# Payout headline referral experiment: a test plan | SourceX

Canonical: https://refer.sourcex.si/guides/what-happens-when-a-referral-page-removes-the-maximum-payout-headline
Last updated: 2026-11-04

Nobody knows yet for this program: SourceX has not run the test, and no results exist. What is known is narrower. Readers can take a headline figure as a typical result: an FTC-commissioned study found that many consumers believe "up to" claims promise the maximum. Our thesis, untested, is that dropping "Earn $1,500,000+" would lower registrations but not qualified introductions. The three-arm design below would settle it on the outcome that matters: introductions that pass eligibility review.

Three kinds of statement run through this page, and they should not be confused.

- **Known.** The reward on a company's first qualifying transaction is the greater of $50,000 plus 5% and 15% of Actual Supplier Payout, with no cap, further qualifying transactions first agreed within the 12-month Repeat Window pay 15%, and SourceX's fees or revenue never enter the calculation. Nothing becomes payable on a lead, a meeting or a signed agreement by itself.
- **Conditional.** Whether any introduction leads to a reward depends on SourceX's review, a completed handoff and the payment conditions in the terms.
- **Thesis.** Everything this page says about how the headline changes behavior. It stays a question until a study answers it.

This page itself carries the program's standard offer block, and the live site's headline leads with "Earn $1,500,000+". That headline is the control arm.

## Terms used in the three-arm headline test

| Term | Meaning here |
|---|---|
| Headline figure | A headline that leads with a large reward number, such as "Earn $1,500,000+" |
| Arm | One version of the landing page, shown to a randomly assigned share of first-time visitors |
| Qualified introduction | An introduced company that submits a verified application and passes SourceX's eligibility review by the cutoff |
| Comprehension score | Correct answers, out of three, on what the reward is calculated on, when it becomes payable and whether the headline figure is typical |
| Open case | An introduction with no final status at the cutoff; reported, never dropped |

## Why the headline is worth testing

**A headline figure is not a typical amount.** The program has no cap, and a company's first qualifying transaction pays the greater of $50,000 plus 5% and 15% of Actual Supplier Payout: USD 100,000 paid yields USD 55,000, USD 400,000 yields USD 70,000 and USD 1,000,000 yields USD 150,000 (illustrative arithmetic, not a forecast). SourceX has published no typical payouts, so nobody outside SourceX can say how often any figure is reached, and this page does not guess.

**Readers can hear a maximum as a promise.** In 2012 the FTC published a commissioned study of replacement-window advertising that claimed savings of up to 47%, and said the research showed many consumers are likely to believe they will achieve the maximum "up to" result. The FTC's position was that advertisers making such claims should be able to substantiate that consumers are likely to achieve the maximum under normal circumstances. That research concerned consumer savings claims, not referral programs. It is still the clearest public evidence that a headline figure can be read as a typical outcome, which is why comprehension is one of the outcomes below.

**Two opposite risks, both plausible.** A large headline figure may attract registrations from people whose networks hold no qualifying companies, which costs everyone screening time; the [poor-fit guide](/questions/industries-poor-fit-data-licensing) lists the patterns that fail. Or removing it may lose the investors and advisors whose networks hold large, long-established companies, who notice the program because one introduction could justify their time. Neither risk has been measured.

**Partner programs disagree about what to count.** In PartnerStack and Wynter's survey of 100 senior B2B SaaS leaders, 42% said they use multi-touch attribution for partner revenue, 31% first-touch and 19% last-touch. A headline test judged on clicks answers a question nobody is paid on. Completed, paid handoffs are what the program rewards, but they arrive months later and are too rare for a page test. The primary outcome is therefore the earliest point that is both defined by the terms and screened by SourceX: a qualified introduction.

## The qualified-conversion headline study

Three arms. Every arm is a true statement of the same terms, shows the calculation base, the no-cap rule and the payment trigger above the fold, links to the [program terms](/terms) and uses the same page body and registration form below the headline.

| Arm | Headline | Line beneath | What it tests |
|---|---|---|---|
| A, control | Intro a company. Earn $1,500,000+ | Default reward on a first qualifying transaction: the greater of $50,000 plus 5% and 15% of Actual Supplier Payout, with no cap | The current framing |
| B, formula first | Earn the greater of $50,000 plus 5% and 15% of what SourceX pays the company you refer | No cap, the USD 50,000 floor once per company, and paid only after the company is paid | Whether stating the calculation base first changes who registers |
| C, fit first | Know a company with 50+ employees and years of records? | Introduce its CEO; reward the greater of $50,000 plus 5% and 15% of Actual Supplier Payout, with no cap | Whether leading with fit filters out poor-fit registrations |

No arm may show a typical or expected reward, because none is published, and no arm may move the calculation base or the payment conditions out of sight.

| Element | Specification |
|---|---|
| Question | Does the headline change qualified introductions per visitor, and how well partners understand the reward? |
| Unit and allocation | First-time visitors to the referral landing page, randomized in equal thirds; assignment sticks for 90 days, matching the attribution window |
| Exclusions | Staff and known bots, removed before assignment; nothing removed after assignment |
| Entry window | A fixed period set in advance, for example ten weeks |
| Primary outcome | Qualified introductions per 1,000 assigned visitors at the cutoff |
| Secondary outcomes | Registrations per 1,000 visitors; introductions per registered partner; share of introductions failing the 50-employee test; duplicates; comprehension score |
| Cutoff | One date, for example 150 days after the entry window closes, chosen to cover the 90-day window plus review time |
| Analysis | By assigned arm (intention to treat); each arm against control with a 95% interval; Holm correction for the two comparisons |
| Stopping | Fixed duration; no stopping early because a result looks good |
| Preregistration | This table, the analysis code and the launch date published before the first visitor |

**Consent and comprehension.** After registering, partners are offered an optional three-question check, with consent asked first and answers stored against the arm, not against any company. Skipping is recorded as missing, not as wrong. The questions:

1. The reward is calculated on (a) what SourceX actually pays the referred company, (b) SourceX's fees, (c) the buyer's price. Correct: (a).
2. A reward becomes payable after (a) a meeting, (b) a signed agreement, (c) conditions that include the company being paid. Correct: (c).
3. The $1,500,000+ figure is (a) the typical reward, (b) an illustration of what the uncapped program can pay, not a typical result, (c) a minimum. Correct: (b).

Nobody is deceived by the design: every arm describes the same terms, and every partner sees the same terms after registering, whichever headline they saw.

## Complete outcomes, including the cases that went nowhere

The result is the whole table, per arm, at the cutoff. A study that reports only the winning ratio hides the cases that explain it. This shell is the reusable part: copy it, fill it, publish it with the cutoff date.

| Count at cutoff | Arm A | Arm B | Arm C |
|---|---|---|---|
| Visitors assigned | Not yet collected | Not yet collected | Not yet collected |
| Registered, never introduced a company | Not yet collected | Not yet collected | Not yet collected |
| Introductions submitted | Not yet collected | Not yet collected | Not yet collected |
| Duplicates or outside the attribution window | Not yet collected | Not yet collected | Not yet collected |
| Verified applications | Not yet collected | Not yet collected | Not yet collected |
| Passed eligibility review | Not yet collected | Not yet collected | Not yet collected |
| Declined in review | Not yet collected | Not yet collected | Not yet collected |
| Withdrawn by the company | Not yet collected | Not yet collected | Not yet collected |
| Still open | Not yet collected | Not yet collected | Not yet collected |
| Comprehension check skipped or declined | Not yet collected | Not yet collected | Not yet collected |

Two reporting rules. First, open cases stay in the denominator and the result is shown as a range they allow, as the [conversion benchmark method](/guides/data-referral-conversion-benchmark) sets out. Second, only the prespecified comparisons between arms are causal. Cuts by partner role, traffic source or company size are descriptive, labeled as such, and never used to declare a winner. Decline reasons belong in the separate [rejection reasons study](/guides/data-referral-rejection-reasons-study).

## How many visitors the test needs

Illustrative inputs, invented to show the arithmetic; SourceX has published no conversion rates. Suppose the control arm produced 4 qualified introductions per 1,000 visitors and you wanted to detect an arm producing 6. With two-sided 5% significance and 80% power, the standard two-proportion formula gives about 19,500 visitors per arm, roughly 58,600 across three arms, before the correction for two comparisons. Halve the baseline rate and the requirement roughly doubles.

That arithmetic is the most important stress test in the design. If the site cannot reach the required traffic in the entry window, the primary outcome will be underpowered, and an honest preregistration says so in advance: report qualified introductions descriptively, with intervals, and power the test on comprehension, which every registering partner can answer. A small test that announces a winner on four qualified introductions is worse than no test.

## Stress test: a counterexample, the limits and what would change the answer

**The counterexample, as a question.** Could removing the headline figure lose the partners who matter most? A PE operating partner with access to thousand-person portfolio companies may register precisely because one introduction could be worth the effort, while a partner with a small network may respond better to arm C's fit-first framing. If arm C produces fewer registrations but the same qualified introductions, the headline figure mostly attracted poor fit. If arm C produces fewer qualified introductions, the headline figure carries real signal for high-value introducers. The design can tell these apart; intuition cannot.

**Limits.**

- **Contamination.** Partners copy headline wording into emails and posts, so visitors may meet another arm's message off the page. Record the referrer of each visit and report the share arriving from partner-shared links.
- **Traffic mix.** A newsletter mention in week three changes who arrives. Randomization protects the comparison, not how far it generalizes.
- **Missing assignment.** Visitors who return on a new device may be counted twice; report the estimated share.
- **Time.** Qualified introductions take months; the cutoff is a compromise between speed and completeness.
- **One program.** A result on refer.sourcex.si says little about other programs with different terms.

**Disclosure travels with the wording.** The FTC's endorsement guidance says affiliate relationships should be disclosed close to the links, treats "Paid link" as adequate and "Commissionable link" as probably not clear, and expects companies running affiliate networks to tell affiliates how to disclose. Whichever arm wins, partners who reuse its wording still disclose that they may earn a reward; see [whether you have to disclose a referral fee](/questions/do-i-have-to-disclose-a-referral-fee).

**Decision rule, fixed in advance.**

| Result | Decision |
|---|---|
| A matches or beats B and C on qualified introductions, and question 3 scores are no worse | Keep the current headline |
| B or C matches or beats A on qualified introductions, with better question 3 scores | Propose the new headline to the owner |
| Intervals overlap on every outcome | Inconclusive; keep the current block and say so |
| Registrations fall but qualified introductions do not | Report that the headline figure mostly attracted poor-fit registrations |

Changing the live offer block is the owner's decision. It stays on every page until a study like this has been run, published and reviewed.

**A measurable next step.** Before any split test, add only the three-question check to the current page for a set period. If nearly every registering partner already answers question 3 correctly, the headline matters less than this page assumes; if many answer (a), the test is worth its traffic.

## Which introductions count as qualified

The eligibility pass in the primary outcome is SourceX's, applied to the published fit facts: a peak of 50+ full-time employees, not counting contractors; an operating history backed by documents; rights to license the records; and an authorized contact, the CEO, who wants the conversation. 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.

Partners never handle a company's records; discovery stays at metadata level (which records, how many, how many years, which system, who administers it). What stays unknown until SourceX reviews a company: whether its records match current demand, what the rights review finds and what any license would cover. Paid handoffs are a follow-up outcome rather than the primary one: payment is due 60 calendar days from whichever is later, completed handoff or the day the Acquisition Partner gives its final written acceptance of the complete, validated dataset, and both events land far beyond any page test's cutoff. Estimates from the [referral earnings calculator](/tools/referral-earnings-calculator) are projections, not promises, and the [expected earnings worksheet](/guides/referral-expected-earnings-calculator) asks you to supply your own odds rather than borrow a headline.

## 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.

Read past the headline before you introduce anyone: [open a partner account](/register), measure the company against the [qualifying criteria](/who-qualifies) and work through the full [program terms](/terms), so that your own introductions would count as qualified in any study of this page. Related methods sit in the [guides hub](/guides).

## Sources

- https://refer.sourcex.si/terms
- https://refer.sourcex.si/who-qualifies
- https://www.ftc.gov/news-events/news/press-releases/2012/06/ftc-report-many-consumers-believe-claims-promise-maximum-results
- https://www.ftc.gov/reports/effects-bristol-windows-advertisement-savings-claim-consumer-take-away-beliefs
- https://www.ftc.gov/business-guidance/resources/ftcs-endorsement-guides-what-people-are-asking
- https://partnerstack.com/resources/research-lab/charts/multi-touch-is-the-most-common-partner-tribution-method-for-senior-leaders-in-b2b-saas