# Founder data partnership decline research | SourceX

Canonical: https://refer.sourcex.si/guides/why-do-founders-decline-data-partnership-conversations
Last updated: 2026-11-04

Nobody has published dependable evidence on why founders turn down a conversation about licensing their company's records, and this page claims none. It publishes the method for finding out: an anonymous survey, sent by an independent collector to founders and CEOs who said no after an introduction, with a yes-or-no item for each possible reason, one deciding-reason question, a sealed record of what the introducer guessed, and rules that hide any small group. No responses have been collected, and no results exist.

For a platform research director, a decline is the most informative event in a referral program and the one least often recorded honestly. Introducers explain a no in shorthand such as "bad timing", and a program that acts on those guesses can fix the wrong thing. The survey asks the founder, and measures how far the guesses miss.

## Where in the conversation the no arrives

A founder can decline at five points, and a no at each point rests on different knowledge. The survey records the stage, because a reason given before any contact with SourceX means something different from the same reason given after a call.

Two groups sit outside this frame on purpose. Owners who never answered the introducer at all belong to [the unanswered-offer study](/guides/why-business-owners-ignore-referral-offers); companies that said yes and then went quiet belong to [the stall study](/guides/the-most-common-reasons-company-introductions-stall). Single declines read as lessons for the next introducer belong to [the consent-first case protocol](/guides/what-a-declined-referral-can-teach-the-next-partner), and what founders believe before any call to [the belief survey](/guides/what-founders-think-data-licensing-means-before-the-first-call).

| Stage | The no came | What the founder knew by then |
|---|---|---|
| S1 | At the first mention | Only the introducer's description |
| S2 | After reading SourceX's own pages | Plus SourceX's published process and conditions |
| S3 | After a first call with SourceX | Plus SourceX's questions about systems and history |
| S4 | During scoping or the rights review | Plus the company's own rights picture |
| S5 | After seeing proposed terms | Plus the terms on offer to the company |

## Definitions fixed before the first invitation

Every term is defined in writing and dated before the entry window opens. A definition changed mid-study splits the cohort.

| Term | Definition used here |
|---|---|
| Decline | An explicit no to a data partnership conversation from the company's decision-maker, logged by the introducer with a date |
| Decision-maker | The founder or CEO who said no; under the program, the CEO authorizes a licensing discussion |
| Data partnership conversation | Any discussion with SourceX about licensing the company's records, from a first call onward |
| Introducer | The partner, adviser or investor who raised the idea with the company |
| Case code | A random code the collector assigns; it links one invitation to one sealed guess and nothing else |
| Contributing reason | Any reason item the founder answers yes |
| Deciding reason | The single item the founder says mattered most |

## Frame, consent and the anonymity design

The frame is every decline logged by participating introducers across two calendar quarters, at any stage from S1 to S5. Introducers opt in beforehand and agree to log every decline, the embarrassing ones included. Fit facts are not a filter: a company the introducer later learns was under 50 full-time employees at peak still enters, flagged, because the founder's reason is the point.

An independent collector, such as a research firm or an academic partner, sends the invitation. It never comes from the introducer, the fund or SourceX. There is one invitation, one reminder ten days later and a close thirty days after the invitation. The rules the collector follows:

- **Five promises in the consent text.** Answers stay anonymous to the introducer, the fund and SourceX; no company name is asked, only a size band and sector; any question can be skipped; no answer is used to approach the company again; free text is quoted only if the founder ticks a box allowing it.
- **The key is destroyed.** The collector deletes the table that links contact details to case codes as soon as invitations close.
- **Small groups stay hidden.** No cell with fewer than 11 respondents is published, and no percentage that would let a reader work one out. CMS applies the same threshold in its data use terms, which bar showing any cell of 10 or fewer and any figure that reveals one (paraphrased).
- **Nothing from the company.** No records, documents, samples or system details are requested at any point.

## The anonymous decline-reason survey

The reason section uses one yes-or-no question per reason rather than a single tick-all list. Pew Research Center tested that choice: in a 2019 experiment it assigned half of 4,581 U.S. adults to a select-all list and half to separate yes-or-no questions, and its estimates for the experiences it asked about ran 8 percentage points higher on average with the forced-choice format. Pew reports that it generally avoids select-all lists as a result. The survey follows that lead and adds a third answer, not sure, so uncertainty is counted rather than forced.

Each item reads "Did this play a part in your decision?" Items appear in random order within each family, and the families in random order, so no reason always comes first.

| Code | Family | Did this play a part in your decision? |
|---|---|---|
| B1 | Fit | We did not think the company was large or established enough |
| B2 | Fit | Most of our records belong to our clients, not to us |
| B3 | Fit | Our records are scattered, incomplete or hard to describe |
| B4 | Rights | Customer or vendor contracts might not allow it |
| B5 | Rights | Concern for the privacy of staff or customers in the records |
| B6 | Rights | The data was already licensed or promised elsewhere |
| B7 | Timing | The moment was wrong: a sale, a raise, a migration or a crisis |
| B8 | Capacity | Nobody inside could own the work |
| B9 | Trust | Unease about AI developers using our records at all |
| B10 | Trust | Worry about how customers or staff would react |
| B11 | Trust | Unsure who SourceX is or whether to trust it |
| B12 | The offer as understood | The likely payment did not seem worth the effort |
| B13 | The offer as understood | Concern about agreeing to exclusivity |
| B14 | The offer as understood | Unclear what would happen to the records |
| B15 | The messenger | Felt pressed by the person who raised it |
| B16 | The messenger | Did not want the introducer to be paid for it |

## Stage, deciding reason and what would have changed the answer

Around the reason battery sit four short parts, each optional:

1. **Stage.** Which of S1 to S5 best describes when the founder said no, and who raised the idea, by role only.
2. **Deciding reason.** "Of the items you answered yes, which mattered most?", with "none of these" and a short text box.
3. **What would have changed it.** Nothing would; a later date; clearer information; a different person raising it; different terms; something else.
4. **Anything else.** One open box, quoted only with permission and edited to remove anything identifying.

## The sealed introducer's guess

Before the collector sends an invitation, the introducer records two things under the case code: the stage, and their best guess at the founder's deciding reason, using the same B codes. The guess goes to the collector, is sealed and is never shown back to the introducer.

After the close, the collector compares each guess with the founder's deciding reason, by item and by family. Only the program-wide agreement rate is published, under the same suppression rule, so no partner sees a case or is ranked. The comparison answers a practical question for every other study in this program: when partners report why a company said no, how often are they right?

## Analysis written down before fielding

Everything below is fixed before the window opens; anything added later is labeled exploratory.

1. **The flow.** Declines logged, invited, opened, completed, partial and never answered, each with its count.
2. **Reason items.** For each item, yes, no and not sure among those who answered it, with the base shown beside every share.
3. **Deciding reasons.** The distribution by family, with "none of these" reported as its own row.
4. **Stages.** Compared only where every stage in the comparison has 11 or more respondents.
5. **Guess agreement.** Item-level and family-level agreement between sealed guesses and founders' answers.
6. **Who is missing.** Respondents compared with nonrespondents on stage, size band and introducer role, the only facts known for both.
7. **No causal claims.** Results describe what respondents said; nothing is weighted to stand for founders who were never asked.

## Three predictions and the counts that would overturn them

The predictions are written down now so the results cannot quietly redefine them. They are hypotheses, not findings.

| Prediction | Overturned if |
|---|---|
| Timing or capacity (B7, B8) is chosen as the deciding reason more often than any part of the offer as understood (B12 to B14) | An item from B12 to B14 is the deciding reason at least as often as B7 and B8 together |
| Sealed guesses match the founder's deciding-reason family in fewer than half the cases | Family-level agreement reaches half or more |
| Founders who said no at S1 name the messenger or timing more often than those who said no at S3 to S5 | That share is no higher at S1 than at S3 to S5 |

## The results table, empty until the survey closes

**Method note.** The frame, definitions, items and rules above draw on SourceX's published fit criteria and authorization rule (the terms and the who-qualifies page, checked 2026-10-28), Pew's 2019 forced-choice experiment and CMS's cell suppression threshold. The survey measures founders' stated reasons and introducers' guesses. It does not measure SourceX's review decisions, demand for any company's records, completed licenses or rewards.

Every cell below stays empty until the window and the survey have closed with real answers.

| Reported item | Status |
|---|---|
| Flow from declines logged to completed surveys | Not collected |
| Yes, no and not sure for each of the sixteen items | Not collected |
| Deciding reason by family | Not collected |
| Stage comparison, where cells allow | Not collected |
| Sealed-guess agreement | Not collected |
| Respondent and nonrespondent profile | Not collected |

## Program facts the collector should have ready

Founders sometimes reply with a question, so the collector's script carries only published facts. The CEO decides whether the company talks to SourceX, and a decline is no black mark: SourceX's screening page says a company that does not fit one program may fit another later. Section 4 of the terms reads: "The referral reward is paid separately by SourceX and is never deducted from the supplier's agreed compensation." A company is screened on a full-time headcount that reached 50 at its peak with contractors left out, its own records and its rights to license them. 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. A company needs no prepared dataset to start, as [the dataset question page](/questions/can-a-company-discuss-a-partnership-before-preparing-a-dataset) explains.

## Where the decline survey can mislead

- **Tidy reasons.** A founder may name the most presentable reason rather than the real one; the not-sure option and the open box only partly help.
- **Courtesy to the introducer.** Even anonymous, some founders will spare a friend; B15 and B16 may be undercounted.
- **Small numbers.** Two quarters of declines may leave whole stages hidden by the suppression rule. Report the counts that exist and stop there.
- **Who opts in.** Introducers who volunteer to log every decline are probably the more careful ones.
- **Moving terms.** SourceX versions its terms; a change during the window is dated, and the cohort is split at it.
- **Demand can shift.** If AI developers come to need fewer real business records, the reasons founders give may matter less to anyone; the survey cannot test that.

## 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.

If your network will log its declines for the survey, start with companies whose CEO has already heard the idea: [register as a partner](/register), screen them against [who qualifies](/who-qualifies) and read the [program terms](/terms); further research designs are in the [guides hub](/guides).

This survey studies founders' decisions, not SourceX as a business; anyone weighing SourceX as an investment can ask for 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://www.pewresearch.org/methods/2019/05/09/when-online-survey-respondents-only-select-some-that-apply/
- https://www.cms.gov/files/document/lds-dua-terms-and-conditions.pdf
- https://refer.sourcex.si/terms
- https://refer.sourcex.si/who-qualifies
- https://sourcex.si/investors
