening cohort analysis](/guides/when-better-screening-lowers-applications-but-improves-outcomes) explains why the denominator matters more than the pile. SourceX publishes no results from this design.

## Risks and limits of a framing test

- **Small networks.** Most platform teams can message dozens of CEOs, not hundreds, so the likeliest verdict is inconclusive. That is a result, not a failure.
- **Contamination.** CEOs in one network compare notes. A founder who sees both versions can blur the comparison; note it and keep them in their assigned version.
- **Self-reported trust.** The two follow-up questions measure what founders say, after the fact, among those who answer.
- **Moving conditions.** The conditions are SourceX's, follow each program and can change; a winning frame becomes wrong the day the wording does.
- **Fit is decided elsewhere.** The test measures messages, not companies. Only SourceX's review decides whether a company fits, and the screening page warns that a line such as "our companies qualify" implies an acceptance nobody has given.
- **No benchmark.** Nobody has published framing results for data licensing introductions, so there is nothing to compare your counts against.

**Methods note.** The variant pair, measures and rules were built on 2026-10-28 from SourceX's screening page, how-it-works page and program terms, with FTC disclosure guidance borrowed as a test standard. They describe a design, not measured results. Whoever sends the introductions signs the SourceX Channel Partner Agreement before submitting any, and every introduction concerns one company whose CEO agreed to it.

## 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 the first send, [register](/register), copy rows 1 to 8 fresh from [who qualifies](/who-qualifies) on the audit date, and have the sender read the [program terms](/terms); other network experiments are collected in the [guides hub](/guides).

Testing how a network describes SourceX's conditions is referral work. Judging SourceX as a company to invest in is a different task, and the [SourceX investor overview](https://sourcex.si/investors) is where investor materials are requested.

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/who-qualifies
- https://sourcex.si/how-it-works
- https://refer.sourcex.si/terms
- https://www.ftc.gov/business-guidance/blog/2014/09/full-disclosure
- https://www.ftc.gov/system/files/documents/plain-language/bus41-dot-com-disclosures-information-about-online-advertising.pdf
- https://sourcex.si/investors
