So set the limit on the front stages, not on the length of the list. The calculator below has no revenue line and ends with a team whose own log tells it to stop.
What is scarce: operator hours, not company names
The founder benefit is narrow. A CEO learns whether the company's own operating records could be licensed for AI training and evaluation and, if the CEO wants it, gets an introduction to SourceX. From there SourceX reviews the company and, if a deal follows, manages the agreement, rights review, data preparation, delivery and payout, as what SourceX manages after an introduction sets out. The operator never touches records.
The constraint is attention, and most of it is already committed. In a 2020 NBER working paper, Gompers, Gornall, Kaplan and Strebulaev reported that venture capitalists worked about 58 hours a week and spent more of it on portfolio companies, about 19 hours, than on any other activity. A referral track draws on those same hours and on two ledgers the operator does not control.
| Ledger | Whose time | In the calculator? |
|---|---|---|
| Operator hours | The platform lead or analyst running the track | Yes: this is the capacity limit |
| Relationship-owner hours | The partner, LP or founder who asks the CEO | Logged beside it; about 0.5 hours per ask in this Illustrative plan |
| Founder hours | The CEO who takes the conversation | Logged beside it; about 1.5 hours per conversation in this Illustrative plan |
| Referral income | Nobody's time | Left out; the base case is $0 |
The stage-weighted workload calculator
Each stage needs three inputs:
- T, the weeks a referral typically stays in the stage;
- w, the operator hours per week one referral needs while it sits there;
- λ (lambda), the new referrals entering the first stage each week.
Little's law does the rest. MIT's John D.C. Little proved in 1961 that the number of items in a system equals their arrival rate multiplied by their average time in it, L = λW. Applied stage by stage, the referrals sitting in a stage equal λ × T, and that stage's weekly load equals λ × T × w. Add the stages and the weekly load is λ times the hours one referral consumes over its life. Turned around, that is capacity: the new referrals per week you can sustain equal your weekly hours divided by hours per referral.
Illustrative, at one new referral a week:
| Stage | Who carries it | Weeks in stage (T) | Operator hours per week per referral (w) | Hours per referral (T × w) | Active referrals (λ × T) | Weekly operator hours |
|---|---|---|---|---|---|---|
| 1. Prepare: screen from data the fund holds, brief the relationship owner | Operator | 1 | 1.5 | 1.5 | 1 | 1.5 |
| 2. Ask: the owner asks the CEO, then a metadata-only conversation | Operator and relationship owner | 2 | 0.75 | 1.5 | 2 | 1.5 |
| 3. Introduce: written introduction and portal submission, until the company applies | Operator | 2 | 0.2 | 0.4 | 2 | 0.4 |
| 4. SourceX review: founder questions and status checks | SourceX, operator on call | 8 | 0.1 | 0.8 | 8 | 0.8 |
| 5. The company's deal, delivery and payout | SourceX and the company | 16 | 0.05 | 0.8 | 16 | 0.8 |
| Total | All stages | 29 | Varies by stage | 5.0 | 29 | 5.0 |
Twenty-nine referrals count as active, but stages 1 to 3 hold five of them and take 3.4 of the 5.0 weekly hours, or 68%. The other 24 sit in stages that SourceX and the company run. A cap of ten on the total count would halt this track in its tenth week, although its front-stage load has not changed since week five.
| Operator hours per week | New referrals per week sustained | Per month | Active at steady state | In stages 1 to 3 |
|---|---|---|---|---|
| 2 | 0.4 | 1.7 | About 12 | 2 |
| 5 | 1.0 | 4.3 | 29 | 5 |
| 8 | 1.6 | 6.9 | About 46 | 8 |
The T values for stages 4 and 5 are placeholders; they change the size of the active list but barely the hours. Referrals leave the count only when they end: SourceX declines the company, the CEO decides not to go ahead, the deal ends without Closing, or the company is paid. Record those exits with dates, or the list grows for ever.
What each stage costs when no revenue is assumed
Prepare, 1.5 hours. Read what the fund already holds, such as headcount history, years in operation, main systems and who runs them, against the published fit tests (a peak full-time headcount of 50+ excluding contractors, documented operating history, licensing rights and an authorized contact), plus the location rule on who qualifies. Then brief the relationship owner on one page.
Ask, 1.5 hours. The owner asks the CEO, who is the person who authorizes a licensing discussion. If the CEO wants to talk, the conversation stays at metadata level: which records exist, roughly how many, over how many years, in which system and who owns it. No files, samples or logins.
Introduce, 0.4 hours. One written introduction to the CEO and a portal submission the same day. Credit goes to the first valid introduction of a new company once it submits a verified application within the 90-day attribution window, so this stage should not linger.
Review, deal and payout, 1.6 hours together. Founder questions and status checks. SourceX does the work; the partner involvement comparison shows how little falls to the partner.
The calculator has no income row, on purpose. A pilot plans a quarter at a time, and a reward cannot buy hours back inside that quarter. It becomes payable only after Acquisition Partner acceptance, Acquisition Partner payment received by SourceX in cleared funds, payment of the supplier compensation, SourceX finance approval and the partner's payment and tax details, and even then the money can take until 60 calendar days after the last of the two closing steps, completed handoff and final written acceptance of the complete, validated dataset from the Acquisition Partner. A track that pays for its hours only with an assumed reward does not fit them.
Replace the placeholders with a six-week time log
SourceX has not published stage durations, stage conversion rates or typical payouts, and no cohort of platform time logs has been published either. The figures above are planning inputs until your own log replaces them. Keep one row per entry, under a code rather than a client name.
| Field | Illustrative entry | What it updates |
|---|---|---|
| Referral code | R-07 | Keeps one row per company |
| Stage | 2, Ask | Which T and w the entry feeds |
| Dates the referral entered and left the stage | 3 March to 19 March | T, measured |
| Minutes and who spent them | 40 operator, 30 relationship owner | w for each ledger |
| Founder feedback two weeks after the conversation | Useful: yes, plus one sentence | Whether the hours bought anything |
| Exit and reason | Declined in SourceX review | Removes the referral from the active count |
After six weeks, set w to the median weekly minutes per referral in each stage divided by 60, and T to the median days in stage divided by 7. Ask founders the feedback question through someone other than the operator, so politeness does not inflate it.
Expand, hold or stop when the queue moves
Write the thresholds before the first ask; these defaults are planning choices, not SourceX benchmarks.
| Signal in the log | Decision | What follows |
|---|---|---|
| Front-stage hours within budget for four straight weeks, founder questions answered within two business days, no relationship owner declining for lack of time | Expand intake | Raise new referrals per week by no more than a quarter, then measure again |
| The queue of screened companies waiting for an ask grows two weeks running, or measured hours run 120% to 150% of plan | Hold | Add no new companies until the queue falls |
| Core founder requests wait longer than before the track started, hours exceed 150% of plan for a month, or a founder reports feeling pressed | Stop new intake | Finish the asks already promised; submitted companies stay in SourceX's review |
A team whose log says stop (Illustrative)
Illustrative: an invented two-person platform team.
The operator budgets 4 hours a week and plans 0.75 new referrals a week, a planned load of 3.75 hours at the calculator's 5.0 hours per referral. Six weeks of logging change two rows. Preparing takes 2.0 hours, because the fund's records on headcount and systems are thinner than assumed. The ask stage takes three weeks at 1.5 hours a week, because CEOs send follow-up questions and relationship owners want a briefing call. Hours per referral rise to 8.5, and the load at the planned intake reaches 6.4 hours a week, 160% of budget.
The rest of the log agrees. The queue of screened companies waiting for an ask grew from two to seven in four weeks. Two founders' requests for candidate introductions, the platform's core service, waited more than a week. Reward income is $0, as planned, and it decides nothing.
Cutting intake to 0.47 a week, about two a month, would fit the hours. But the team's written rule says stop when core requests slip, and they did. It stops new intake, finishes the three asks already promised, and leaves the companies already submitted in SourceX's review, since stopping withdraws nobody. Whether a lighter, on-request version deserves a place later is the question in when a referral program distracts from core portfolio support.
Where the calculator breaks
- Averages take time to arrive. Little's law describes long-run averages in a system whose inflow and outflow balance. In the first months the list is still filling, so the table overstates the active count, though not the front-stage load.
- Hours are not attention. A 20-minute question that breaks a focused afternoon costs more than 20 minutes, and questions cluster when a company reaches contracting, so an average w of 0.1 can hide a week of three hours. Batch review-stage questions into a set weekly slot.
- Small numbers. Six weeks and a handful of referrals give noisy weights; treat gaps under half an hour as ties.
- Criteria move. SourceX screens against the specific program's criteria, and a company outside them today may match a later program, so screening time can shift with the program.
- Other ledgers run out first. A relationship owner asked twice a month or a CEO asked at a bad moment can end a track that the operator's hours would still allow.
Method note. The stage list follows SourceX's published referral stages, and the payment conditions come from the program terms, Version 2026-10-22.2, since replaced by version 2026-11-04, both checked 26 October 2026. T and w are Illustrative planning inputs, not measurements. Not measured: SourceX review durations, stage conversion and rewards, none of which is published. To count hours per qualified introduction instead of per stage, see whether data referral work is worth a platform team's time; to choose how asks reach founders in the first place, see research, events or direct introductions; and the overview for venture platform teams covers the program from the fund's side.
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 the calculator shows front-stage hours that fit your week with income set to zero, register for a partner link, check each company against who qualifies and read the program terms before the first ask; more operating guides sit in the guides hub.
Sizing a platform team's referral workload is a referral decision, and assessing SourceX as a company to invest in is a separate one; investor materials 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.