How to test a referral CTA without creating false urgency: an accurate CTA variant test that audits every closing ask before it is sent, changes one thing at a time and counts a clear no, a withdrawn yes and a CEO who felt rushed beside every agreement

Short answer

Test only closing asks that would still be true if the CEO waited a month. Run every candidate through a five-gate audit covering time, scarcity, value, approval and a truthful description of the next step; pair two survivors that differ in one respect; and judge them on informed decisions rather than clicks. Agreed introductions and clear declines count for a version, while a CEO who felt rushed, a yes later withdrawn or a request that ignores a stated fit condition counts against it. Nothing in SourceX's published terms or screening page gives a company a date to apply by, so no CTA may imply one.

Visual briefFull-size diagram
Five candidate closing asks run toward an audit column of five gates labeled time, scarcity, value, approval and step. Three asks, apply within 90 days, few spots left and see what it's worth, are struck through in orange and stop at the audit. Two asks, introduce you and the same ask plus an exit line, pass through to boxes A and B, beside a scoreboard that counts agreed and clear no for a version and rushed and withdrawn yes against it.
Illustrative. The candidate asks are examples of the audit, not tested wording; no results exist yet.

This page is for partner growth managers who run outreach for a firm's client relationships and want to know which closing line helps the right CEOs act. A CTA test that scores only yes rates will drift toward pressure, because pressure raises yes rates for a while. The design below builds the guard into the measure.

The closing ask is the variable, and nothing else moves

The CTA here means the last sentence that asks the reader to do something, plus any button or link label that carries the same ask. Everything above it stays fixed: sender, subject, reason for writing, description of SourceX, conditions, disclosure and links.

Neighboring pages cover the rest of the message. The false urgency review edits every time claim in a whole draft by asking whose clock it runs on. The eligibility messaging test varies how the conditions block is framed. The calculator test deals with anchoring on the largest reward. This test starts once a draft has passed that review and asks a narrower question: of two honest closing lines, which leads more CEOs to a decision they understand?

Five gates for every candidate ask

A candidate enters the test only if it passes all five. Reviewers who did not write it run the audit and sign it.

GateQuestion for the askIt fails whenBasis
1. TimeDoes it set or hint at a time limit?Any reply-by date, "before the window closes", "this quarter"No published SourceX rule gives a company a date to apply by; the 90-day window in section 2 of the terms decides which partner is credited, not when a company must act
2. ScarcityDoes it imply a limited number of places?"Few spots", "only some companies will be chosen", "while programs are open"SourceX's screening page says data needs differ by program and a company that misses one may fit another later
3. ValueDoes it promise a price, valuation or reward figure?"Find out what your data is worth", any amountSourceX describes its early estimate as indicative and labels its fit illustration "Not a valuation"
4. ApprovalDoes it imply the company is accepted, selected or endorsed?"Pre-approved", "you've been picked", "just confirm"Fit is decided in SourceX's own review
5. StepDoes it describe the next step truthfully and leave a free exit?"Takes two minutes" without a basis; no way to wait or declineSourceX describes its first step as a few questions, no files and an indicative estimate

The FTC's announcement of its 2022 staff report on dark patterns describes countdown timers that make people believe an offer is time-limited when it is not (our paraphrase). It was written about consumer shopping, not referral messages, and is borrowed here as a test standard: if a closing line would make a CEO believe a door is closing, and no door is closing, it fails gate 1.

The accurate CTA variant test

Each pair changes one thing. Run one pair at a time; a second pair starts only after the first is read. For the pair under test, write the hypotheses before launch, two-sided: H0, the closing line makes no difference to decided answers, agreed introductions or the guard measures; H1, it changes decided answers in either direction, and a change counts only if no guard measure rises.

PairVersion AVersion BThe single differenceWhat it tests
1. Exit line"Would you like me to introduce you to SourceX?"The same, followed by "If now is not the moment, there is no need to reply."An explicit no-cost exitWhether naming the exit lowers agreed introductions or only lowers pressure
2. Size of the ask"Would you like me to introduce you to SourceX?""Would a 15-minute call with me first be useful?"Introduction now, or a conversation firstWhether a smaller first step leads to more informed decisions
3. Who moves first"Would you like me to introduce you to SourceX?""If the published criteria look right, the company fit checker gives a preliminary screen; tell me afterward if you would like an introduction."The CEO screens the company before askingWhether self-screening cuts requests that ignore a fit condition

Pair 3 points to the company fit checker, which SourceX describes as preliminary and non-binding; the line must keep both words or their meaning.

Asks rejected at the audit (Illustrative). These never enter a test, whatever they might do to yes rates.

Rejected askGate failedWhy
"Apply within 90 days so my introduction counts"TimeThe partner's credit window, presented as the company's deadline
"Only a few companies are being taken on this quarter"ScarcityNo published quota or quarter exists
"See what your records are worth"ValueNo valuation exists before review, and pricing stays indicative until a Dataset Deal Confirmation is signed
"You're pre-qualified; just confirm your interest"ApprovalNobody has qualified the company
"Click below; it takes two minutes"StepThe time is unsupported, and the application involves describing systems and years of history

What both versions keep. A disclosure sentence saying SourceX may pay the partner a reward separately and the company's payout would be the same either way, which rests on section 4 of the terms; one ask per message; no reward figure in the closing line.

Who receives which closing line, and what they agree to

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

Consent. The message is an ordinary personal ask. Record the reply and its date. After the CEO's decision, and only with their agreement, ask two questions: "Did the message make you feel you had to decide quickly?" and "Was it clear what would happen if you said yes?" A CEO who asks to be left out is removed, and only the removal is counted. Nobody collects company records, systems or headcounts to score the test.

Measures that let a clear no count as success

MeasureDefinitionDenominator
Decided answer (primary)A written yes or no from the CEO by day 21CEOs messaged in that version
Agreed introduction (primary)The CEO asks in writing to be introduced, or says the company will applyCEOs messaged
Felt rushed (guard)"Yes" to the first follow-up questionCEOs who answered
Withdrawn yes (guard)A CEO agrees, then retracts within 30 daysCEOs who agreed
Ignored-condition request (guard)The CEO asks to proceed while stating a fact the message lists as ruling a company outCEOs messaged
Objection (guard)A complaint or a request to stop receiving messagesCEOs messaged

Opens and clicks are not measured. The terms say an earlier link click alone does not decide credit, and a click on a closing line shows curiosity, not a decision.

Stops and the reading rule, fixed in advance

Stop rules.

  1. A sent ask fails a gate on re-audit. Halt both versions, set aside every send so far and restart after a fresh audit.
  2. One CEO felt rushed. Pause both versions; someone outside the test rereads both closing lines before any further send.
  3. An approval read. Any reply taking either ask as acceptance or endorsement: halt.
  4. An objection. Pause and review.
  5. SourceX changes a page the ask relies on. The fit checker description or the terms change: halt and re-audit.
  6. Ceiling reached. Stop when either the planned number of CEOs or the planned end date arrives.

Reading rule, in order. First, version B may not exceed A on felt rushed, withdrawn yeses or objections; a version that raises any guard loses, whatever else it does. Second, B must beat A on decided answers by the margin written before launch, for example at least three more per 30 CEOs. Third, B may not trail A on agreed introductions by more than the margin set for that, for example two per 30. Anything else is inconclusive and the current ask stays. The tie-break, also written in advance: when the counts cannot separate the versions, keep the one with the explicit exit.

Declines, silence and requests that ignore a condition

Every CEO messaged stays in the denominator of the version they received. Silence is an undecided answer, not a no. A decline is a decided answer and counts for the version, because the CEO understood enough to choose. A request that ignores a stated condition counts against the version, even though it looks like a yes.

Because assignment is random, a clear difference can be read as caused by the closing line, but only for this sender group, this period and this wording. It does not tell you what would happen with CEOs who do not know the sender.

Reading uncertainty and choosing what to change

A partner firm can usually message a few dozen CEOs, so most honest results will be inconclusive. The five-replies arithmetic shows how easily a small split misleads, and the qualified reply rubric shows how to score replies blind. An inconclusive read is still useful: if neither version raised a guard measure, either is safe to keep, and the tie-break decides.

Report every count with its denominator, every stop rule that fired, every audit date, and what did not work, including asks that confused CEOs and any CEO who said they felt pushed.

Methods note. The gates, pairs, measures and rules were written on 2026-10-28 from SourceX's program terms, screening page and how-it-works page, with the FTC's dark patterns announcement borrowed as a test standard. They describe a design. SourceX publishes no results from it.

What the closing line can never change

No ask changes who decides or whether a company fits. Agreeing to a conversation with SourceX, and signing afterward, stay with the CEO. On fit, SourceX looks at peak full-time headcount (50 or more, without contractors), years of records the company produced itself, its right to license them and whether they match a program running now. 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. A sender needs a signed SourceX Channel Partner Agreement before the first introduction is submitted.

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 auditing the first candidate ask, register, keep who qualifies open for gates 2 and 4, and read section 2 of the program terms so the credit window stays on your own task list and out of the closing line.

  1. Step 1Share your linkSend your personal link to a company you know.
  2. Step 2Company appliesThe company applies itself at /apply.
  3. Step 3Company signs and is paidThe company signs, delivers its data and is paid by SourceX under its agreement.
  4. Step 4You get your rewardthe greater of $50,000 plus 5% and 15% of the actual supplier payout, with no cap. A reward becomes payable after SourceX has paid the referred company under its signed agreement and the payout conditions in the program terms are met, including finance approval and your tax and payment details.

Common questions

Can a CTA mention the company's own system shutdown date?

The date can appear in the reason for writing, attributed to whoever told you, but not in the ask itself. Keeping it out of the closing line stops a real company event from being turned into a deadline for replying to you, and keeps both versions of the test equal.

Why count a clear no as a good outcome?

A CEO who understood the message and declined has made an informed decision, and the message did its job. Counting only yeses rewards the ask that pushes hardest; counting decided answers rewards the ask that makes the choice easy to make either way.

Can the CTA say 'see if your company qualifies'?

Only if the next step is a check rather than a verdict and the message says so: the company fit checker gives a preliminary, non-binding screen, and SourceX decides fit in its own review. Words such as pre-approved or selected fail the approval gate.

Sources

  1. refer.sourcex.si/terms
  2. refer.sourcex.si/who-qualifies
  3. sourcex.si/how-it-works
  4. ftc.gov/news-events/news/press-releases/2022/09/ftc-…

Free resources

Know a company with valuable proprietary data?

Become a referral partner, get your link and introduce the company’s CEO (in an owner-run company, usually the owner). Referral partners can join from most countries. We can't work with partners in countries or regions under U.S. sanctions or on U.S. restricted-party lists. The companies you introduce should be based in the U.S., Canada and the EU, or in the UK, Australia, New Zealand, India, Singapore, South Africa, Latin America, the UAE and Saudi Arabia with at least 70% of their data in English.

Refer a company →

I own a business

Explore licensing your company's data directly to SourceX for AI training and evaluation. Start a short assessment; no uploads needed.

Start an assessment