Budgeting a pilot with income locked at zero is covered in the zero-revenue pilot budget, and naming why a finished pilot fell short in the failure-type review. This page covers a pilot that was simply read before its outcomes could exist.
Why the completion column is empty on review day
Everything a platform does sits early, from screening to the introduction. A completed transaction sits at the far end, after the company signs with SourceX and delivers its dataset, the Acquisition Partner gives its final written acceptance and SourceX pays the company. The terms (version 2026-10-22.2, since replaced by version 2026-11-04, section 5) set a clock only at that end, a payment deadline that runs 60 calendar days from whichever comes later, completed handoff or that acceptance, and they give no duration for the stages before it. A pilot planned for one quarter can therefore close with every introduced company somewhere between application and handoff.
Statisticians call this censoring. The NIST/SEMATECH handbook describes a test run to a fixed time in which the number of units reaching the event is random, possibly zero, and units still running at the cutoff are censored because their event times are missing (our paraphrase). A company in review on your review date is such a unit; calling it a failure or a near-win invents an outcome.
Here a completed transaction means the dataset reached Closing and completed handoff and SourceX paid the company. A verified application, a negotiation or a signature is not a partial completion; section 5 of the terms is blunt: "A lead, meeting or signed agreement alone does not earn a payable reward."
The plan a learning review is graded against
A learning review is only as honest as the plan written before the first ask. Without one, call the review an observational account in its first line and drop words such as "confirmed". With one, the review grades the pilot against seven fields fixed in advance.
| Field | What the plan states | Why the review needs it |
|---|---|---|
| Question | Does the network yield CEOs who want the conversation, and verified applications, at an acceptable effort? | A money question cannot be answered in one quarter |
| Hypothesis | A number written in advance, such as at least three CEOs agreeing out of the first eight asked | Without it, any count can be called encouraging |
| Unit | One company, named before outreach, with its relationship owner and the date it was screened | Companies added after early wins inflate the record |
| Cohort rule | Companies that pass the published fit facts on metadata alone | Keeps companies that could never fit out of the denominator |
| Comparison | The plan's own thresholds, or the platform's previous quarter | A baseline borrowed from another fund cannot be checked |
| Truthful variants | Any compared approaches say the same true things: the CEO decides, SourceX decides fit, nothing is promised, the payout does not depend on the partner | A variant that wins by overstating teaches nothing |
| Stop and decision rules | Triggers that pause asking, and what each possible result leads to | Stops the review from writing a rule after seeing the numbers |
The learning review: observed, open, not observable
Every sentence gets a tag before it is written. Observed: a dated record supports it now. Open: the company is still at a stage, so the sentence says only where and since when. Not observable: nothing the platform can see would support it, so it is cut.
Method note. The tags rest on SourceX's terms, for the stages and the payment clock, and the NIST handbook's account of fixed-time observation, both checked on 2026-10-28. The review never estimates the chance that an open company completes: SourceX publishes no conversion rates or timing averages, and the platform cannot see inside its review.
| Sentence the review wants | Record that would support it | Tag on review day | How it is written |
|---|---|---|---|
| We asked N CEOs | Dated ask log, one row per company | Observed | A count and the date range |
| X CEOs wanted the conversation | Each CEO's written yes | Observed | X of N asked |
| Y companies reached a verified application | The stage shown in the partner dashboard | Observed | Y of the companies introduced |
| Some companies were already known to SourceX | The duplicate outcome SourceX communicated | Observed | A count, logged as a cost of how the cohort was found |
| SourceX declined a company | The outcome as SourceX communicated it | Observed | A count, with no guessed reason |
| Founders found the conversation useful | Two-line notes in the founder's words, coded by a second person | Observed once coded | Counts per code, negative ones included |
| Several companies are progressing | The last dated stage change for each | Open | Each company with its stage and the date it got there |
| The pilot will produce rewards, revenue or a value per company | Nothing the platform can see; no conversion rates or typical payouts are published | Not observable | Not written |
| A transaction completed | Completed handoff and the company paid | Not yet observed | None by the review date, beside the open count |
What the platform can measure before anything completes
Two families of measure need no completed transaction. Qualified next steps are things the CEO did or asked for, each with a date: a written yes to a conversation, a question the CEO sent SourceX, a verified application SourceX did not return as a duplicate. A meeting the platform booked with itself does not count. Founder trust is logged after every conversation, whatever its result, and effort (platform hours, relationship-owner asks, founder time) sits beside both.
Every measure carries its denominator (of the CEOs asked, of those who agreed, of the companies introduced), and each open company carries its stage age, which describes and predicts nothing. The three trust items are these:
- Did the CEO describe the offer back correctly? Correct means the CEO decides, SourceX decides fit, nothing is promised, and the company is paid the same with or without a partner. On that last point section 4 of the terms says "The referral reward is paid separately by SourceX and is never deducted from the supplier's agreed compensation."
- Would the CEO hear another introduction from the same relationship owner? Yes, no or unsure, in the CEO's words.
- Did anything go wrong? A complaint, a request not to be contacted again, or any sign the CEO felt pressed.
Stop rules while the pilot is still asking
Written into the plan, these can all fire inside a quarter because none waits for a completion. Each pauses or ends asking; none changes how open companies are read.
- A founder says the offer was misdescribed. Pause every ask until the wording is fixed and that founder has had a reply.
- Any record, file, sample or login reaches the platform. Stop, delete it, tell the sender, and log a protocol failure.
- Relationship owners decline most of the first asks. Set the number in advance; when it is reached, stop asking and revisit the cohort rule.
- SourceX asks for a change in how the program is described. Pause until the change is made everywhere.
What the observed rows can decide
The decision uses observed rows only. Open companies never vote: one still in review neither rescues a weak pilot nor sinks a strong one, and the review says so, so nobody reads the open count as a promise.
| What the observed rows show | Decision for the next cohort | What happens to open companies |
|---|---|---|
| Qualified next steps reach the planned threshold and no trust item tripped | Ask the next cohort under the same plan | Read on their fixed dates |
| Next steps fall short, but CEOs described the offer correctly and nobody complained | Hold new asks; fix the cohort rule or the relationship-owner step first | Read on their fixed dates |
| A trust item tripped: a misdescription, pressure or records sent | End asking until the cause is fixed and the plan rewritten | Read on their fixed dates, unchanged |
| Too few companies to read against the threshold | Write that no answer exists yet about the network | Read on their fixed dates |
Read the open rows again on dates set at review
The review date ends the asking, not the observing. Each open company gets read dates and a horizon, after which it is reported as open at horizon rather than turned into a result; a read every 60 days is one workable cadence (Illustrative). Nobody edits the original review: each read appends a dated line on which companies moved and which did not, recording only what SourceX communicated, because credit is decided from dated records (terms, section 2) and silence is not an outcome.
A review written on day 120 (Illustrative)
Illustrative. A generic cohort; the companies are letters, and every count and day is invented to show the format, not a result.
Twelve companies were screened; three failed a published fit fact and one relationship owner declined to ask, leaving eight CEOs asked against a planned threshold of three agreeing. Tagged, the review reads: five of eight wanted the conversation (observed); three companies are open in review and none has completed (open); one CEO heard the offer as selling customer data, so the opening line was rewritten (observed). No sentence mentions rewards. Decision: ask the next cohort under the same plan, and read A, B and C again on days 180 and 240.
| Company | Furthest stage, with day | Tag | Review wording |
|---|---|---|---|
| A | Verified application, in SourceX review since day 41 | Open | In review, 79 days at review |
| B | Verified application, in review since day 66 | Open | In review, 54 days |
| C | Verified application, in review since day 95 | Open | In review, 25 days |
| D | Verified application, declined by SourceX on day 88 | Observed | Declined in review, reason as SourceX gave it |
| E | Introduced, returned as already known to SourceX | Observed | Duplicate, not credited |
| F | CEO declined after describing the offer correctly | Observed | Informed no |
| G | CEO declined, saying it sounded like selling customer data | Observed | Misreading, logged as a wording fault |
| H | No reply after the single follow-up the plan allowed | Observed | No reply |
Which companies a pilot cohort may include
A company enters the cohort only on facts the relationship owner already knows. Its full-time headcount must have reached 50 or more at its peak, with contractors left out. It needs years of operating records of its own making and the right to license them, and its CEO must know about the idea and want to talk. 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.
Screening stays at metadata, never files, samples or logins. Client-owned material is usually not a fit, each SourceX program applies its own criteria, and a failed screen carries a date, not a verdict.
Small samples, negative results and where the review misleads
With eight companies, one company moves a percentage by more than ten points, so the review reports counts ("five of eight"), never rates; how to avoid declaring a winner from five replies works through the arithmetic, and a cohort built from ten hand-picked introductions reads the same way. What did not work goes in at full size.
The review can still mislead: open companies may never complete, first cohorts draw on the warmest relationships, trust items are self-report (hence the second coder), and a decline in SourceX's review describes a program today, not a sector. When the review is ready to leave the team, publishing a retrospective investors can trust sets out the format.
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.
If the review supports a next cohort, register as a partner before its first introduction, check each company against who qualifies and keep the program terms beside you; other designs sit in the partner guides.
Reviewing a referral pilot is a different exercise from judging SourceX as a business. Anyone doing the second can ask for investor materials on SourceX's investor page, which notes that company and commercial metrics go 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.