# Measure referral quality, not volume, for data licensing

Canonical: https://refer.sourcex.si/guides/how-to-measure-referral-quality-instead-of-referral-volume
Last updated: 2026-11-04

To measure referral quality, focus on whether referred companies in the U.S., Canada, the EU, the UK, Australia, New Zealand, India, Singapore, South Africa, Latin America, the UAE and Saudi Arabia (outside the U.S., Canada and the EU, at least 70% of the data must be in English) meet SourceX's eligibility criteria for data licensing, possess original data rights, have an authorized sponsor, and align with active SourceX's current data needs, rather than simply the number of introductions made.

Referral programs often focus on the sheer number of leads, but for specialized data licensing, quality always trumps volume. Understanding what makes a high-quality referral for SourceX helps you focus your efforts where they matter most, leading to more successful outcomes for you and the referred business.

### Why quality matters more than quantity

Unlike traditional sales where many leads might eventually convert, data licensing requires specific conditions to be met for an opportunity to progress. A large volume of unqualified referrals can consume time without yielding results. By focusing on quality from the outset, you increase the likelihood of your referred company completing a qualifying data licensing deal and your reward becoming eligible for payment.

### Key indicators of a high-quality referral

Instead of just counting introductions, measure your referrals against these criteria:

1.  **Alignment with eligibility profile:** Does the referred company fit the [initial screening profile](/who-qualifies)? This includes operating maturity (e.g., 50+ FTEs at peak, several years in operation), and whether they produce original, proprietary data. Companies that do not meet basic criteria are less likely to progress.

2.  **Possession of original data rights:** A crucial quality indicator is whether the company owns or can license its data without legal or ethical breaches. This means the data isn't client-owned, scraped, or personal consumer data. High-quality referrals always involve companies in the U.S., Canada, the EU, the UK, Australia, New Zealand, India, Singapore, South Africa, Latin America, the UAE and Saudi Arabia (outside the U.S., Canada and the EU, at least 70% of the data must be in English) with clear data ownership and rights.

3.  **Authorized sponsor engagement:** A high-quality referral comes with the explicit knowledge and backing of an owner, executive, or designated decision-maker within the referred company. Without an authorized sponsor, discussions about data licensing (which is a strategic decision) rarely move forward.

4.  **Relevance to SourceX's current data needs:** While SourceX ultimately assesses this, you can gauge relevance by considering if the company's operational data seems unique or valuable for AI development in a specific industry. High-quality referrals present data types that are likely to attract AI developers and buyers seeking proprietary insights.

### How SourceX evaluates quality

SourceX's six-stage referral process inherently filters for quality at each step. Your referral's progress through these stages is a direct measure of its quality:

*   **Stage 3: SourceX reviews the opportunity.** This is where initial quality checks on the company's operating history, data, data rights, and relevance to SourceX's data requirements are performed. A high-quality referral is more likely to pass this stage efficiently.
*   **Stage 4: The company completes a qualifying data licensing deal.** A truly high-quality referral results in a company successfully negotiating terms and agreeing to a deal.
*   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.

### Tools to help you assess quality

To improve the quality of your referrals, consider using the [company fit checker](/tools/company-fit-checker). This tool provides a preliminary, non-binding screening to help you identify businesses in the U.S., Canada, the EU, the UK, Australia, New Zealand, India, Singapore, South Africa, Latin America, the UAE and Saudi Arabia that are a better match for SourceX's programs, before you even make the introduction. By pre-screening, you increase the likelihood of referring companies that genuinely qualify.

Focusing on these quality indicators not only streamlines the process for SourceX but also increases your potential for earning rewards. Investing time in identifying suitable companies in the U.S., Canada, the EU, the UK, Australia, New Zealand, India, Singapore, South Africa, Latin America, the UAE and Saudi Arabia yields better results than a high volume of unqualified leads.

> Rewards are calculated as the greater of $50,000 plus 5% and 15% of the compensation SourceX actually pays the company you refer, with no cap. A lead, meeting or signed agreement alone does not trigger payment. See the [program terms](/terms).

## Why is a fund's address book not a qualified pipeline?

Because an address book lists people, and a qualified introduction needs much more than a person: a company that passes the published fit tests, a named colleague who owns the relationship with its CEO, records the company controls itself, and a CEO who knows about the conversation and wants it. Each is a separate stage, and most entries stop before the last. Count the network as a waterfall from contact to consenting CEO, and call only the bottom row pipeline.

Pipeline here means introductions you can make, not the [data pipeline](/proof/what-is-a-data-pipeline) that moves records after a company signs. Three kinds of claim run through this section:

- **On the record (checked 2026-10-26).** SourceX's screening profile asks for roughly 50 or more full-time employees at peak, several years in operation, original documentation the company created itself, rights to license it and an authorized sponsor. Material that belongs to clients is usually not a fit, nor is an introduction the decision maker never heard about. The CEO authorizes a licensing discussion.
- **Decided later.** Acceptance, scope and any reward, settled by SourceX's review and signed agreements.
- **Our working claim.** In most venture address books the steepest drop comes at the size-and-history stage, because venture networks lean toward young companies. The Census Bureau's Business Dynamics Statistics show startups creating jobs at higher rates while older, larger firms employ most workers. That is a national pattern, not a measurement of any fund; the waterfall tests it on yours.

| Term | Meaning in this section |
|---|---|
| Operating company | The business that employs people and keeps records, not a holding vehicle or a brand |
| Relationship owner | The person in your network who can ask that company's CEO and be heard |
| Consenting CEO | A CEO who knows a licensing conversation is proposed and wants it |
| Qualified pipeline | Rows at the last stage of the waterfall, and nothing above them |

## Which company sits behind each entry, and who owns the way in?

Resolve every entry to a company before you score it. People change jobs, three partners know the same CFO, and many LP entries are investors with no operating company to introduce. For each entry, write down the operating company today, who controls it (founder, private equity owner or parent group), its CEO and the person who can ask that CEO. Keep one row per operating company; a parent and its subsidiaries count as one company for the reward cap.

Owning shares never makes the fund the authorizer. The CEO decides, and the relationship owner is whoever the CEO will actually listen to, often a founder or LP rather than the deal partner. The [overview for venture platform teams](/for/venture-platform-teams) covers the platform side once the owner is found.

## The relationship-to-eligibility conversion waterfall

Each stage asks one question, and a row moves down only when the answer is a confirmed yes.

| Stage | Question that moves a row down | Who can answer it | Most common reason a row stops | Illustrative count |
|---|---|---|---|---|
| 1. Address-book entries | None yet: everyone you can email | Your CRM | Not applicable | 420 |
| 2. Operating companies | Which company, today, does this person lead or work for? | Public records, then the contact | Duplicates, job changes, investors and advisers with no company to introduce | 240 |
| 3. Size and history | 50 or more full-time employees at peak, contractors excluded, and several years of operations? | Public facts, then the company | Too young or too small; a staffing firm counting placed workers | 64 |
| 4. Relationship owner | Will a named person ask the CEO? | Colleagues, founders, LPs | Nobody owns the CEO relationship, or the owner will not spend the goodwill | 29 |
| 5. Records it controls | Did its own staff create the records, rather than holding them for clients? | The company, through the owner, at metadata level | Records live in clients' systems or under client contracts | 17 |
| 6. Consenting CEO | Does the CEO know about the conversation and want it? | The CEO | Not now; or the company already talks to SourceX, which makes it a duplicate | 7 |

Read the counts as a shape, not a rate. They describe one fictional fund's address book, invented to show how the method behaves; SourceX has not published conversion rates, and no cohort data stands behind them. The lesson holds anyway: the 7 at stage 6 are the pipeline, the 29 at stage 4 are a to-do list and the 240 at stage 2 are a directory.

**Method note.** Stages follow the published fit criteria (who qualifies and the terms, checked 2026-10-26) in order of cost: public facts, then a colleague's time, then the company's attention, then the CEO's. The order also protects permission, since asking about records before a relationship owner agrees means contacting a company without consent. The waterfall measures reachability and fit, not demand, record quality or payout, which only SourceX's review assesses.

## Reachable contacts or mandate-fit companies: three rows tested

A reachable company has cleared stage 4 at best; a mandate-fit company clears stages 3 and 5 on confirmed facts. A row needs both, and funding status is never a stage.

Illustrative. Generic companies, invented facts, no outcomes implied.

**A. An owner-led company.** A 180-person civil engineering consultancy, run by its founder as CEO since 1998, keeps project files, inspection reports and change logs in its own document system. The founder co-invests as an angel in two of the fund's deals, and the partner beside him on both owns the relationship. Within a week every stage is confirmed: pipeline.

**B. A network intermediary.** A 450-person medical billing company is tied to the fund only through a portfolio company's CFO, who worked there for six years and still speaks with its CEO. Stage 4 depends on the CFO agreeing to ask. Stage 5 is one the partner cannot settle: billing records touch patient information and the contracts of the providers it serves, which is what SourceX's rights review is for. The partner asks only which record types exist and which system holds them.

**C. The no-fit counterexample.** A franchise brand with roughly 600 people across its system is the fund's best-known name. Its franchisor employs 45 full-time staff, and the operating records sit with 80 independently owned franchisees, each a separate company with its own CEO. A warm relationship with the franchisor's CEO does not help: the row fails stage 3 on the franchisor's headcount and stage 5 for the network's records. A big brand is not a big company; a franchisee that passes on its own is a separate row.

## How do you use industry knowledge without stereotyping data quality?

Use a sector to choose your questions, never to answer them. Industry labels describe what a company sells; how deep its records go, how well they are documented and who owns them vary inside every industry. Treat each sector belief as a hypothesis with a named public clue and a named person who can confirm it.

Industry codes show why. Under the 2022 North American Industry Classification System manual, an establishment is classified in an industry when its primary activity meets that industry's definition. A code describes activity, not record depth or ownership, and so does the sector label in a partner's head.

### The evidence-led sector screening worksheet

| Sector belief | What the belief skips | Public clue worth logging | What only the company can confirm | Route |
|---|---|---|---|---|
| "Healthcare businesses are data-rich" | Much of the material is patient information governed by client and provider contracts | Years in operation; a named practice or billing system in job ads | Which records are its own operating documents | Ask the owner; flag rights early |
| "SaaS companies are data companies" | Product content usually belongs to customers | A public changelog or engineering blog going back years | Whether its own engineering and support history survives | Ask; product data stays out |
| "Construction runs on paper" | A firm may keep years of requests for information, submittals and change orders in project software | Job ads naming a project management platform; dated project lists | How far back the digital record reaches | Do not drop on the label; ask |
| "Staffing firms are large" | Placed workers do not count toward the 50 | Internal recruiter and back-office roles in job ads | Own full-time headcount at peak, stated separately from the placed workforce | Keep in when records are strong |
| "Professional firms have the best documents" | Much of the work product belongs to clients or sits under confidentiality | Practice areas and years in operation | Which material is the firm's own: methods, templates, internal guides | Ask; expect a narrower scope |
| "Old family businesses have nothing digital" | Some have migrated systems more than once | A dated system migration or sale notice | What survived each migration | Ask; a dated migration is real urgency |

It adds one rule to the [screening worksheet for consulting businesses](/resources/a-referral-screening-worksheet-for-professional-consulting-businesses): every row starts as a belief and ends confirmed, dropped or marked "no contact".

### Observed, inferred and confirmed stay in separate columns

Log what anyone can see, with source and date, as **observed**; your reading of it as **inferred**; and only what the company said through the relationship owner, with date and speaker, as **confirmed**. An inference never moves a row down the waterfall, and purchased contact lists never enter it.

### Ask permission and route only plausible introductions

- **Confirmed at stages 3 and 5, owner agrees:** the owner asks the CEO; on a yes, a partner who has signed the Channel Partner Agreement makes the introduction.
- **A sector clue and nothing else:** the owner asks one metadata question: which records, which system, how many years.
- **A strong clue and no relationship:** mark the row "no contact". A 300-person dental lab group profiled in a trade journal, with nobody in your network who knows its CEO, stays there.
- **A confirmed fact fails:** drop the row, record why and name the trigger that would reopen it.

Whether to approach a sponsor first or the CEO directly is weighed in [introducing through a sponsor or directly to the CEO](/compare/introduce-through-sponsor-or-directly-to-ceo).

## Rank the next conversations without projecting licensing proceeds

Score each row by the stages it has cleared on confirmed facts, counting an inferred pass as zero. Break ties by which next question costs the relationship owner least, then by a dated event such as a scheduled system migration. Never rank by guessed proceeds: SourceX has not published typical payouts or conversion rates, so that ranking would sort your guesses.

| Rank | Row (Illustrative) | Stages cleared on confirmed facts | Next question | Who asks it |
|---|---|---|---|---|
| 1 | Case A, engineering consultancy | 6 of 6 | None: introduce | The co-investing partner |
| 2 | 210-person claims administrator owned by an LP | 4 of 6 | Which records its staff created | The LP principal |
| 3 | Case B, medical billing company | 3 of 6 | Whether the CFO will ask | The platform lead, to the CFO |
| 4 | Dental lab group from a trade article | 1 of 6 | None until a path exists | Nobody: no contact |
| 5 | Case C, franchise brand | Fails stage 3 | Reopen only for a franchisee that passes alone | Not applicable |

## Where this waterfall could be wrong

- **The counts are invented.** Your fund's shape may differ sharply; a growth-stage portfolio loses fewer rows at stage 3 and more at stage 5.
- **Rights.** Visible records often belong to customers or clients, which SourceX's screening guidance treats as usually not a fit.
- **Demand.** SourceX applies each program's own criteria and checks relevance in review, so a row confirmed at all six stages can still be declined today and fit a later program.
- **Synthetic data.** Epoch AI's report on AI in 2030 finds the likely scaling bottlenecks surmountable and says combining synthetic and multimodal data should close the data gap. If so, demand for private business records could narrow.
- **Relationship cost.** Asking an LP or a portfolio CFO to raise this spends goodwill, and no data on their reactions has been published; record each one.
- **Stereotypes cut both ways.** Dropping a "paper" sector loses real rows; promoting a "data-rich" one wastes asks on client-owned records.

**A measurable next step.** Run your first 100 entries through the waterfall in one sitting, noting the count and main reason at each stage. Thirty days later, find the largest drop. At stage 4, your network needs mapped relationship owners more than new contacts; at stage 3, the address book is the wrong list, and your LPs' and operators' companies are the place to look. Count rows that moved on an inference; the target is zero. The [explainer for venture platform teams](/guides/how-venture-platform-teams-can-explain-company-data-partnerships) helps when a CEO asks what the program is, and the [company fit checker](/tools/company-fit-checker) gives a quick, non-binding read on one row.

## 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 a row reaches stage 6 with a confirmed CEO yes, secure a [partner link](/register), score the company against [who qualifies](/who-qualifies) and work through the [program terms](/terms) before the relationship owner sends anything; more network playbooks are in [guides](/guides).

Evaluating SourceX itself as an investment is a separate decision from making an introduction; for that, 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://epoch.ai/files/AI_2030.pdf
- https://refer.sourcex.si/terms
- https://refer.sourcex.si/who-qualifies
- https://sourcex.si/investors
- https://www.census.gov/library/stories/2022/02/united-states-startups-create-jobs-at-higher-rates-older-large-firms-employ-most-workers.html
- https://www.census.gov/naics/reference_files_tools/2022_NAICS_Manual.pdf
