A community lead at a venture fund can easily run the opposite event: a session that counts attendees as prospects and follows up with everyone. That inflates a referral funnel and teaches founders to avoid the next invitation. The runbook below treats the room as a source of evidence about what founders do not yet know, and its output is a short set of public answers, not a list of companies.
What the roundtable is for: questions, not prospects
The audience question is narrow: what do founders in this network actually want to know about licensing their company's own records for AI training and evaluation, and which of those questions deserve a careful public answer? Pick the format for that question, not for reach.
| Format | Good at | Weak at |
|---|---|---|
| Roundtable of six to twelve founders | Surfacing questions people would not type into a form, and hearing which ones several founders share | Representing the whole network; anything that needs a company's own facts |
| Webinar | Explaining published rules to many people at once | Learning what the audience did not already know to ask |
| One-to-one office hour | One company's situation, privately | Seeing which questions recur across companies |
| Written memo | A stable answer people can forward | Discovering new questions |
The roundtable size borrows from evaluation practice. CDC's evaluation brief on focus groups describes a group interview of roughly six to twelve people who share similar characteristics or interests, and treats the choice of participants and the discussion guide as the two planning decisions that matter most (our paraphrase). Here, "similar" means founders whose companies have been operating for some years, because the published screening profile looks for companies with 50 or more full-time employees at peak and years of their own records. Funding plays no part in that profile, so a portfolio company and a non-portfolio founder from the wider network belong at the same table. The live guide on choosing between research events and direct introductions compares the channels over a quarter.
The research-first roundtable runbook
Seven steps, each with an owner and an output. The hypotheses come first so the session can prove you wrong.
| When | Step | Owner | Output |
|---|---|---|---|
| 21 days before | Write five hypotheses about what founders will ask, and what you would publish if each held | Community lead | A dated hypothesis sheet, kept unseen by attendees |
| 14 days before | Send the invitation | Community lead | States that attending is optional and commits no company, that nothing about a company is collected, whether the host is a registered SourceX referral partner that could be paid, and that no one's absence is noted |
| 7 days before | Write the discussion guide | Community lead and one founder who will not attend | Neutral prompts such as "What would you need to know before taking this to your CEO?", never "Why haven't you licensed your data yet?" |
| Session, 75 minutes | Run five rounds (below) | Facilitator | Anonymous question cards and dot-vote counts |
| 2 days after | Build the question log | Facilitator | Clusters, card counts and votes, with no names |
| 14 days after | Publish the digest | Community lead | Answers to every question that passed the test, sent to all invitees, attendees or not |
| 30 days after | Score the hypotheses and the measures | Community lead | Which hypotheses held, and the counts in the measurement section |
The five rounds. House rules read aloud (10 minutes): no company names or figures on cards, no recording, nothing collected for anyone. Silent writing (15): each founder writes up to three questions, one per card. Clustering (25): cards go on a wall, the group merges duplicates, and each founder places three dots on the clusters they most want answered. Next-step round (15): for the top clusters, "If you had the answer, what would you do differently?" Close (10): when the digest arrives, and how to ask anything privately afterwards.
The worth-answering test
A question earns a public answer only when it passes all four parts.
- Recurrence. At least three of the founders in the room raised it on a card.
- Decision. In the next-step round, someone named something they would do differently with the answer.
- Source. A named source can answer it: SourceX's published pages, the program terms, or a profession the founder can consult.
- No private facts. Answering it needs no company's own records, contracts or figures.
A question that fails the last part goes to a private route: the founder's own counsel, a self-check the founder runs alone, or the company applying to SourceX itself. A question that fails the third part gets a one-line public answer saying nobody has published it. A question that fails the second is logged and not built into content. One that fails only the first is answered one-to-one if the founder asks again.
An Illustrative question log. Ten founders, eight clusters; invented, and no outcome implied for any company.
| Question cluster | Founders who raised it | Changes a next step | Verdict | Answer or route |
|---|---|---|---|---|
| "What would our records be worth?" | 8 | Yes | Say nobody publishes it | SourceX publishes no typical payouts; pricing stays indicative until a Dataset Deal Confirmation is signed. The live page on what company data is worth explains why |
| "Who at our company would have to say yes?" | 7 | Yes | Answer for everyone | The CEO authorizes a licensing discussion; while a sale is pending, the company's own CEO, and after closing, the acquirer's CEO |
| "Would licensing mean handing over our customers' data?" | 6 | Yes | Answer for everyone | Material that belongs to clients is usually not a fit; the company's own operating records are what SourceX screens |
| "Do our customer contracts allow it?" | 5 | Yes | Private route | The company's counsel, then SourceX's rights review after the company applies |
| "How many years of records do we even keep?" | 4 | Yes | Private route | A self-check the founder runs alone, such as the company fit checker |
| "Would our investors see the data?" | 3 | Yes | Answer for everyone | Not through this program: section 4.2 of the partner agreement withholds any right for a partner to pass a company's records along, and applying needs no files |
| "Is synthetic data making this moot?" | 2 | No | Note only | Logged for a later reading list |
| "Would licensing affect our next round?" | 1 | No | Note only | Not a question for this room |
Method note. The four-part test and the log were built for this page from the program terms (Version 2026-10-22.2, since replaced by version 2026-11-04), SourceX's screening and process pages and the unsigned partner agreement sample, all read on 2026-10-27, with the group size taken from CDC's focus group brief. They measure what founders ask and whether a reliable public answer exists. They do not measure whether any company fits, what its records would be worth or how SourceX would decide, which only SourceX's review can establish.
The hypothesis sheet gets scored against this log. If the lead expected price to dominate and it did, that is a finding; if nobody asked who decides, the digest still answers it, because the rule is short and settled.
Optional seats and what never leaves the room
Optional has to be visible in the design, not just in the invitation.
- No sign-in sheet beyond a name. No company size, systems, record counts or "areas of interest" fields.
- No attribution. The host agrees in the invitation never to tie a question to a person or a company, in the digest or anywhere else.
- Redirect on the spot. When a founder starts describing their own systems, the facilitator thanks them and asks for the question underneath.
- Disclose the host's interest. If the host or the fund is a registered SourceX partner and could be paid for an introduction, say so in the invitation and again in the house rules.
- No promises from the chair. The partner agreement bars a partner from telling a company that an estimate, interest, a signature or an upload secures a purchase or a reward. The chair answers only from published pages.
- Private escalation, later. After the digest, a founder can reply to the host alone. The host's moves are limited: point to the self-check, or, when the company's CEO asks in writing, make one introduction, after which the company applies itself. Before submitting anything, the host must have signed the SourceX Channel Partner Agreement.
Founders sit with their investor in this room, and a question asked in front of a backer is not always a free question. That is a further reason the digest answers questions, not people. The live backlog for founder questions a platform team cannot answer handles the private questions that arrive afterwards.
Measuring requested follow-up and what founders understood
Count with denominators, and count requests rather than outreach.
| Measure | Numerator | Denominator |
|---|---|---|
| Attendance | Founders who came | Founders invited |
| Question yield | Question cards written | Founders who came |
| Public answers | Clusters answered in the digest within 14 days | Clusters that passed the test |
| Private follow-up | Founders who wrote to the host afterwards | Founders who came |
| Introductions requested | CEOs who asked in writing to be introduced | Founders who came; counted, never targeted |
| Understanding | Anonymous before-and-after cards answered in line with the published rules on two items: who decides, and whether customer data is involved | Cards returned |
SourceX has not published results from founder events, and no benchmark for these figures exists yet; the first roundtable sets the network's own baseline.
Where a research-first roundtable can mislead
- Small rooms. Ten founders are not the network. Treat the log as a set of questions to test in the next session, not as a survey result.
- Who shows up. The curious attend and the skeptical stay away, so the log under-represents the questions behind a quiet no.
- The host's position. An investor who could be paid for introductions shapes what founders ask aloud. Disclosure helps; it does not remove the effect.
- Rights. The most common answer to "can we license this?" depends on customer contracts that only the company's counsel and SourceX's review can read.
- Price. The most-asked question may have no public answer at all, and a digest that guesses one would mislead every reader.
- Fit moves. Each SourceX program applies its own criteria, so an answer about fit is dated and can change.
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 a founder writes after the digest and the company's CEO asks to be introduced, register as a partner, compare what the CEO has told you with who qualifies, and keep the program terms open while you write the single introduction. The overview for venture platform teams, the live roundtable ideas page and the guides library cover adjacent formats.
Hosting founders to learn what they ask is community work; judging SourceX as a potential investment is a separate question, and anyone pursuing it can ask for materials through the SourceX investor overview.
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.