The design is for SourceX together with partners who opt in. It needs SourceX's first-call notes and the partners' briefs, and it changes nothing about how any company is reviewed.
More detail can save a call and still cost an introduction
The case for a fuller brief is simple: if the first conversation already knows the company's size, history and systems, it can start further along. The case against is quieter. Every line a partner writes about someone else's company is a line that can be wrong, and every extra field is room for things a partner should never pass on: a file, a figure learned in confidence, a guess about fit written as if it were settled. SourceX's screening guidance says fit questions are answered from a conversation with the relationship holder, "never from files, samples or logins," and its process page says partners do not need to export, upload or describe confidential records.
So "better" cannot mean "more useful" alone. The evaluation treats an introduction as better only when it saves effort without adding corrections or oversharing, and it measures all three together. That is what separates it from the long-versus-short email experiment, which compares two notes to a CEO and counts applications.
Four levels of brief detail
Each level adds lines to the one below it. Every line at levels 2 and 3 carries a tag saying the CEO stated it, and the CEO sees and approves the brief before it reaches SourceX.
| Level | What the brief contains | Who supplies each line |
|---|---|---|
| 0. Link only | Nothing beyond what the company enters when it applies through the partner's link | The company |
| 1. The route | Company name, the CEO's name, how the partner knows the CEO, the date of the CEO's yes, and the partner's stake disclosed | The partner |
| 2. Fit facts | Level 1, plus peak full-time headcount with contractors left out, years in operation and location, as the CEO stated them | The CEO, quoted by the partner |
| 3. Record outline | Level 2, plus the record types, the systems holding them, the years each spans and who runs each system, as the CEO described them | The CEO, quoted by the partner |
Out of bounds at every level: records, samples, screenshots or exports; anything the partner learned through a board or advisory seat, which may prompt a question to that company's own CEO but never travels further; customer names; the partner's view of what the records are worth; and any statement that the company qualifies. How to write a volume line honestly at level 3 is covered in when record volume is unknown, and labelling estimates in the estimate field guide.
Assignment by partner, not by introduction
The unit of assignment is the partner. Partners who opt in are grouped by type (investor, adviser, services firm) and by how many introductions they made in the previous two quarters, then assigned to a level by a draw recorded before the entry window opens. Each partner uses the brief template for their level for the whole window.
| Element | Specification |
|---|---|
| Unit of assignment | One registered partner who has signed the Channel Partner Agreement |
| Unit of analysis | One introduction submitted from that partner's dashboard |
| Strata | Partner type crossed with prior volume (low, high) |
| Entry window | Two quarters, fixed in advance |
| Follow-up | Each introduction followed for 120 days from submission: the 90-day attribution window plus time for review |
| Analysis | Every introduction analysed at its partner's assigned level, whether or not the partner followed the template |
| Adherence | A reviewer records, for each brief, whether it matched the assigned level |
Because outcomes are grouped by partner, the analysis has to account for that clustering, and the report would follow the CONSORT extension for cluster randomised trials published in the BMJ in 2012, which the CONSORT group cites in its reference list. In practice the number of partners per level matters more than the number of introductions: the plan is at least ten partners per level, and results would be reported as intervals, never as single figures.
Three scores for every introduction
| Score | How it is counted | Who counts it |
|---|---|---|
| Repeat questions (primary) | Questions SourceX's first call had to ask whose answers the assigned level was meant to carry, from a fixed list per level | SourceX's intake reviewer, using the list |
| CEO corrections (guard) | Brief lines the CEO changed, at approval or on the first call, divided by lines written | The partner logs approval edits; SourceX logs first-call corrections |
| Oversharing (guard) | Any out-of-bounds item in the brief or attached to it | A reviewer who does not see the partner's name |
| Verified application (secondary) | The company applied and SourceX verified the application within the follow-up period | SourceX's dated records |
| Partner minutes (secondary) | Time the partner spent writing the brief, self-reported | The partner |
Level 0 carries no brief, so its repeat-question list is empty by definition; it serves as the floor against which the fuller levels are compared on corrections, oversharing and applications.
Decision rule, written before the window opens. A level is preferred over the level below it only if it cuts repeat questions by a margin set in advance, while corrections per line and oversharing do not rise. Any oversharing in a level disqualifies it until its template is rewritten, whatever its other scores show.
Floor, read date and stop rules
- Floor: ten partners per level. Below that, the report shows raw counts per level and no comparison.
- Read date: 120 days after the last introduction in the window is submitted. Nothing is compared before it.
- A file, sample or export arrives at any level: that level pauses, the material is not kept and the template is rewritten before the level resumes.
- A brief repeats something learned in a board or advisory seat: the brief is withdrawn and the partner's level pauses.
- A brief says the company qualifies, or that SourceX sent the partner: the template for that level is rewritten.
- A new terms version or a revised screening page: the window closes, templates are re-dated and a new window opens.
Every introduction stays in the count
Introductions the CEO never approved a brief for, introductions that went unanswered, and companies that declined all stay at their partner's assigned level. A company SourceX already knew is a duplicate under section 3 of the terms and credited to nobody, but its brief is still scored, because brief quality is the subject. Nothing leaves the count after assignment.
| Count on the read date | Level 0 | Level 1 | Level 2 | Level 3 |
|---|---|---|---|---|
| Partners assigned | Not yet counted | Not yet counted | Not yet counted | Not yet counted |
| Introductions submitted | Not yet counted | Not yet counted | Not yet counted | Not yet counted |
| Briefs the CEO approved | Not applicable | Not yet counted | Not yet counted | Not yet counted |
| Repeat questions per introduction | Not applicable | Not yet counted | Not yet counted | Not yet counted |
| CEO corrections per line written | Not applicable | Not yet counted | Not yet counted | Not yet counted |
| Introductions with any oversharing | Not yet counted | Not yet counted | Not yet counted | Not yet counted |
| Verified applications | Not yet counted | Not yet counted | Not yet counted | Not yet counted |
| Duplicates | Not yet counted | Not yet counted | Not yet counted | Not yet counted |
| Still open on the read date | Not yet counted | Not yet counted | Not yet counted | Not yet counted |
Because levels are assigned by draw, a difference between them on these scores can be read as an effect of the level for the partners in the study. Differences inside one partner type, one industry or one company size are descriptive only: the draw did not balance those subgroups, and they are too small to carry a conclusion.
Released with the results: the four templates, the per-level question lists, the draw record, adherence rates and the full table above, zeros included.
What the evaluation cannot tell you
- Skill differs by partner. With ten partners per level, one unusually careful writer can tilt a level. Stratifying the draw helps; it does not remove the risk.
- The intake reviewer knows the level. A brief's length gives it away. A fixed question list per level limits how much judgment enters the primary score.
- Being watched changes writing. Partners in a study may write more carefully than they normally would, at every level.
- The approval step helps everyone. Having the CEO approve each brief may cut errors at all levels and shrink the differences between them.
- One program, one process. Results would describe SourceX's intake as it ran in the window, not introductions in general.
A brief that holds up until the evidence exists
Until results are in, the defensible default is the shortest brief that names the route and quotes the CEO for anything else, with every out-of-bounds item left out. The facts, questions and permission entry keeps each line attributed to who said it, and the language-evidence brief shows the same discipline applied to record language.
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 writing any brief, register and sign the Channel Partner Agreement, confirm the company against who qualifies with its CEO's own answers, and keep the program terms version beside the brief; other study designs sit in the guides hub.