# Data partnership skeptic interviews: a pilot guide | SourceX

Canonical: https://refer.sourcex.si/guides/how-to-invite-skeptics-into-a-data-network-pilot
Last updated: 2026-11-05

Invite skeptics to design and judge a data network pilot, not to be converted by it. Give each one a named seat with a single job: a doubtful founder edits the ask, counsel draws the line on what may be asked, an investment committee member writes the stop rule, and a data-market doubter writes the prediction that would sink the thesis. Run a premortem with them before the first ask, interview them without ever requesting an introduction, and let their objections become the pilot's rules.

## Why skeptics belong inside the pilot, not on its mailing list

A platform team that has decided its network holds data-licensing opportunities usually treats skeptics as an audience: send them the memo, answer their objections, move on. That wastes the most useful people in the room. A skeptic has already done the work of imagining how the idea fails, and a pilot exists to find out whether it does.

Translate the category thesis into a company question and the skeptic's role becomes obvious. The thesis might be "established companies in our network hold records that could be licensed directly." The company question is narrower: will a CEO who has heard the strongest objections still want a conversation? A pilot designed only by enthusiasts never tests that question, because the objections were never put to anyone.

Some objections rest on evidence worth taking seriously. Longpre and colleagues audited about 14,000 web domains behind three large training corpora and found that, in a single year (2023 to 2024), restrictions on AI crawling rose sharply: roughly 5% of all tokens in the C4 corpus, and over 28% of its most actively maintained sources, became fully restricted. That study is about public websites, not company records, so it does not show how any company feels about licensing its own material. It does show that data owners care who uses their work for AI and on what terms, which is exactly the doubt a skeptical founder will raise, and exactly why a direct license the company controls, or a decision not to license at all, has to be the company's call.

## The skeptic participation guide

Four seats cover the objections a data network pilot most often meets. Each seat has one job, one bounded veto and a clear list of what its holder is never asked to do.

| Seat | Typical holder | The objection they bring | Their job in the pilot | Their veto covers | Never asked to |
|---|---|---|---|---|---|
| The doubtful founder | A portfolio CEO who has said their data is not for sale | Investors pitching vendors to founders erodes trust | Edit the ask note and the reward disclosure before anything is sent | The wording of the ask | Introduce their own company or name others |
| The boundary keeper | Fund counsel or a privacy lead | Partners will end up touching confidential records | Write the list of permitted questions and review the first three notes | The question list | Give a legal view on any company's rights |
| The committee skeptic | An investment committee member or LP adviser | The platform team will be seen endorsing a vendor | Write the stop rule and the line between referral work and investment views | The stop rule | Approve SourceX as an investment |
| The data doubter | A technical partner who expects synthetic data to win | Demand for real business records will not last | Write the prediction that, if it comes true, ends the thesis | Nothing; their job is to be scored | Change their public view |

**What every seat receives.** The pilot's counts at the end, including declines and non-replies; the log showing what was done with each of their objections; and a named date when the pilot is judged. Without those three, a seat is decoration and the skeptic will notice.

**What no seat sees.** Company names, board materials, anything the fund learned under confidentiality, and anything resembling a record. Skeptics shape rules and wording. They do not review companies.

**Method note.** The four seats map onto the points where a referral introduction needs a decision, as version 2026-10-22.2, since replaced by version 2026-11-04 of the program terms and the who-qualifies page lay them out (both checked 2026-10-26): what may be asked, who authorizes, what is credited and when the work stops. No survey of skeptics' views on data licensing has been published, so the objections in the table are the ones the program's own rules anticipate, not measured frequencies.

## Run a premortem before the first ask

A premortem gives skeptics a structured way to be right early. Gary Klein, writing in Harvard Business Review, describes it as the hypothetical opposite of a postmortem: at the start of a project, the leader announces that it has failed, and everyone independently writes down every reason they can think of. The article cites 1989 research finding that this kind of prospective hindsight increased people's ability to correctly identify reasons for future outcomes by 30%.

A 45-minute version for a data network pilot:

1. **Set the scene (2 minutes).** "It is six weeks from today. The pilot failed. Nobody wants to run it again."
2. **Write alone (8 minutes).** Every participant, enthusiasts included, lists reasons. Silent writing keeps the most senior voice from setting the list.
3. **Read one reason at a time (15 minutes).** Go round the table until every reason is on the board, without debate.
4. **Convert each reason (15 minutes).** Every reason becomes a check that runs before an ask, a stop rule that ends the pilot, or a note that it cannot be tested in six weeks.
5. **Assign an owner (5 minutes).** Each check and stop rule gets one person's name.

Illustrative. One possible board after step 4:

| Premortem reason | Converted into | Owner |
|---|---|---|
| Founders felt pitched by their own investor | Stop rule: if two of the first five CEOs say the ask felt like pressure, pause and rewrite | Doubtful founder |
| A partner asked a company for a sample export | Boundary rule: questions come only from counsel's list; any request for files ends that partner's part in the pilot | Boundary keeper |
| Most companies turned out to know SourceX already | Check: ask each CEO about earlier contact with SourceX before anything is submitted | Platform lead |
| LPs read the pilot as a fund endorsement | Separation rule: referral updates never mention SourceX as an investment | Committee skeptic |
| Demand for the record types collapsed | Logged as untestable in six weeks; the data doubter's prediction tracks it instead | Data doubter |

## The skeptic interview guide

Before the premortem, spend twenty minutes alone with each skeptic. The interview finds objections that people will not voice in a group, and it is the only place where a skeptic can name a founder who should not be asked.

1. What would have to be true for this pilot to be a bad idea for our founders?
2. Which kinds of company should we not approach at all, and why?
3. What should the note to a CEO never say?
4. After six weeks, what result would convince you to stop, and what result would convince you to continue?
5. What would you need to see before you trusted the counts we report?
6. Who else should hold a seat?

**Rules for the interviewer.** Ask permission before taking notes. Code each answer into the objection log the same day, using the skeptic's words. Never ask the skeptic for an introduction, an endorsement or information about a specific company's records. Quotes stay inside the pilot; nothing a skeptic says is published or attributed without their written agreement. A skeptic can leave the pilot at any time, and the log records that they left, not why.

## What relationship and ownership evidence the pilot still needs

Skeptics shape the rules. Each company in the pilot still needs its own evidence before an ask, gathered at metadata level only (what records exist, roughly how much, how many years, which system and who owns it).

- **The relationship owner,** and whether they agreed to make the ask.
- **The CEO's answer.** Authorizing a licensing discussion and signing for the company both fall to the CEO, and SourceX usually does not treat a company as a fit when the decision maker never knew about the introduction.
- **The published fit facts:** a peak headcount of 50 or more full-time employees, not counting contractors; an operating history the company can document; the rights to license; and a qualifying location. 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.
- **Novelty.** Any company with an earlier application, relationship, lead, account or email correspondence with SourceX is a duplicate and is not credited, which is why the premortem check above asks first.

Records that belong to a company's clients are usually not a fit, and nothing about any company's records passes through the pilot. The [shortlist worksheet](/resources/a-business-network-referral-shortlist-worksheet) is a good place to hold these facts per company.

## A six-week pilot the skeptics can end

Keep the pilot small enough that its skeptics can read every row: six to eight companies over six weeks.

- **Week 0.** Interviews, premortem, the permitted-question list, the edited note and the stop rules, all dated. The data doubter's prediction is written and sealed.
- **Weeks 1 to 5.** Relationship owners make the asks. Every outcome is logged: agreed, declined, no reply, duplicate, failed a fit fact.
- **Week 6.** Skeptics and enthusiasts meet to score the pilot against what was written in week 0.

What would change the thesis: if CEOs who saw a plain disclosure and a clear choice still decline at the rate the doubtful founder predicted, the thesis that CEOs want this conversation is wrong for this network. If a stop rule fires, the pilot stops, whatever the enthusiasts feel. If neither happens, the result is a direction, not a rate; six to eight companies cannot support a percentage worth quoting. For a fuller pre-registered design, see [the ten-introduction experiment](/guides/the-ten-introduction-experiment-for-an-ai-data-network-thesis), and for writing to a doubtful operator directly, [the referral message for a skeptical operator](/guides/the-referral-message-for-a-skeptical-operator).

None of the pilot's asks can be submitted until the person submitting has signed the SourceX Channel Partner Agreement. A pilot member whose copy carries the older title (Channel Partner Non-Circumvention and Non-Disclosure Agreement) is still bound by it exactly as signed. Credit is settled when an introduced company submits a verified application, and it goes to one partner only: the one whose valid introduction of that new company came first inside the 90-day attribution window.

## Where inviting skeptics can backfire

- **Token seats.** Skeptics see quickly when objections are logged and never acted on. The objection log has to show a conversion or a reason for every entry.
- **Veto creep.** A seat that grows its veto beyond its scope turns the pilot into a standing argument. Write each veto's scope down in week 0.
- **Confidentiality.** A skeptic from a portfolio company must never learn which other companies were approached. Share counts only.
- **Stretching the evidence.** The consent research covers public web data. Using it to claim how companies feel about licensing their own records overstates it.
- **The skeptics may be right.** Sometimes the premortem shows the pilot should not run. That is a result, and a cheap one. The [disconfirming-evidence checklist](/guides/what-would-disprove-your-network-s-data-opportunity) helps decide whether a finding ends one company's row or the whole thesis.

## 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 skeptics have edited the ask and signed off the stop rules, each submitter should [register as a partner](/register), every pilot company should be screened against [the qualification page](/who-qualifies), and the [program terms](/terms) should be read before the first relationship owner reaches out. The [platform team overview](/for/venture-platform-teams) and the [guides library](/guides) cover the rest of the program.

Running a pilot with skeptics is a referral decision; a skeptic who wants to examine SourceX as a business to invest in is making a separate one, and 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://hbr.org/2007/09/performing-a-project-premortem
- https://arxiv.org/abs/2407.14933
- https://sourcex.si/investors
