A network can list 38 partners and still depend on one firm's client list. The name count measures recruiting; the test below measures what one bad quarter could take away.
Count exposures by what partners share, not by who they are
Bank supervisors already think this way. The Comptroller's Handbook booklet on concentrations of credit groups loans into pools that share a characteristic, such as one borrower group, a common repayment source, an industry or a region, because a single development can make the whole pool behave the same way. The booklet also warns that the measured size of a concentration depends on how broadly you draw the pool, so a bigger number is not automatically a bigger threat.
A referral network has the same structure. Four common drivers decide whether partners rise and fall together:
- A shared relationship source. Partners at one firm, one fund or one association draw on overlapping clients. If the firm changes its policy on outside referrals, all of them stop at once.
- A shared sector. If most fit companies come from one industry, a change in which records a current program needs affects all of them together.
- A shared criteria version. SourceX applies the criteria of the specific program, and they differ by size, history, geography, language and data category. Companies screened against one program's criteria move together when that program changes.
- A shared queue. Introductions from every partner pass through the same people who check them before submission. One partner's surge delays everyone.
Draw the pools before measuring, and write down why each partner sits where they do. Too narrow a pool hides the risk; too wide a pool makes everything look concentrated.
What this test counts as an active partner and a fit company
Five terms carry the whole analysis. Fix them first, with the date, so a later rerun compares like with like.
| Term | Definition used in the test | Why it matters |
|---|---|---|
| Signed partner | Has signed the SourceX Channel Partner Agreement | The only partners who can submit introductions |
| Active partner | Proposed at least one company in the last two quarters | Separates recruiting from output |
| Fit company | A unique company that passes the published fit facts on metadata, counted once however many partners named it | The unit every shock is measured in |
| Credited company | Only the earliest valid introduction inside the 90-day attribution window counts, and only after the company itself submits a verified application | What the program actually recognizes |
| Pool | A group of partners sharing one of the four drivers above | The unit of exposure |
Every pool in the network screens against the same facts: at least 50 full-time employees at the company's peak, with contractors left out of the count; a documented operating history; the right to license those records; and an authorized contact, who in this program is the CEO. 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. Partners get there by asking which records exist, about how many, spanning how many years, held in which system and owned by whom. Nobody in the network handles files, samples or logins, and whether the records match a current program stays unknown until SourceX reviews them.
The partner dependency scenario analysis
Illustrative. A generic network with invented counts; no real partners, companies or outcomes.
Over two quarters, 38 signed partners produced 52 proposals naming 41 unique companies, of which 34 passed the fit facts on metadata. Seventeen partners were active. Grouped by shared source:
| Pool | Partners active | Unique fit companies | Shared driver |
|---|---|---|---|
| Pool 1: one regional accounting firm | 3 (A with 7, B with 5, C with 3) | 15 | One client base, mostly freight and warehousing |
| Pool 2: one private equity operating group | 2 | 7 | One fund's portfolio |
| Pool 3: MSPs from a spring association campaign | 4 | 5 | One association's members |
| Independent partners | 8 | 7 | None found |
By name, the largest partner holds 7 of 34 fit companies, about a fifth. By pool, one firm holds 15 of 34. Sixteen of the 34 are freight and warehousing companies, and most of those came through Pool 1.
Now apply six shocks, each measured as the fit companies the network would still have if nothing replaced the loss for two quarters.
| Shock | What it tests | Fit companies left (of 34) | Early signal in dated records | Agreed response |
|---|---|---|---|---|
| 1. Partner A goes quiet | Name concentration | 27 | No proposal from A in 45 days | Normal follow-up; no action on the network |
| 2. Pool 1's firm stops outside referrals | Shared source | 19 | All three Pool 1 partners quiet in the same month | Recruit where Pool 1's clients are not; never pressure the firm |
| 3. A program's criteria change for freight records, and 10 of the 16 freight companies stop matching | Shared sector and criteria version | 24 | Decline notes citing program fit cluster in one sector | Mark the ten "revisit", since a company that misses one program may fit another later |
| 4. Duplicate-risk case: Pool 1's partners name each other's clients | Overlap | 34, but Pool 1 shows 19 names for 15 companies | The same company proposed twice in one quarter | One shared pre-check list inside the firm |
| 5. Underperforming cohort: the spring campaign stays where it is | False breadth | 34, with 10 of 14 campaign partners never proposing | Signed count up 58%, fit companies from the cohort at 5 | Stop counting the cohort as diversification |
| 6. Capacity limit: Partner A brings 18 companies from one client seminar | Shared queue | 34, but other partners wait about two months | One partner holding more than half of a month's checks | Cap any one partner at 4 of the 10 monthly pre-check slots |
The capacity figures are your own: in this example the people screening introductions before submission can check 10 a month. SourceX's internal review capacity is not published, and none of these numbers describe it.
The working line. Call the network strategically weak when one shock would leave it with less than two-thirds of its fit companies (about 23 here), or when one partner could take more than half of a month's checking capacity. Only shock 2 crosses the first line, and shock 6 crosses the second. These thresholds are editorial working rules, not an industry standard; set your own, write them down and keep them fixed between runs.
Method note. The test uses dated proposal records, unique-company counts after duplicate removal, the criteria version each company was screened against and your own workload data. It does not measure demand for any company's records, record quality, payout or partner earnings, none of which a network can see before SourceX's review.
Where support hours, review slots and overlap pile up behind one pool
Concentration rarely shows in one metric. It shows in four places at once, all pointing at the same pool.
| Dimension | What to count | Reading in the example | Weak when |
|---|---|---|---|
| Output | Unique fit companies by pool | 15 of 34 from Pool 1 | One pool holds more than a third |
| Support | Hours spent answering partner questions, by pool | Pool 3 takes the most hours for 5 fit companies | Hours rise in a pool whose output does not |
| Queue | Share of next month's pre-check slots, by partner | Partner A would take 18 of the next 22 | One partner above half |
| Overlap | Proposals naming an already-proposed company | 4 inside Pool 1 | Overlap repeats in the same pool two quarters running |
Overlap has a second, quieter edge. Several of Pool 1's clients may belong to corporate groups. For the reward, a parent and its subsidiaries are one company, so the USD 50,000 floor applies once, to the group's first qualifying transaction, however many entities license; later qualifying transactions by sister entities first agreed within the 12-month Repeat Window pay 15% of their Actual Supplier Payout. The terms are silent on credit when the parent had earlier contact with SourceX, or when another partner already introduced a sister company. SourceX works that out in duplicate review from its dated records, so a network should flag group links in its own list and assume no answer. The subsidiaries attribution guide covers the records to keep.
Evidence gates before calling the network diversified
Treat "diversified" as a claim that needs evidence, and open each gate only on dated records.
- Pools defined and dated. Every active partner sits in exactly one pool, with the reason written down.
- Unique companies only. Duplicates are removed before any share is calculated. The terms credit one partner per company, and a company SourceX already knew through an application, lead, account or email is not credited to anyone new.
- Two consecutive quarters under the line. No pool above a third of fit companies, and no shock below two-thirds, twice in a row.
- Criteria version recorded. A change in program criteria resets the comparison; rerun the test on the new version.
Correction path. When duplicate review moves a company out of a pool's count, restate that pool's figures with the date and keep the earlier version. When a pool goes quiet, record the date and the cause if known, and do not delete its history; a quiet pool is information about your exposure. When a campaign cohort stays inactive past two quarters, report it as its own line rather than folding it into the total.
Where the scenario analysis misleads
- Small numbers. With 34 fit companies, one company moves a share by three points. Read shocks as directions, not measurements.
- Pool boundaries. Two partners at different firms who serve the same industry association may still overlap. The booklet's warning about pool definitions applies in both directions.
- Quiet is not lost. A partner who pauses for a quarter may return; the shock asks what happens if they do not.
- Context from software channels. In PartnerStack and Wynter's survey of 100 senior leaders at B2B SaaS companies with $50M+ in revenue, mid-market and enterprise respondents reported that about 35% of new pipeline was partner-influenced or partner-sourced. That is a software-sales figure, not a data-referral one, but it shows how much weight a channel can end up carrying before anyone stress-tests it.
- Company concentration is a different question. How much of one company's records come from a few customers is a fit and rights question covered in customer concentration in M&A. This page is about where your introductions come from.
Related: the partner's side in depending on one large referral, one company distorting a dashboard in one large introduction, unique-company counting in qualified introductions, not logo counts, queue rules in limited review capacity, firm pre-checks in preventing duplicate claims, and 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.
When your shock test shows a pool you cannot afford to lose, recruit partners outside it: point them to register for a partner link, have them screen each company against who qualifies, and make sure every one of them reads the program terms before introducing an authorized CEO who wants the conversation.