A field guide to failed data introductions: an identification key that names each failure by the dated evidence it left, sorts preventable failures from correct stops, and a case collection protocol that sets every failure beside an introduction that nearly failed

Short answer

A failed data introduction teaches something only when its mechanism can be shown from dated records and told apart from a correct stop, such as a company the published criteria ruled out. This field guide gives an identification key with eight types, the evidence each type leaves and the type it is most often mistaken for, plus a protocol for collecting consented, anonymized cases that sets every failure beside a comparable introduction that nearly failed but went ahead. No cases have been collected or published yet.

Visual briefFull-size diagram
A branching identification key. Starting from a failed introduction, yes-or-no questions lead to labeled endpoints: CEO skipped, prior contact, knowable misfit, information leak, CEO declined, stalled after yes and review outcome, with preventable endpoints outlined in orange, correct stops in green and complex outcomes in slate.
The key's questions and types are this page's design; no cases have been collected or classified yet.

Platform teams already have failed introductions in their notes. What they rarely have is a way to tell which failures were mistakes, which were the screening working, and which no one could have avoided. Without that sorting, a "lessons learned" deck mixes all three and draws the wrong lesson from most of them.

Failure, correct stop or still open

Three outcomes look alike in a CRM and mean different things.

OutcomeDefinition used on this pageExample
FailureThe introduction ended without the result the CEO wanted, for a reason someone on the introducing side could have acted onBoard-pack details reached SourceX before the CEO knew about the idea
Correct stopIt ended because a published criterion or an informed CEO decision said it shouldA 40-person firm, below the full-time headcount SourceX requires, was not introduced
Still openInside the 90-day window that section 2 of the terms allows for credit, or still in SourceX's reviewAn application submitted last month

Amy Edmondson's account of organizational failure gives a useful lens. Her Harvard Business Review article sorts failures into three categories: "preventable ones in predictable operations, which usually involve deviations from spec; unavoidable ones in complex systems, which may arise from unique combinations of needs, people, and problems; and intelligent ones at the frontier". The same summary says some failures are inevitable and some even good, and that people need to feel safe reporting them. Applied here, a knowable misfit is preventable, a rights problem nobody could see is complex, and a well-reasoned ask to an unusual company that SourceX declined can be intelligent.

The identification key

Work through the couplets in order and stop at the first endpoint that fits. Each question must be answered from a dated record, not memory; if no record answers it, the case goes to type H.

  1. Did SourceX hear about the company before its CEO knew about the idea? Yes: type A. No: go to 2.
  2. Had the company any earlier application, lead, account or email exchange with SourceX? Yes: type B. No: go to 3.
  3. Did a published fit fact fail that the introducer could have checked before asking? Yes: type C. No: go to 4.
  4. Did confidential material travel with the introduction? Yes: type D. No: go to 5.
  5. Did the CEO, fully informed, say no? Yes: type E. No: go to 6.
  6. Did the CEO say yes, yet no application followed inside the 90-day window? Yes: type F. No: go to 7.
  7. Did the company apply and SourceX find no fit with a current program, or rights it could not clear? Yes: type G. No: type H.
TypeField mark (the dated evidence that must exist)Commonly mistaken forLensWhat it can teach
A. CEO skippedAn introduction record dated before the CEO's first contactType E, when the CEO later says noPreventableAsk the CEO first; SourceX's screening page says an introduction the decision maker did not know about is usually not a fit
B. Prior contactSourceX's duplicate outcome, plus the date of the company's earlier contactType GCorrect stop for creditUsually nothing the partner could know; section 3 counts any earlier lead, account or email correspondence
C. Knowable misfitA public or CEO-given fact, dated before the ask, that fails a published criterionType GPreventableCheck headcount at peak, record ownership and location before asking
D. Information leakA message that carried board-pack, client or system detailsAny type it hides behindPreventable, and seriousIntroductions carry a name, a contact and a reason; the process page says no confidential records need to travel
E. CEO declinedThe CEO's dated no, after a clear explanationType ACorrect stopNothing to fix unless the ask itself was pressured
F. Stalled after yesA dated yes and no application within 90 daysType E, when the CEO quietly changed courseComplexRead with the stall study's stage clocks
G. Review outcomeAn application date and SourceX's outcomeType CComplex, sometimes intelligentProgram criteria and rights are settled in review, not by the partner
H. UnclassifiedToo few records to chooseNoneUnknownBetter logging, not a lesson

Types A, C and D are the casebook's real subject: failures with a preventable mechanism. Types B and E are kept to stop them being mislabeled as mistakes. Because the key stops at the first match, a leak could hide behind an earlier type, so question 4 is asked of every case and a leak is recorded as a second tag wherever it occurs.

The case collection protocol

Frame. Participating networks contribute every introduction or ask they logged from the protocol's start date that ended as types A to H, plus near misses: introductions that reached an application although a field mark of A, C, D or F was present (for example, a CEO brought in late who still agreed). Networks do not choose which cases to send; they send all of them.

Consent. A case narrative needs the introducer's consent and, where the company is identifiable from any detail, the CEO's. A company can withdraw its case at any time before publication. Counts by type need no narrative and no company consent, because they carry no identifying detail.

Evidence standard. A type is assigned only from records: dated messages, the introduction record, the application date and SourceX's published outcome stage. Two people classify each case independently with the key; disagreements are discussed and the resolution recorded, and a case still in dispute becomes type H.

Anonymization review. Before anything is shared beyond the collecting team: company and people names removed, industry widened to a broad sector, headcount given as a band, years of records as a range, and dates shifted to month and year. A third person reads each narrative asking one question: could someone in this network recognize the company? If yes, it is rewritten or dropped.

What the case file never holds. No company records, samples, screenshots, contracts or reasons a CEO shared in confidence. The casebook studies how introductions were made, never what a company owns.

Failures beside near misses: what the comparison can say

A field mark becomes a lesson only if it appears more often among failures than among near misses drawn from the same kind of network, company size band and relationship route. If late CEO involvement is as common in near misses as in failures, it does not explain the failures, however vivid the failure stories are.

Even then the comparison is descriptive. Networks that volunteer cases are unusually careful, cases with consent differ from cases without it, and small counts move a great deal with one case. The report will say "appears more often among failures in these cases" and never "causes". Where a mechanism appears in only one network, it is reported as that network's pattern.

This casebook is neighbor to four other studies, each answering a different question: why introductions stall measures time in stage, why founders decline asks the founders, ownership blockers counts one family of causes, and the declined-referral protocol builds lessons from the introducer's own steps. One partner can map a single year stage by stage with the funnel map of why good introductions fail, and a platform whose whole pilot ended at zero can start with what an unsuccessful pilot can teach; the casebook pools cases across networks instead.

One case written to the template (Illustrative)

Illustrative. An invented case; no real network, company or outcome.

FieldEntry
Case codeNetwork 2, case 14
Company bandRegional logistics, 200 to 400 full-time employees at peak, records spanning 10 to 15 years
RouteA portfolio founder supplies the company; the platform lead asked through the founder
Dated sequenceMonth 1: platform lead submits the introduction. Month 1, nine days later: the founder first mentions it to the CEO. Month 2: CEO declines and asks not to be contacted again
Key pathQuestion 1: yes. Type A, with a second tag D because the submission described the company's dispatch system, learned from the founder
ComparisonNetwork 2, case 9, a near miss: CEO also told after submission, agreed anyway and applied
Lesson tagPreventable: the CEO hears first, and the introduction carries only a name, contact and reason
What stays unknownWhether the CEO would have agreed if asked first

The comparison row matters most. Case 9 shows that a late CEO conversation does not always fail, so the casebook cannot say lateness alone ended case 14; the leaked system detail is the stronger candidate, and the report would say so with that caveat.

How the casebook will be reported

The report gives counts by type and lens for failures, near misses and correct stops, then a narrative for each type that reaches five consented cases, with its comparison cases beside it. Types with fewer than five cases are reported as counts only.

TypeFailuresNear missesCorrect stopsNarratives published
A to HNot yet collectedNot yet collectedNot yet collectedNone

No cases have been collected, classified or published. Any statement about SourceX's own results comes only from its verified records, as partner results research explains.

Where a field guide misleads

  • Volunteer networks. The careful networks that join will show fewer preventable failures than the field has.
  • Outcomes SourceX does not explain. Type G rests on the outcome stage a partner can see; the reasons behind a review decision may never reach the casebook, so G stays broad.
  • Hindsight. A fit fact can look "knowable" after the fact. Type C requires a dated record from before the ask, not a fact discovered later.
  • Rules that change. SourceX applies each program's own criteria, and those move; a type C case is judged against the published facts on the date of the ask.
  • Rare events. Leaks are, one hopes, uncommon; one case can dominate a small casebook, and the report will flag single-case types.

Methods note. The key's endpoints follow SourceX's terms (version 2026-10-22.2, since replaced by version 2026-11-04), its screening page and its process page, read on 2026-10-28, with Edmondson's three categories as the lens. The couplets, field marks, near-miss comparison and five-case threshold are this page's design and have not been tested on real cases.

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.

The key works best before the ask: if a company would land in type A, C or D today, fix that first, then register as a partner, confirm the company against who qualifies and read the program terms so the dated record you leave is the one the casebook would want. Other research designs are in the guides library.

Studying how introductions fail is referral work; evaluating SourceX as a company to invest in is a separate decision, and materials for that are available on request through 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

Is a company that SourceX declines a failed introduction?

Not always. If the published criteria ruled it out and the partner could not have known, the introduction was a correct stop or a complex outcome, not a mistake. The key separates those from failures someone could have prevented, because only the second kind teaches a partner to act differently.

Why collect introductions that worked?

Because a warning sign is only a lesson if it is rarer where things went well. If late CEO involvement shows up as often in near misses as in failures, it does not explain the failures. Without the comparison, a casebook collects coincidences.

Can the casebook say how often introductions fail?

No. It collects cases to understand mechanisms, not to estimate rates, and participating networks are not a random sample. Rate questions belong to stage-by-stage studies with a full denominator.

Sources

  1. hbr.org/2011/04/strategies-for-learning-from-failure
  2. refer.sourcex.si/terms
  3. refer.sourcex.si/who-qualifies
  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