The experiment behind a better partnership landing page: a frame test that keeps every sentence identical, shows it either as a portfolio perks listing or as a plain explainer, and scores qualified conversions beside the founders who think the fund won them a better deal

Short answer

Test the frame, not the words. A platform may be tempted to list SourceX in its portfolio perks catalog, beside negotiated discounts, and a perks catalog tells readers the fund bargained for something. Publish the same facts in two settings, a catalog listing and a plain explainer, send each founder to one by a logged coin flip, and judge the result on qualified conversions (a CEO's own request that becomes a verified application) and on how many founders come away thinking the fund got them a better deal.

Visual briefFull-size diagram
Two page mockups side by side carry the same four grey text lines. Page A sits in a grid of perk tiles marked as other offers; page B stands alone with a resources label. A coin between them assigns each founder to one page, and both pages feed two meters at the bottom: qualified conversions in green and better-deal misreadings in orange.
The design, not a result. Pages and meters are Illustrative; no founder, partner or SourceX results for this test are published.

Earlier tests changed other things. The landing page test for different decision-makers changed who each version addresses, the personal link comparison changed the path a note takes, and the consolidation decision settles what to do with the pages afterwards. In this test not one word changes. Only the setting around the words does.

What a fund's page may call the relationship before any test

Settle the name first, because a "partnership" page can be untrue before anyone reads it. Section 4.5 of the sample Channel Partner Agreement says: "Neither Party may publicly identify the other as a partner or disclose non-public deal details without written authorization, except as legally required." Section 7.1 adds: "Participation creates no partnership or employment relationship." So the fund asks SourceX in writing how the page may describe the relationship, and both versions use that approved wording unchanged.

The page also has jobs it must never do. It may not suggest that a company qualifies, that SourceX endorses the fund, that coming through the fund buys preferential access, or that the fund can decide anything for a company. SourceX's screening page warns that a portfolio-wide line saying "our companies qualify" implies an acceptance nobody has given (our paraphrase). Each company's CEO decides whether to talk, and SourceX decides fit after its own review.

The fact block both versions carry word for word

Write the block once, have counsel and SourceX's approved wording checked against it, then freeze it for the test. Neither version may add, cut or reorder a sentence. These are the sentences:

  1. Who decides. SourceX reviews each company; the company's CEO decides whether to talk and signs for it.
  2. What the fund is. A registered referral partner in SourceX's program, described in SourceX's approved words, that may be paid by SourceX if a company it introduced licenses data and the terms' conditions are met. The fund cannot commit SourceX to anything.
  3. What the company is paid. The same amount whether it comes through the fund or applies on its own. Section 4 of the terms puts it this way: "The referral reward is paid separately by SourceX and is never deducted from the supplier's agreed compensation."
  4. No special access claimed. The page promises no priority review, no different terms and no pre-approval: "may be a fit; SourceX decides after review".
  5. What nobody at the fund sees. Records, files, samples or logins.
  6. Who may be a fit. The published fit facts and the geography sentence, quoted.
  7. Three ways forward. Ask the fund for an introduction, apply directly, or do nothing.

Two frames, one set of sentences

Version A is a listing in the platform's perks catalog: a tile in the grid with the catalog's card layout, opening a detail page in the catalog's template. Version B is a standalone explainer in the platform's resources section, laid out like any other article. Everything a catalog template normally adds that would be untrue here comes out of version A before launch: "exclusive", "portfolio discount", "redeem", "claim your offer", "priority" and "pre-approved". What remains is the variable under test, the suggestion carried by the catalog setting itself.

That is a real question for a platform team. The catalog is where founders are used to finding what the fund negotiated, and comparing a data referral with other platform benefits shows why platforms are tempted to file it there. If the setting alone makes founders think they got a deal, the frame is doing the misleading even though every sentence is true.

The qualified-conversion experiment

Every element below is fixed before the first invitation goes out. The search rules follow Google Search Central's guidance on testing pages, which recommends a canonical link on alternate URLs, a temporary redirect rather than a permanent one, and no cloaking (our paraphrase). Credit is never read from the page either: section 2 of the terms says an earlier link click alone does not decide credit, and attribution comes from dated records.

ElementSpecification
HypothesisThe catalog frame produces more founders who think coming through the fund changes their terms, and no more qualified conversions than the explainer
UnitOne founder-CEO the platform invites, whose company passed a metadata screen against the published fit facts
Version ACatalog listing: tile and detail page in the perks catalog layout
Version BExplainer: standalone page in the resources section
Held constantThe fact block, the invitation note, the sender, the sending weeks and the approved description of the relationship
AssignmentA coin flip per CEO, logged before the invitation is written; the invitation links only to the assigned version
Search hygieneThe alternate URL carries rel=canonical to the original, any redirect that routes test traffic is a temporary 302, and crawlers see exactly what people see
Primary measureQualified conversions per CEO invited: the CEO's own written request for an introduction, or own application, that becomes a verified application SourceX does not return as a duplicate
Trust measureBetter-deal misreading: when a CEO requests an introduction, they answer in their own words whether coming through the fund changes what SourceX pays their company; a second person codes each answer correct, wrong or unsure without knowing the version
Guard measures"Are we approved?" questions; applications started without the CEO's yes; any record or file sent to the fund; complaints and do-not-contact requests
Reading dateAfter every 90-day attribution window from the last invitation has closed, plus any review time the plan sets
Never countedPage views, clicks, time on page, tile impressions

Floor, stop rules and the decision rule for the page test

Set a floor before launch, such as 25 invited CEOs per version (Illustrative). Below it there is no winner, only counts. Four stop rules apply in both arms at once: a founder says they were told they would get better terms, priority or approval (pause both versions, fix, restart the count); any record or file reaches the fund (stop, delete, log); SourceX asks for a wording change (pause until made); complaints pass a number set in advance (pause and review the invitation volume). A conversion built on a misreading is not counted as qualified, which is why the trust measure can overrule the primary one.

Result on the reading dateDecision
The explainer matches or beats the catalog on qualified conversions, with fewer misreadingsKeep only the explainer; the catalog tile becomes a plain link to it, or goes
The catalog beats the explainer on qualified conversions by more than the preset margin, with no more misreadingsKeep the catalog listing and recount misreadings every quarter
The catalog leads on conversions but carries more misreadingsThe explainer wins
Below the floor, or zero in both armsNo winner; keep the explainer, the frame least likely to suggest preferential access

Which founders to invite into the test

Invite only CEOs whose companies could plausibly fit, or the test measures curiosity rather than conversion. The published facts are a full-time headcount that peaked at 50 or more with contractors excluded, years of operating records of the company's own, the right to license those records, and a CEO who would welcome 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.

The screen runs on what the platform already knows: what records exist, roughly how many, since when and in which system. Nobody asks for files or logins. Funding stage is not a criterion, so network companies the fund never backed can be invited on the same terms. Whoever submits an introduction during the test must already have signed the SourceX Channel Partner Agreement, as section 1 of the terms requires.

Reading a small test, and publishing what did not work

With 25 CEOs per arm, a split of three conversions to one is a story, not a result; how to avoid declaring a winner from five replies works through why. Report counts per arm, the misreadings in the CEOs' own words where they agree to it, every pause and its cause, and the crossovers. Analyze each CEO in the arm they were assigned, whatever page they later found.

The design has limits worth stating in the write-up. The misreading question reaches only CEOs who request an introduction, so it cannot see founders who misread and walked away. Founders in one cohort share links, so some will see both versions. Catalog layouts differ between platforms, so one fund's result may not carry to another's. And SourceX's review decides every verified application, on criteria that follow its current programs.

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 page goes live, register as a partner under the entity that will submit introductions, ask SourceX in writing how the relationship may be described, check the invited companies against who qualifies and keep the program terms open while you freeze the fact block. More test designs sit in the partner guides.

A landing page test is referral work; deciding whether SourceX is worth backing is another job entirely. The second belongs on SourceX's investor page, where investor materials can be requested and the company says it hands company and commercial metrics to investors directly.

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.

  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 the perks version keep words like exclusive or priority if the rest of the copy is accurate?

No. Any company can apply to SourceX directly, the company is paid the same with or without the fund, and the fund has no basis to promise priority, so those words would be untrue. Strip every template element that implies a negotiated benefit; the test keeps only the catalog's placement and layout.

Why ask the misreading question only when a CEO requests an introduction?

Because that is the moment a misreading does harm: a CEO who goes ahead believing the fund secured better terms has been misled at the point of decision. Asking every visitor would mean tracking people who never made contact, which the test avoids.

Does the fund need SourceX's permission to run the test?

The test itself needs none, but the page's description of the relationship does. The sample Channel Partner Agreement bars publicly identifying the other party as a partner without written authorization, so get the wording approved first and use it, unchanged, in both versions.

Sources

  1. refer.sourcex.si/terms
  2. sourcex.si/documents/partner/SourceX_Channel_Partner_Ag…
  3. refer.sourcex.si/who-qualifies
  4. developers.google.com/search/docs/crawling-indexing/website-testing
  5. sourcex.si/investors

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