# Small sample referral reporting: counts first | SourceX

Canonical: https://refer.sourcex.si/guides/how-to-report-a-small-pilot-without-exaggerating-percentages
Last updated: 2026-11-04

Report a small pilot as counts over named denominators, not as percentages. "1 of 9 introduced companies has been paid; 4 of 9 are still open" says more than "11% conversion", because it shows how few cases sit behind the figure and how many can still change it. Count unique companies, give each stage its own population and reporting date, keep unresolved companies in their own column, show no percentage below 30 companies, and report rewards only as amounts actually paid, never annualized.

A fund COO or platform analyst usually writes this report for a partners' meeting or an investment committee that is used to percentages. The template below is designed so that a skeptical reader can rebuild every number from the rows, and so that the pilot's real lessons are not drowned by a ratio nine companies cannot support. Choosing the pilot companies and budgeting the work are covered in the [ten-company pilot guide](/guides/choose-a-ten-company-pilot-before-a-portfolio-wide-launch); this page starts when the results come in.

## Why nine companies cannot carry a percentage

One paid company out of nine introductions is 11%. Change one outcome and it becomes 22%, or 0%. With four of the nine still unresolved, the numerator can rise for months after the report is written. A percentage on that base mostly describes chance and timing.

The uncertainty is measurable. For 1 success in 9, an exact (Clopper-Pearson) 95% interval, which we calculated for this example, runs from about 0.3% to 48%. A reader shown "11%" will not imagine a range that wide. Federal statisticians face the same problem with small samples: the National Center for Health Statistics' presentation standards for proportions suppress an estimated proportion when the sample behind it is smaller than 30, and use the width of the Clopper-Pearson interval as a further test. A referral pilot is not a health survey, but the presentation lesson transfers. Below 30 companies, write counts.

## Stages, populations and the reporting date

Every number in the report belongs to one stage, one population and one date. Define them before any result arrives, so nobody chooses the denominator that flatters the outcome.

| Stage | Who is counted | Counting unit | Record that proves it |
|---|---|---|---|
| Screened | Companies the team checked against the published fit facts | One company per legal entity, a group counted once | The team's screening log |
| Introduced | Companies whose CEO knew about and agreed to the introduction | One company | The written introduction record |
| Applied | Companies that submitted a verified application within 90 days of the recorded introduction | One company | Status in the partner portal |
| Qualification decided | Companies whose baseline review SourceX finished, pass or decline | One company | Status in the partner portal |
| Closed and paid | Companies paid by SourceX for selected data that reached Closing | One company | Status in the partner portal |
| Reward paid | Rewards with the status Paid | Dollars, by payment date | Reward status |

The population is a cohort fixed by introduction date: every company introduced between two dates. Outcomes are read on a separate as-of date. A company introduced in the pilot stays in the pilot cohort even if its result arrives a year later, which keeps later good news from being credited to a different period.

## The counts-first pilot reporting template

Illustrative. A venture platform team ran a two-quarter pilot: introductions from January 1 to June 30, outcomes read on September 30. It screened 16 portfolio and network companies, and nine CEOs agreed to an introduction. The template reads left to right: who was eligible for the stage, who reached it, who resolved without reaching it, and who is still open.

| Stage | Out of | Reached | Resolved without reaching | Still open | Sentence for the report |
|---|---|---|---|---|---|
| Passed the known fit facts | 16 screened | 11 | 5 failed a fit fact | 0 | "11 of 16 screened companies passed the fit facts we could check" |
| Introduced | 11 that passed | 9 | 2 CEOs declined | 0 | "9 of 11 CEOs agreed to an introduction" |
| Applied | 9 introduced | 7 | 1 window lapsed | 1 inside its 90 days | "7 of 9 applied; 1 can still apply" |
| Passed qualification | 7 applied | 4 | 2 declined | 1 pending | "4 of 7 passed qualification; 1 decision pending" |
| Closed and paid | 4 that passed | 1 | 1 declined in evaluation | 2 in evaluation | "1 of 4 has closed and been paid; 2 are still in evaluation" |
| Reward received | 1 company | USD 55,000 on USD 100,000 paid to the company | Not applicable | Not applicable | "One reward of USD 55,000 was paid in Q2" |

Every cell is a count of companies a reader can trace. The last row is arithmetic from the published terms: a company's first qualifying transaction pays the greater of $50,000 plus 5% and 15% of Actual Supplier Payout, with no cap, so on the USD 100,000 SourceX paid that company the reward is USD 55,000. It is a single Illustrative observation, not an average.

The same pilot, written the way small results usually get written, and what each line hides:

| Line that exaggerates | What it hides | Counts-first replacement |
|---|---|---|
| "11% of introductions converted" | Nine cases, four still open, a range of 0.3% to 48% | "1 of 9 introduced companies has been paid; 4 of 9 are open" |
| "78% application rate" | Two companies, one of which can still apply | "7 of 9 applied; 1 can still apply" |
| "33% decline rate" | Whether the base is 6 decided cases or 7 applicants | "2 of 7 applicants were declined at qualification; 1 is pending" |
| "Annualized reward run rate of USD 220,000" | One USD 55,000 reward multiplied by four quarters | "One reward of USD 55,000 was paid; no run rate is estimated" |

## Reconciling unique companies and unresolved outcomes

At the as-of date, every introduced company sits in exactly one state, and the states add up to the cohort. If they do not, a company has been counted twice or lost.

| State on September 30 | Companies |
|---|---|
| Paid | 1 |
| Declined by SourceX, at qualification or in evaluation | 3 |
| Did not apply within 90 days | 1 |
| Open: can still apply, pending qualification or in evaluation | 4 |
| Total introduced | 9 |

Three rules keep the rows honest. Count legal entities, not contacts, meetings or applications, and count a parent company with its subsidiaries once. Record a company already known to SourceX as a duplicate rather than a decline: under section 3 of the [program terms](/terms), any earlier lead, account or email makes a company a duplicate, which says nothing about the pilot's judgment. And report declines as SourceX's decisions on its current data needs, not as verdicts on the company: [who qualifies](/who-qualifies) says SourceX applies each program's own criteria, so a company declined for one program can still suit another at a later date.

## Report the learning, not an annualized reward

A two-quarter pilot cannot estimate income, and the program gives it no basis to try: SourceX does not publish conversion rates or typical payouts, and a reward is payable only after the Acquisition Partner's final written acceptance, completed handoff and further conditions. What a small pilot can report well:

- **Time between stages, listed rather than averaged.** Days from introduction to application for the seven applicants: 3, 5, 8, 12, 15, 21 and 40. The median is 12; with seven values, show all seven.
- **Hours spent per stage** by the platform team and by founders, from a log kept during the pilot.
- **Reasons for each resolution** in categories: fit fact failed, CEO declined, window lapsed, declined by SourceX.
- **Founder usefulness** as counts of answers to one question asked the same way each time.
- **Money only when paid**, in the period paid, with no average built from one observation.

The [learning report for an unsuccessful pilot](/guides/what-an-unsuccessful-pilot-can-teach-a-venture-platform) sorts outcomes into failure types, and the [no-revenue pilot budget](/guides/a-referral-pilot-with-no-revenue-assumptions) keeps income at $0 while the pilot runs; both pair with this template.

## How the counts were defined, and what they leave out

The template measures stage counts for one introduction cohort, read on one as-of date, from three records: the team's screening log, written introduction records and the statuses SourceX shows in the partner portal. The 30-company floor is borrowed from the NCHS presentation standards for proportions, checked on 2026-10-26, which were written for survey estimates; we use the threshold as a presentation rule, not as a statistical requirement. The interval of 0.3% to 48% is our own exact binomial calculation for the Illustrative numbers. The template does not measure demand for any company's records, record quality, payout size or what SourceX would decide about a company still open.

## Where counts-first reporting can mislead

- **It can hide a real signal.** A pilot that genuinely works looks modest in counts. Once a cohort passes 30 companies, show percentages with their intervals.
- **Selection.** Pilot companies are usually chosen for clean evidence, so a result does not extrapolate to the whole portfolio.
- **Changing criteria.** SourceX's requirements vary with its current data needs, so cohorts screened under different criteria are not comparable without saying so; the [qualified-cohort growth method](/guides/measure-growth-in-qualified-introductions-not-logo-counts) handles cohorts across quarters.
- **The as-of date cuts deals mid-flight.** Payment follows acceptance, handoff and several conditions, so a report closed at quarter-end will show open cases that later resolve either way.
- **Confidentiality.** The partner agreement bars disclosing non-public deal details without written authorization, and SourceX shares updates as generic stages or reference codes. A report to an investment committee or LPs should carry counts and reference codes, never company names or deal terms.

## 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 next cohort is ready, [register for a partner link](/register), screen each company against [who qualifies](/who-qualifies), record the cohort dates before the first introduction, and keep the [program terms](/terms) beside the template so every stage matches the program's own definitions. The [funnel stages page](/resources/referral-funnel-benchmark-report) describes what each portal status means.

Reporting a referral pilot and evaluating SourceX as a company are separate decisions; readers assessing SourceX itself can request investor materials through the [SourceX investor overview](https://sourcex.si/investors).

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.

## Sources

- https://refer.sourcex.si/terms
- https://refer.sourcex.si/who-qualifies
- https://sourcex.si/documents/partner/SourceX_Channel_Partner_Agreement_Sample.pdf
- https://www.cdc.gov/nchs/media/pdfs/2024/09/sr02_175.pdf
- https://sourcex.si/investors