Why one concentration number hides where a network depends on a few people
A team can spread its asks across a dozen people and still find that almost every CEO who said yes came through two of them. Counting partners or companies at one point in time cannot show that. The report treats concentration as a profile across five stages, so a reader sees where a network narrows and who it would lose with one departure.
| Statement | Standing | Basis |
|---|---|---|
| Partners sign the Channel Partner Agreement before submitting introductions, and one partner at most is credited per company | Program fact | Terms, sections 1 and 2 |
| Credit is decided from dated records; a link click alone never settles it | Program fact | Terms, section 2 |
| Whether any counted company is accepted, licenses records or produces a reward | Conditional | SourceX review and the company's own signed agreement |
| Concentration rises between the asking stage and the agreeing stage | Hypothesis | Unmeasured; this design tests it |
Two live pages answer neighboring questions: the six-shock dependency test groups partners by a shared source inside one network, and the one-large-introduction dashboard handles a single company swamping a results view. This report measures people, at every stage, and is built to be repeated and pooled.
Definitions fixed before any record is read
| Term | Definition used in the report |
|---|---|
| Relationship owner | The person who asks, or would ask, a company's CEO for the conversation; recorded by role and a random code, never by name |
| Partner of record | The person or entity that signed the Channel Partner Agreement and submits the introduction; the sample agreement leaves room for an individual or an entity |
| Network | One fund platform, advisory firm or association, with all its owners |
| Window | Two calendar quarters, fixed before the frame is built |
| S1 Named | A company an owner proposed for an ask during the window |
| S2 Asked | The owner put the question to the CEO, with a dated record |
| S3 Agreed | The CEO said yes to a conversation, in writing |
| S4 Applied | The company's application shows as verified in the partner's own portal status |
| S5 Credited | The portal status shows the company credited to this network's partner of record |
| Failed ask | Any S2 company that never reached S3: declined, unanswered by the cutoff, or withdrawn |
Each company is counted once per network, grouped by parent, and assigned to the owner who held the relationship on the day it was named. If two owners claim it, the report records both and assigns it by a tie-break written in advance (the earlier dated ask); credit itself is never inferred from that tie-break, because only SourceX decides overlapping cases.
The concentration and dependency analysis
Four measures run at every stage, and two run once per window.
| Measure | How it is computed | What it answers |
|---|---|---|
| Top-1 share | The largest owner's count divided by the stage total | Could one person's absence stall the stage? |
| Top-3 share | The three largest owners' counts divided by the stage total | How narrow is the core? |
| Effective number of owners | 1 divided by the sum of every owner's squared share | How many equal-sized owners would produce the same concentration |
| Concentration gradient | Top-3 share at S3 minus top-3 share at S2, then S5 minus S2 | Does the network narrow as companies move on? |
| Remove-one loss | For each owner, the S3 and S5 counts with that owner's companies taken out; report the largest loss | What one departure would cost, measured on outcomes, not asks |
| Single-path share | Of S3 companies, the share with only one owner able to reach the CEO, counted when the company was named | How many yeses had no second route |
The top-share and effective-owner measures borrow their arithmetic from official statistics. For its 2017 Economic Census, the Census Bureau gave firm concentration ratios for the top 4, 8, 20 and 50 firms and based its Herfindahl-Hirschman Index on the top 50 (paraphrase). The report borrows the arithmetic and no cut-offs: Census applies these measures to companies in an industry, a network of fifteen people is a different population, and a report should never call itself "highly concentrated" by importing a threshold.
Arithmetic demonstration (Illustrative, not data). Six owners asked 60 CEOs: A asked 12, B, C and D asked 10 each, E and F asked 9 each. Twenty CEOs agreed: 8 through A, 6 through B, 2 each through C and D, 1 each through E and F.
| Stage (Illustrative) | Total | Top-3 share | Sum of squared shares | Effective owners |
|---|---|---|---|---|
| S2 Asked | 60 | 32 of 60, about 53% | about 0.168 | about 5.9 |
| S3 Agreed | 20 | 16 of 20, 80% | 0.275 | about 3.6 |
The asks look spread across six people; the yeses behave as if fewer than four people held the network. Removing owner A would take S3 from 20 to 12. None of these counts describes any real network; they show how the columns are filled.
Sampling frame, consent and the records each network supplies
Frame. Networks with at least one signed partner of record and at least 30 dated asks in the window, invited in writing. Each network that agrees names who compiles its log; networks that decline are counted by type only.
What each network supplies. One row per S1 company: network code, owner code, partner of record, the date of each stage, the number of owners able to reach the CEO on the day the company was named, a sector band, a peak full-time headcount band, and the status on the cutoff date. The rows follow the record the sample agreement already asks for. Section 3.3 of the sample Channel Partner Agreement says: "Each introduction record must identify the company, domain, Partner, opportunity, introduction method and timestamps." Company names and domains stay inside the network; the researcher receives codes.
Where S4 and S5 come from. Only from the partner's own view of its referrals. Section 4.6 of the sample agreement gives the partner information reasonably needed to track its own referral and reward, through generic stages or reference codes (paraphrase). No network sees another's statuses, and SourceX is asked for nothing beyond what each partner already sees.
Consent. Each owner agrees before their rows are compiled and may withdraw them until the data freeze. No company or CEO is contacted for the study, and nobody handles company records for it; the fit fields hold only what the owner already knows from talking with the CEO.
What a counted company has to show. The companies in the log are those an owner believed could pass SourceX's screen: 50 or more full-time employees at peak, contractors left out; records the company created itself over several years; the rights to license them; and a CEO who authorizes the conversation. On location, SourceX's sentence reads: 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. Relevance to a current program is assessed only in SourceX's review, and a company that misses one program may fit another later.
Every failed ask stays in the denominator
A concentration share computed only on successes flatters whoever was lucky. Each S2 company carries one final code on the cutoff date, and every code stays in the stage it reached.
| Code at cutoff | Counted in | Why it stays |
|---|---|---|
| Owner declined to ask | S1 only | Shows whose relationships are guarded, not whose asks fail |
| CEO declined | S2 | A failed ask; the most informative row for each owner |
| No answer by the cutoff | S2, flagged | Read as a range: once as declined, once excluded |
| Applied, then not a fit after review | S4 | A company the owner's screen missed |
| Duplicate ruling | S4, flagged | SourceX already knew the company; credited to no later partner under terms section 3 |
| Still in review | Its last stage, censored | Never counted as a success or a failure |
Per-owner success (S3 divided by S2) is printed only with its interval and only for owners with ten or more asks; smaller owners are pooled as "others". The report is descriptive: a high top-three share at S3 says where yeses came from, not that those owners ask better.
How results will be reported, and why the table is empty
Results: not published. No network has been measured, and no number on this page is a finding. Each report will print, per network and then pooled: the frame and the networks that declined, the window and cutoff, this table filled in with intervals, the remove-one loss for every owner above the pooling floor, the single-path share, the no-answer range and dated departures from this protocol.
| Stage | Total | Top-1 share | Top-3 share | Effective owners | Largest remove-one loss |
|---|---|---|---|---|---|
| S1 Named | Not published | Not published | Not published | Not published | Not published |
| S2 Asked | Not published | Not published | Not published | Not published | Not published |
| S3 Agreed | Not published | Not published | Not published | Not published | Not published |
| S4 Applied | Not published | Not published | Not published | Not published | Not published |
| S5 Credited | Not published | Not published | Not published | Not published | Not published |
Any cell resting on fewer than five companies prints as a count, never a share. Networks are identified by type, never by name, unless a network asks in writing to be named.
Counterexamples and weak points in a concentration report
Counterexamples the report must print. A network whose top owner holds most asks but few yeses; a network whose gradient falls, so it widens toward outcomes; owners whose S3 rows are mostly single-path even though their asks looked spread. If most networks show a flat or falling gradient, the hypothesis above is wrong and the stage profile adds little to a simple count.
- Small numbers. Thirty asks and eight yeses make every share jumpy. The report reads directions, not decimals, until two windows agree.
- Owner assignment. Introducer chains blur who owned a relationship. The tie-break keeps the report consistent but cannot make it correct; the sponsor-reach study codes the path itself.
- Consent bias. Networks willing to share logs may be the better-run ones, and owners who withdraw rows may be the ones with weak results.
- Portal lag. S4 and S5 depend on statuses that can change after the cutoff; a later restatement keeps the earlier version and its date.
- Frame bias. Companies nobody named never enter, so the report says nothing about the network the log cannot see.
Methods note. The report measures counts of companies per owner at five dated stages, from network logs and each partner's own portal statuses, over one fixed window. It does not measure demand for any company's records, record quality, payouts or rewards, none of which a network can observe before SourceX's review, and it uses no reward figures. Formulas: shares on stage totals; effective owners as the inverse of the summed squared shares.
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 stage table shows your network leaning on one or two people, widen the set of owners who can ask before you widen the set of companies: register whoever will sign as partner of record, check each company against who qualifies, read the program terms, and find related study designs in the guides library.
Building a concentration report is a channel decision; deciding whether SourceX itself merits investment is separate work, and readers doing it can request investor materials 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.