What makes a referral brief credible to an executive? A blinded evaluation protocol that masks briefs partners actually sent, has executives score them without knowing sender, company or outcome, audits every claim in each one, and links the files only after all are locked

Short answer

Nobody knows yet, and this page does not guess. SourceX has published no study of how executives judge a referral brief, the note a partner sends a CEO proposing a conversation about licensing the company's records. The protocol below would find out from briefs partners actually sent: each is masked, scored for credibility by executives who know nothing about its sender, company or fate, audited sentence by sentence for where its claims come from, and only then linked to the dated reply the CEO gave.

Visual briefFull-size diagram
A masked brief with redaction bars sits at the center, surrounded by four role cards: the anonymizer sees the original, the panel sees the masked brief, the claim auditor sees the masked brief and the terms, and the outcome coder sees only the dated reply; a bar beneath reads linked only after all four files lock.
Who sees what in the protocol. No brief has been collected or scored, and no findings are published.

Why real briefs, and why every file is blind

A message experiment varies wording someone wrote for the test. The five-factor trust study does that well, and it can say which wording readers prefer. It cannot say whether briefs that read as credible are the ones CEOs answer, because its messages were never sent. This protocol works the other way round. It takes briefs that were sent, including those that met silence or a no, and asks whether credibility as strangers judge it, and the kinds of claims a brief makes, line up with what happened.

Real briefs carry identities, so every person in the study sees only part of the picture.

RoleSeesNever sees
AnonymizerThe original briefScores, the claim ledger, the outcome
Panel raterThe masked briefSender, company, outcome, ledger
Claim auditorThe masked brief and SourceX's terms in force on the date it was sentScores, outcome
Outcome coderThe partner's dated record of the CEO's reply and whether the company appliedThe brief, scores, ledger

The research lead links the four files only after each is locked, and the analysis code is written and dated before that link is made.

Definitions, fixed before the first brief arrives

TermHow the protocol uses itWhat it is not
Referral briefThe first written message a partner sent a CEO about SourceXLater follow-ups, or the company's own application
Masked briefThe brief with every name, place, distinctive system and identifying detail replaced by a neutral tagA summary or rewrite
Credibility scoreA rater's mean on three statements scored 1 to 7: "What this says is accurate", "The sender is being straight with me", "I could check this for myself"Whether the rater likes it
OutcomeThe CEO's first dated reply (agreed to talk, asked a question, declined, no reply in 30 days) and whether the company applied within 90 days of the introductionWhether SourceX accepted the company

The third statement takes its cue from the first of Stanford's Web Credibility Project guidelines: "Make it easy to verify the accuracy of the information on your site." By the lab's own account (our paraphrase), it is one of ten guidelines drawn from three years of research with more than 4,500 people, and it recommends supporting claims with citations and source material, preferably linked. Whether that holds for a short note to a CEO is one of the things this protocol can test.

The claim ledger: every sentence sorted by where it comes from

The auditor marks each sentence in a masked brief that asserts something, using five codes. This ledger is the protocol's main asset, and partners can use it on their own briefs today.

CodeClaim typeExample (Illustrative)Who could check it
PPublished program fact"Any reward is paid by SourceX separately and does not reduce what your company receives"Anyone, against SourceX's terms
CCompany fact, as the CEO has stated it"You have kept dispatch notes in one system since 2012"The CEO
SThe sender's own position"I may be paid a referral reward if you license data"; "We worked together in 2019"The CEO, by asking
UA stated unknown"I don't know whether your customer contracts allow reuse"Not a claim of fact; it marks calibration
OOverreachAny sentence implying the company qualifies, SourceX endorses the sender, a payout figure, a timeline or special accessNobody: it asserts what SourceX has not decided

Two refinements. A P claim that misstates the terms in force on the day the brief was sent is coded P-wrong; SourceX publishes new versions of its terms, so the auditor checks against the dated version. And a C claim the partner could only have learned from the company's files, rather than from the CEO, is flagged for the research lead, because partners describe records only at the level of type and system and never handle them.

Methods note. The five codes were set on 2026-10-28 from SourceX's terms, its screening page and the Stanford guideline above; tone, length and layout are deliberately not coded.

Code O follows SourceX's own warning that a line such as "our companies qualify" implies an acceptance nobody has given. From each ledger come five numbers per brief: total claims, the share a CEO could check (P, C and S), any P-wrong, any U, and any O.

The blinded brief evaluation protocol

Sampling frame. Every first brief sent by participating partners during two quarters, whatever happened next. Partners submit their send log with their briefs, and the research lead checks one against the other, so nobody contributes only their best work. Partners have signed the Channel Partner Agreement, as the terms require before any introduction.

Consent. The partner agrees to contribute. The CEO who received the brief agrees to research use of the masked version, since it describes their company. Briefs without the CEO's agreement are counted by number and outcome category only.

Masking and the re-identification attempt. After masking, a second person tries to identify the company from the masked brief using public information alone. If they succeed, the brief is masked further; if it cannot be masked without losing its meaning, it is excluded and counted.

Panel. CEOs of companies with 50 or more full-time employees at their peak, recruited through participating networks; other executives may join as a separate group, analyzed apart. No rater scores a brief about a company in their own sector and region, and no rater is introduced to SourceX while the study runs. Each rater scores ten masked briefs in random order, in uniform plain text, and every brief is scored by at least three raters. After the set, raters answer one open question: "Which note did you trust least, and why?"

Covariate recorded before outcomes. The partner rates their relationship with the CEO (worked together, met a few times, met once) when contributing the brief, before the outcome is coded, because the CEO knew the sender and the panel does not.

Analysis plan, dated before the files are linked

  • Agreement first. How closely raters agree on each brief's score is published before anything else. If agreement is poor, the credibility score is not measuring one thing and the remaining analysis is labeled exploratory.
  • Three planned questions. Do briefs with any O claim score lower? Do briefs with a stated unknown score lower, the same or higher? Is a brief's panel score associated with the CEO agreeing to talk, after allowing for relationship strength? Everything else is exploratory.
  • Every brief counted. The report states briefs sent, briefs contributed, CEO consent declined, briefs excluded at masking, and briefs scored, at every step.
  • Counterexamples published. Briefs that scored low and still led to a conversation, and briefs that scored high and met silence, are shown as masked excerpts beside the averages.

What kind of claim this supports. Briefs differ in sender, company, timing and fit as well as wording, so every result is an association. A brief that scored high and led to a call may have succeeded because the company fitted well. A causal claim about one feature needs a test that changes only that feature, which is what the message experiments linked below are for.

What the linked report will show, once there is one

MeasureCompared acrossStatus
Rater agreement on credibilityAll scored briefsNot yet published
Credibility, briefs with and without an O claimBriefs with any O versus noneNot yet published
Credibility, briefs with and without a stated unknownBriefs with any U versus noneNot yet published
CEO agreed to talk, by credibility bandThirds of the panel score, with relationship strength alongsideNot yet published
P-wrong claimsShare of briefs with at least oneNot yet published

Cells with fewer than five briefs are folded into "other", and below 40 scored briefs only counts are reported.

Results that would overturn the advice on briefs

  • Credibility that predicts nothing. If panel scores show no association with CEOs' replies once relationship strength is allowed for, what strangers find credible is not what moves the CEOs who know the sender, and partners should invest in the relationship before the wording.
  • Overreach too rare to test. If almost no brief contains an O claim, the first planned question cannot be answered, and the report says so rather than guessing.
  • Raters who disagree. If agreement is poor, the three statements need rewriting before a second cohort.
  • Unknowns that help. If briefs naming what the sender does not know score as high or higher, the common fear that admitting limits weakens a brief is wrong, and the guidance should say so.

Where a masked-brief study can mislead

  • Masking removes cues a real CEO used, such as the sender's name and the company's own details.
  • Partners who agree to contribute may write more careful briefs than those who do not.
  • A credible brief is not a fitting company: SourceX's review decides fit, and the outcome measure stops at a conversation and an application.
  • Conventions differ by country and sector, so the report describes the cohort's mix and generalizes no further.

From the study back to your next brief

Whatever the cohort shows, a brief goes only to a company that already passes SourceX's published screen: at least 50 full-time employees at its high point, contractors left out; years of records it created itself; the right to license them; and a CEO who will decide. 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. SourceX's partner process page says a partner need not export, upload or describe confidential records, and the terms say a link click alone does not decide credit; dated records do.

The causal half of the question has its own designs: the five-factor trust study, a test of how much context an introduction should carry, an experiment on long and short introductions and a comprehension test of referral material. More designs sit in the guides hub.

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 your next brief goes out, run it through the claim ledger, check the company against who qualifies, take any program fact from the current program terms and register as a partner so the introduction is on record.

Studying briefs is network research; whether to invest in SourceX is a separate decision, and investor materials for that are available on request from the SourceX investor overview.

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

Why score real briefs instead of writing test versions?

Because only real briefs have real outcomes. A test message can show what readers prefer in the abstract; a masked brief that was actually sent lets the study ask whether what executives call credible lines up with what the CEO who received it did. The cost is that real briefs differ in many ways at once, so every result is an association, not a cause.

Does the CEO who received a brief have a say?

Yes. The brief describes that CEO's company, so the masked version is used only if the CEO agrees. A CEO who declines removes the brief from scoring; the study still counts it, by number only, so the share of briefs that could not be used is visible.

Could a partner improve their own briefs from this before results exist?

Partly. The claim ledger works as a self-check today: count how many sentences in your last brief a CEO could verify, whether you named anything you did not know, and whether any sentence implied a result SourceX has not decided. It cannot tell you which of those matters most; only the study can.

Sources

  1. refer.sourcex.si/terms
  2. refer.sourcex.si/who-qualifies
  3. credibility.stanford.edu/guidelines/index.html
  4. refer.sourcex.si/how-it-works
  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