The referral metrics that survive a denominator check: a seven-test worksheet that keeps three common metrics, repairs four and retires two

Short answer

A referral metric survives a denominator check when its numerator and denominator count the same unit (unique companies, not contacts), every company in the numerator also sits in the denominator, both come from one introduction cohort with one cutoff date, every company in the denominator has had time to reach the numerator's stage, and exclusions are declared rather than dropped. Ratios built from mixed periods or contact counts fail. The worksheet below tests nine common metrics: three survive as written, four can be repaired and two should be retired.

Visual briefFull-size diagram
A matrix of nine referral metrics against seven denominator tests, with passes marked green and failures marked orange; credit rate, duplicate share and open share pass every test, four metrics are marked repair and two, pipeline value and reward run rate, are marked retire.
The worksheet's verdicts as defined on this page. Tests follow the method below; no SourceX data is shown.

This page is for the fund COO or platform analyst who inherits a referral dashboard and has to decide which of its numbers can go in front of partners. The question is narrow: not how to segment a report by network, which tracking accepted introductions by network segment covers, but which ratios are arithmetically honest.

Why a ratio can be wrong while both counts are right

A quarter-end dashboard says 7 companies were introduced and 5 applications were filed, so the application rate is 71%. Both counts are accurate. The ratio is still wrong, because 3 of the 5 applications came from companies introduced the quarter before. They are in the numerator and not in the denominator.

Epidemiologists settled this long ago. The CDC's Principles of Epidemiology glossary defines a proportion as a ratio in which the numerator is included in the denominator, and a rate as the frequency of an event among a defined population per unit of time. A referral conversion figure is a proportion, so every company in its numerator must come from its denominator. "Introductions per partner per quarter" is a rate, so it needs a defined population of partners and a stated period.

The referral program adds two complications of its own. Credit comes only when a company submits a verified application within the 90-day attribution window, so a cohort is not finished until the last window closes. And companies SourceX already knew, through any earlier lead, account or email correspondence, are reviewed as duplicates and not credited, so a denominator that ignores duplicate review overstates what the network found.

Seven tests a denominator has to pass

  1. Unit. Numerator and denominator both count unique companies. Contacts, emails and CRM entries are not companies. A parent and its subsidiaries share one reward cap, so keep them on one row; credit questions between related companies are decided by SourceX in duplicate review.
  2. Subset. Every company in the numerator also appears in the denominator.
  3. Cohort. Both counts come from the same introduction period, measured at the same cutoff date.
  4. Maturity. Every company in the denominator has had time to reach the numerator's stage. For credit, that means its 90-day window has closed.
  5. Deduplication. Companies reviewed as duplicates are shown as their own line, not silently dropped or silently counted as found.
  6. Declared exclusions. Withdrawn, declined and window-ended companies stay in the denominator. If anything is excluded, the report says what and why.
  7. Clock. A rate states its period and population; a proportion states its cohort and cutoff. Mixing the two is a fail.

The denominator audit worksheet: nine metrics tested

Metric as usually writtenTests failedRepaired versionVerdict
Credit rate: credited companies over unique companies introduced, cohort maturedNoneKeep as writtenKeep
Duplicate share: companies reviewed as duplicates over unique companies introducedNoneKeep as writtenKeep
Open share: companies still open at cutoff over unique companies introducedNoneKeep as writtenKeep
Introductions per partner: CRM entries over registered accountsUnit, deduplicationUnique companies introduced in writing, per partner with an agreement in force, per quarterRepair
Application rate: applications filed this quarter over introductions this quarterSubset, cohort, maturityCredited companies from one introduction cohort, reported once its last window closesRepair
Agreement rate: supplier agreements to date over applications to dateCohort, maturityCompanies at a supplier agreement stage over credited companies, by introduction cohort, with open cases shownRepair
Time to payment: median days, paid cases onlyMaturity, declared exclusionsDays to cash for paid companies, shown beside the count and age of every credited company still openRepair
Pipeline value per introduction: summed portal estimates over introductionsSubset, maturityNone: estimates are not outcomesRetire
Reward run rate: one quarter's rewards times fourCohort, maturity, clockNone: report cash received by periodRetire

Two retirements need a word. Portal and calculator estimates are, in the program terms' own words, projections rather than promises, and before supplier payment any reward shown is an estimate based on proposed supplier compensation. No denominator can turn an estimate into an outcome. A run rate fails for a different reason: rewards are with no cap and arrive in lumps after acceptance, handoff and payment conditions, so one quarter's cash says nothing about the next.

A two-quarter cohort, audited (Illustrative)

Illustrative. A generic fund, invented counts. Cohort Q1 is companies introduced January to March; cohort Q2 is April to June. The dashboard was built at 30 June; the audit was run at 30 September, when every window in both cohorts had closed.

CohortUnique companies introducedCreditedReviewed as duplicatesWindow ended without applicationOpen at 30 June
Q194230
Q273134 at 30 June, 0 by 30 September
Both16736Not added: different dates

At 30 June the dashboard showed a 71% application rate. The audit replaces it with two matured credit rates, 4 of 9 for Q1 and 3 of 7 for Q2, and a note that at 30 June the honest Q2 line was "2 of 7 credited, 4 still inside the window, 1 duplicate". The difference is not a rounding issue: the 71% counted three Q1 companies as Q2 successes.

Time to payment shows the second trap. One company, from Q1, has been paid, 151 days after its introduction. A "median time to payment of 151 days" describes one company and ignores the 4 credited companies still open at 30 September, aged 104 to 251 days, and the 2 declined in review. The repaired line reads: 1 paid at 151 days; 4 open, the oldest at 251 days. Anyone reading it can see the clock has not stopped for most of the cohort.

Duplicate share survives untouched at 3 of 16, and it is the most useful number on the page: it tells the platform team how often its network overlaps with companies SourceX already knows. The guide to counting one company across several networks shows how to cut it further.

From 27 CRM entries to 16 companies before any ratio

Run the unit test before computing anything. The CRM held 27 entries tagged for the program. Nine were second or third contacts at companies already listed, one was the same company under its trade name and its legal name, and one was a subsidiary folded into its parent's row. That leaves 16.

StepEntries or companies
CRM entries tagged for the program27
Less extra contacts at companies already listed9
Less a trade name matched to its legal entity1
Less a subsidiary folded into its parent's row1
Unique companies introduced16

Keep the merge log. If a later quarter's denominator moves without a matching log entry, the definition drifted.

What the audited report says, and the reward line it refuses

The audited report leads with counts and denominators, adds percentages only where a cohort is mature, and says when a cohort is too small to support one. Sixteen companies over two quarters is a learning sample, not a benchmark, and SourceX has not published conversion rates to compare it with.

Rewards appear as cash received in the period it arrived, never annualized. In the illustration, a $15,000 reward received in the third quarter is reported as $15,000 received in the third quarter, not as $60,000 a year. Open cases appear as a count with their stage and age, never as an estimated value.

Where the worksheet itself can mislead

  • Small cohorts. Every rate here rests on single-digit numerators. Report counts first.
  • Shifting definitions. If the program terms change, a metric can change meaning between editions. Record the terms version with each report; the guide to saving the applicable terms version explains why.
  • Attribution conventions. Other partner programs credit differently. PartnerStack's survey of 100 senior leaders at B2B SaaS companies with $50M or more in revenue found only 42% use multi-touch attribution across the funnel, so a benchmark built on another program's credit rule is not comparable with one partner credited per company.
  • Partner-side visibility. Stages after credit come from SourceX's generic portal statuses, so the worksheet cannot audit what the fund cannot see.
  • Survivorship in the merge log. Companies never entered in the CRM never reach the audit at all.

Methods note. The seven tests and nine verdicts are this page's own, built from the CDC glossary definitions of proportion and rate and the attribution and duplicate rules in program terms version 2026-10-22.2, since replaced by version 2026-11-04, checked 2026-10-26. The worksheet measures arithmetic consistency only. It does not measure record quality, buyer demand, any company's compensation or the cost of the introductions.

Fit facts that admit a company to the denominator

A clean denominator still needs companies that could fit. Count a company only if it was introduced in writing with the CEO's knowledge and looks plausible on the published facts: 50+ full-time employees at peak (contractors not counted), an operating history it can document, the right to license its own records and a CEO who authorizes the discussion. 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. Funding status plays no part. Relevance to current data needs is checked only in SourceX's review. The fund's partners must sign the SourceX Channel Partner Agreement before introducing anyone, though signed copies that bear the earlier title, Channel Partner Non-Circumvention and Non-Disclosure Agreement, still hold exactly as signed.

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.

Once your metrics pass the worksheet, register for a partner link so every new introduction carries a dated record, check candidates against who qualifies, and keep the program terms beside the definitions page of each report.

Auditing your own referral metrics is a different exercise from assessing SourceX as a business; anyone doing the second can request investor materials from the SourceX investor overview, where the company says it shares company and commercial metrics with investors directly.

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.

  1. Step 1Share your linkSend your personal link to a company you know.
  2. Step 2Company appliesThe company applies itself at /apply.
  3. Step 3Company signs and is paidThe company signs, delivers its data and is paid by SourceX under its agreement.
  4. Step 4You get your rewardthe greater of $50,000 plus 5% and 15% of the actual supplier payout, with no cap. A reward becomes payable after SourceX has paid the referred company under its signed agreement and the payout conditions in the program terms are met, including finance approval and your tax and payment details.

Common questions

Is it wrong to report applications per quarter at all?

No. Applications filed in a quarter is an honest count of activity. It becomes misleading only when it is divided by introductions made in the same quarter, because some of those applications came from earlier introductions. Report the count on its own, and compute ratios by introduction cohort.

When is a cohort mature enough to compute a credit rate?

When every company in it has passed the end of its 90-day attribution window, so each one is either credited, closed or reviewed as a duplicate. For a quarterly cohort that means waiting roughly 90 days after the quarter ends.

Why does the worksheet retire reward run rates instead of repairing them?

Because the problem is the numerator, not the denominator. Rewards arrive in lumps after acceptance, handoff and payment conditions, and come one company at a time, so multiplying one quarter's cash by four describes nothing that has happened. Report cash received by period instead.

Sources

  1. refer.sourcex.si/terms
  2. archive.cdc.gov/www_cdc_gov/csels/dsepd/ss1978/glossary.html
  3. partnerstack.com/resources/research-lab/the-state-of-partners…
  4. refer.sourcex.si/who-qualifies
  5. sourcex.si/investors

Free resources

Know a company with valuable proprietary data?

Become a referral partner, get your link and introduce the company’s CEO (in an owner-run company, usually the owner). Referral partners can join from most countries. We can't work with partners in countries or regions under U.S. sanctions or on U.S. restricted-party lists. The companies you introduce should be based in the U.S., Canada and the EU, or in the UK, Australia, New Zealand, India, Singapore, South Africa, Latin America, the UAE and Saudi Arabia with at least 70% of their data in English.

Refer a company →

I own a business

Explore licensing your company's data directly to SourceX for AI training and evaluation. Start a short assessment; no uploads needed.

Start an assessment