This decision tree is part of the headcount planning template, which produces the monthly counts the tree reads. For the counting rules in question form, see how to count full-time employees.
The tree, read top to bottom
Start with the company's whole history, not this year. Answer each node for the strongest candidate month, then follow the branch.
Node A. Is there any month in which 50 or more people were paid as employees?
- Yes: go to Node B with that month.
- No: the company is below baseline; stop. If today's headcount looks larger than that, the side branch explains why.
Side branch. Are today's numbers inflated by contractors? A team of 30 employees and 45 freelancers looks like 75 people on an org chart but counts as 30 at most. Contractors, agency staff and freelancers paid as vendors never enter the count, whatever their hours. The IRS treats a worker as an independent contractor when the payer has the right to control only the result of the work, not how it is done; for this baseline, follow how the company actually classified and paid each person, not how the work felt. Return to Node A using employees only.
Node B. In that month, were at least 50 of those employees classified full-time by the company?
- Yes: go to Node C.
- No, but close: check the neighboring months; peaks often sit a month either side of a quarter end. If no month reaches 50 full-time, below baseline.
Node C. Who ran payroll that month?
- The company itself: go to Node D with the company's roster.
- A PEO: a certified PEO is generally treated as the employer for employment tax purposes, so the company's own tax filings may show few employees. Use the PEO's worksite employee report and go to Node D. The PEO headcount question covers eligibility for that setup.
- A staffing firm: those workers are usually the staffing firm's employees, not the company's; see whether staffing firm temporaries count. Remove them and return to Node B.
Node D. Can a primary record show the month and the status?
- Yes, the roster shows full-time status with start and end dates: meets baseline, subject to SourceX's review.
- Only indirectly, through tax filings or benefits enrollment: check further before introducing.
Which record closes each node
| Node | Primary record | Cross-check | Held by |
|---|---|---|---|
| A, 50 paid as employees | Payroll register by month | Form 941 line 1, which counts employees paid in the pay period including the 12th of the quarter's last month | Payroll provider, finance |
| Side branch, contractors | Accounts payable vendor list | 1099 or local equivalent filings | Accounts payable |
| B, full-time status | Employment status field in the HR system | Benefits eligibility rules and enrollment | HR, benefits broker |
| C, who ran payroll | Payroll provider contracts by year | PEO worksite reports | Finance, the PEO |
| D, proof of the month | Roster export viewed by the controller | Year-end headcount summary | Controller |
Form 941 is useful as a ceiling, not as proof: it counts everyone paid in one pay period, part-time staff included, and shows nothing about status.
Three illustrative companies through the tree
| Company (illustrative) | Today | Best month on record | Path | Result |
|---|---|---|---|---|
| A 26-employee design agency with 50 freelancers | 76 people working, 26 employees | 31 full-time employees in 2023 | A no, side branch removes 50 freelancers | Below baseline |
| A software services firm that shrank after 2022 | 41 employees, 12 contractors | 88 full-time employees in May 2022 | A yes, B yes, C own payroll, D roster shows status | Meets baseline |
| A home health operator that used a PEO until 2021 | 60 employees, own payroll | 74 workers in 2020 on the PEO's reports, status not recorded | A yes, B unclear, C PEO, D indirect | Check further |
The second company is the one partners overlook: today it looks too small, but its history passes. The third needs the PEO's reports to show full-time status before anyone introduces it, or a later month on the company's own payroll that reaches 50 full-time. The downsized company question covers the shrinking case in more detail.
Peak full-time count versus the ACA's 50
Many finance teams already know one "50 full-time" test: under the Affordable Care Act, an applicable large employer averaged at least 50 full-time employees, including full-time equivalents, in the prior calendar year. That test is not the referral baseline. It averages over a year and adds up part-time hours into equivalents. The referral baseline looks at full-time employees in a single best month, contractors excluded, and ignores equivalents.
So a company with 40 full-time staff and many part-timers might be an ALE and still fall short of the referral baseline; a company that peaked at 60 full-time for one season might pass the baseline without ever being an ALE. Ask which test the controller is answering.
What the partner asks and what stays with the company
The partner asks for two things: the peak number and the month. Everything else stays with the controller or CFO. Headcount is also only one of four baseline checks; the company needs documented operating history, rights to license its records and an authorized contact. 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 90% of the data is in English. The forms above are U.S. examples; elsewhere the national payroll register and social-insurance filings play the same roles.
Passing the tree does not move money. If a company later licenses data, SourceX pays it under its agreement and funds any referral reward separately, calculated on what SourceX actually pays the company. The eligibility checker runs all four checks together.
Next step
Refer a company. Earn up to $100,000. Default reward: 25% of Actual Supplier Payout, capped at $100,000 cumulatively per referred company. Eligibility and payment conditions apply, and a signed Schedule A or addendum can differ. A lead, meeting or signature alone does not trigger payment.
When the tree ends on "meets baseline", register as a partner, confirm the other checks on who qualifies, and read the terms and rewards page before you make the introduction.
Why can peak headcount change a referral shortlist?
Because the eligibility test reads a company's best month, and most shortlists read today. Sorting on current headcount drops companies that once had 50 or more full-time employees and still hold the records from those years, and keeps companies whose numbers today are padded with contractors or agency workers. Re-sort on the documented full-time peak, with its month, its source and the records that survive from it, and the list moves in both directions before anyone has asked a CEO for anything.
The decision tree on this page settles one company. A platform team starts from the other end: forty names from a portfolio review, a co-investor's holdings and the operating network, each with one headcount number of unknown origin. Three people answer three different questions about every row:
| Question | Who answers it | What the partner writes down |
|---|---|---|
| What was the full-time peak, and when? | The controller or CFO, from payroll history | One number, one month, the record behind it |
| Does the company want a licensing conversation? | The CEO, who authorizes it | Whether the CEO knows and agreed |
| Who can ask the CEO and be heard? | The relationship owner: a founder, board member, LP principal or the platform lead | A name, and whether they agreed to ask |
Portfolio membership answers none of them. A shareholding gives the fund neither the controller's number nor the CEO's agreement, and SourceX usually finds no fit where the decision maker was left out of the introduction.
The peak-versus-current headcount evidence sheet
One row per company, metadata only. Nobody sends a roster.
| Field | What goes in it |
|---|---|
| Current people | Everyone working today, employees and contractors, with where the number came from |
| Peak full-time count and month | The highest month of full-time employees on record |
| Grade of the peak | A: payroll roster showing full-time status that month. B: an indirect record such as Form 941, benefits enrollment or a board deck. C: the owner's recollection. D: a public estimate |
| Contractor share today | Contractors and agency workers as a share of current people |
| Records from the peak years | Systems in use then, years covered, and whether still paid for or exported: held, partial or gone |
| Ownership since the peak | Same owner, acquired, a unit divested, or under lender or court control |
| Capacity today | A named person who can still run exports and answer questions, or none |
| Relationship owner and CEO | Who will ask, whether they agreed, and whether the CEO knows |
Grade A matches the tree's "meets baseline" branch and grade B its "check further" branch. Grades C and D are reasons to ask, never results.
Three companies on one sheet
Illustrative. Generic companies, invented figures, no outcomes implied.
| Field | A. Owner-led premium audit firm | B. Fulfillment operator, through a portfolio founder | C. Software reseller, through an LP |
|---|---|---|---|
| Current people | 47 (41 employees, 6 contractors) | 85 (34 employees, 51 agency workers) | 30 employees |
| Peak full-time, month, grade | 96, March 2019, A | 38, November 2024, A | About 71, mid 2020, C |
| Records from the peak years | Eleven years of audit files in one case system | Four years of orders and returns | Ticketing system cancelled in 2023, no export |
| Capacity today | Operations manager can run exports | Systems lead in place | Nobody left who ran the old system |
| Relationship owner | The platform lead, directly; the founder is CEO | The portfolio founder; B is its largest customer | The LP principal |
| Sorted on current headcount | Dropped at 47 | Kept at 85 | Dropped at 30 |
| Sorted on the sheet | First | Off: full-time peak below 50 | Off: the peak-year records are gone |
Company A is the row a current-headcount screen loses: it shrank by half, but its founder still runs it and the files from its larger years are intact. Company B is the row it wrongly keeps: its 51 agency workers are the agency's employees, not B's, and the portfolio founder is spared spending goodwill on an introduction that would fail the first check. Company C is the counterexample. A peak that may well pass is worth nothing when the records created at that size were not kept; it stays off unless another system from those years, such as an email archive, survives.
Rank the next conversations without projecting proceeds
SourceX has not published conversion rates or typical payouts, and demand, rights and record depth are assessed only in its review, so leave proceeds out and sort on five observable keys, in order: a grade A peak of 50 or more; peak-year records held, with the system named; a confirmed relationship owner; capacity to run exports; a dated event, such as a documented cancellation or archive deletion.
| Rank | Company (Illustrative) | Peak and grade | Peak-year records | Next action |
|---|---|---|---|---|
| 1 | A, premium audit firm | 96, A | Held | Confirm the CEO still wants the call, then introduce |
| 2 | 140-person field-services group held by a co-investor | 140, B from a board deck | Held | Controller reads full-time status for that month |
| 3 | Former portfolio company, now 39 people | About 75, C | Unknown | Former board member asks for the month, the record and what survived |
| 4 | B, fulfillment operator | 38, A | Held | Off; revisit if its own full-time employees pass 50 |
| 5 | C, software reseller | About 71, C | Gone | Off unless records from 2019 to 2021 survive elsewhere |
The headcount planning template produces the monthly counts a sheet needs, and the guide to finding a private company's employee count covers public estimates, which stay grade D here.
What can shrinking headcount mean for referral eligibility?
On its own, nothing in either direction. The baseline reads the full-time peak, so a company that passed 50 at its best is judged on that peak however far it has fallen. A shrink changes what surrounds the number: whether the records from the larger years survived the cuts, whether the authorizer is the same, and whether anyone inside can still run the process. Treat it as a discovery signal that tells you what to ask, not as evidence of eligibility or of what the records are worth. The downsized company question gives the short yes; this section reads the shape of the decline.
Where the signal comes from
| Signal | What it can show | What it cannot show |
|---|---|---|
| A portfolio company's board pack | The month of the peak and the slope since | Full-time status, since contractors and open roles are often mixed in |
| A WARN notice | That the employer met a coverage threshold on a date | A company-wide full-time count |
| The controller's answer from payroll | The peak, its month and its status | Whether the records survived, a separate question |
According to the Department of Labor, federal WARN covers employers with 100 or more employees, generally not counting workers with under six months on the job in the past year or those averaging under 20 hours a week, and requires 60 calendar days' notice of plant closings and mass layoffs. A federal filer was large at the time, but the notice describes a site, not company-wide full-time status; the WARN notice guide covers reading a listing.
Peak-history screening examples: two shapes, the same six questions
Illustrative. Two generic companies with the same peak and the same headcount today; only the shape differs.
| Question | Plateau, then slow decline: a 26-year-old title and escrow company | Spike, then correction: a software company founded in 2018 |
|---|---|---|
| 1. Peak full-time, month, grade | 70, April 2019, A | 70, June 2022, A |
| 2. Shape | 55 to 70 full-time from 2012 to 2022; 46 today after automating closings | 18 in 2020; 70 after a funding round; 46 after two layoffs in 2023 |
| 3. Years of records created at 50 or more | About ten of seventeen years on file | About one of eight |
| 4. Peak-year records held | Closing files and escrow ledgers in two systems | Tickets and code kept; a sales tool cancelled in the cuts |
| 5. Authorizer and capacity | Founder is CEO; controller and an IT contractor can export | CEO in place; the engineers who ran the old tools left |
| 6. Whose records | Its own, but some files are held for lenders and clients under contract | Internal records are its own; product data sits in customer tenants |
Question 3 tests whether the signal reveals relevant operating history. A shrink proves only that a larger past existed; the share of the record created at that size shows how much of it reflects the larger operation. SourceX publishes no minimum for years at peak and its screening profile asks for several years in operation, so the ratio picks the next question, not the verdict. Both shapes tie on the first sort key. The plateau company keeps its records and the people who can export them, so its next step is an introduction; the spike company lost a system and its operators, so its first conversation is about what survived.
The cause of a shrink matters as much as its size:
| Cause | What it changes | Question to ask |
|---|---|---|
| A unit was divested | Its systems and years may have gone with the buyer | Which stayed? See carve-out data separation |
| Work moved to contractors or an outsourcer | The peak stands; later records may sit in a vendor's systems | Who holds the records created after the change? |
| A wind-down | Systems are switched off on a schedule | What is the decommissioning date? See winding down without losing records |
| A lender, trustee or court controls the assets | The CEO alone may not be able to authorize | Who controls the assets today? |
A documented decommissioning date moves a shrunken company up only when the other keys hold. A sense that a company is in decline is not a date.
Where this sheet can be wrong
- Unchecked peaks. A grade C number can outrank a checked one; keep C rows below every A and B row until a record confirms them.
- Program thresholds. SourceX says some programs set higher size thresholds and smaller specialists can still be considered, so 50 is a screening line, not a promise either way.
- Unknowns until review. Demand, rights and record quality are decided in SourceX's review; the sheet measures none of them.
- Distress. A sharp shrink can bring lender or court control, where the CEO alone may not be enough.
- Relationship cost. Asking a founder or an LP to raise this spends their goodwill, and no data on how often they decline has been published.
Methods note. The sheet, grades and sort order were built on 2026-10-26 from SourceX's program terms, the who-qualifies screening profile, this page's decision tree and the Department of Labor's description of WARN coverage. The sheet records sources and dates, not outcomes, and no ranking made with it has been tested against results.
When a row reaches grade A with its peak-year records held and a relationship owner who agreed to ask, set up a partner account, measure the grade A row against who qualifies once more and read the program terms in full. Whoever submits signs the SourceX Channel Partner Agreement first; among competing partners, only the author of the first valid introduction made during the 90-day attribution window is credited, and only after the company submits a verified application. The overview for venture platform teams and the guides library cover the rest of the portfolio workflow.
Building a referral shortlist and evaluating SourceX as a company are separate decisions, and the SourceX investor overview takes requests for investor materials from anyone considering the second.
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.