# Industry council aggregate findings on AI data | SourceX

Canonical: https://refer.sourcex.si/guides/when-an-industry-council-wants-to-publish-aggregate-findings
Last updated: 2026-11-04

It can publish, as long as the findings stay the council's own account of what members think and have experienced. Members opt in knowing every use of their answers; the report shows only totals that no reader could trace to a company; it never estimates how many members might qualify for SourceX or anything else; no respondent becomes a referral lead; and wherever SourceX is named, the council's own relationship is disclosed beside it. Pilot the questionnaire with a few volunteers who may decline any part of it.

This page is for directors of industry councils, trade associations and business networks that have joined, or are weighing, SourceX's referral program and want to publish what their members think about AI data licensing. It covers the publication: who owns it, what it may ask and say, and when not to publish at all.

## What the council owns, and what it may not borrow

The findings are the council's research about its own members. Present them that way: nothing in the report should read as SourceX's findings, as SourceX endorsing the council, or as the council endorsing SourceX. Members' answers stay with the council and never travel to SourceX or to any partner, and no member gains preferential treatment by answering.

If the council has joined the program, SourceX's unsigned sample of the Channel Partner Agreement bears on the report twice. Its parties block names the Partner as "an individual, or entity type and jurisdiction", so a council can be the signing partner. And section 4.5 says: "Neither Party may publicly identify the other as a partner or disclose non-public deal details without written authorization, except as legally required." A report that calls SourceX the council's partner therefore needs SourceX's written authorization first, and no report carries deal details.

The simpler route is often the better one: a report on members' views about AI data licensing in general does not need to name any provider.

## Questions to ask, and three to leave out

| Topic | Ask it? | Why |
|---|---|---|
| Whether members have heard of companies licensing business records for AI training | Yes | Awareness is a view, and the council's to describe |
| Concerns: client confidentiality, staff privacy, reputation, effort | Yes | It tells the council what members need explained |
| What a member would want to know before any licensing conversation | Yes | It shapes the council's education, not anyone's pipeline |
| Whether any data buyer has approached the member, yes or no, without names | Yes | Experience, kept anonymous |
| Respondent's role: CEO or owner, other executive, other | Yes | Views differ by seat; the CEO decides for the company, and the report should say whose view each total holds |
| Company size, years of records, systems, data volumes | No | It turns a views survey into a fit estimate and a record inventory nobody asked the council to build |
| Offers, prices or terms a member has received | No | Commercial terms are each company's own; ask the council's counsel before any question that touches them |
| "Would you like an introduction to SourceX?" | No | It converts respondents into leads; interest has its own route, set out below |

## The member-consented aggregate research plan

Written and dated before the invitation goes out, then published with the findings.

| Part | What the council writes down | Illustrative entry |
|---|---|---|
| 1. Purpose | Why the council is asking and what it will publish | To describe members' views on AI data licensing in the annual member report |
| 2. Sponsor and who conducted it | AAPOR's code asks researchers to "Name the sponsor of the research and the party(ies) who conducted it", and to disclose a different funder if there is one | Sponsored and run by the council's research committee; no outside funding |
| 3. Frame and coverage | The code also asks for "A description of the sampling frame(s) and its coverage of the target population", including any segment not covered | All 412 member firms in good standing on 1 November; associate members not covered |
| 4. Invitation and consent | Purpose, every use, what is never asked, who sees member-level answers, and the date withdrawal closes | One invitation to each member's primary contact, who answers in their own name |
| 5. Variables | Views, concerns, information needs and respondent role only | No size, records, systems or commercial terms |
| 6. Analysis | Counts with the number answering each question shown beside them | No percentages on fewer than 20 answers; free text summarized by theme |
| 7. Reporting floor | The smallest group any table may show | No cell under 10; no sector-by-region table if any cell would fall below 10 |
| 8. Lock date | When answers are frozen and aggregates fixed | Withdrawal open until the lock date, stated in the invitation |
| 9. Firewall | Who may see member-level answers, and what they may never be used for | The research committee only; staff who hold the SourceX relationship see totals; answers never used to contact a member about SourceX |
| 10. Disclosure | What the report says about any referral relationship | See the SourceX paragraph below |
| 11. Retention | When member-level answers are deleted | A set number of days after publication, under the council's own policy |

AAPOR's disclosure standards, in paraphrase, apply to surveys whose results are released to the public, which is exactly the council's case. The floor in row 7 is a minimum; before release, run [the anonymization review for network research](/guides/how-to-publish-network-research-without-exposing-companies), which tests every table and quote for a route back to a company.

## A small opt-in pilot, with a way out and a way to object

Before the full survey, ask for six to ten volunteer members to read the invitation and answer the questionnaire.

- **Decline route.** Any volunteer may skip any question or leave the pilot with one reply, and hears nothing more about it.
- **Feedback route.** Comments go to a named staff member who does not hold the SourceX relationship, so criticism of the partnership has somewhere neutral to land.
- **What the pilot counts.** Questions skipped; misreadings, such as "is this SourceX's survey?", "will this decide whether we qualify?" or "will SourceX see my answers?"; and any objection to the topic itself.
- **Go or rewrite.** Field the survey only if no volunteer believes SourceX will see their answers and no question is skipped by more than half of them. Otherwise rewrite and pilot again.

This pilot tests the instrument. Whether the topic helps members at all is a different question, tested in [the five-member association pilot](/guides/a-small-association-can-run-a-useful-pilot-with-five-members).

## Declines, partial answers and unwelcome results

Report the number invited, the number who answered fully, partly or not at all, and any withdrawals, with the response-rate formula stated. Decide before fielding that the council will publish whatever the results show, or nothing; choosing after seeing them turns research into promotion. A finding that most members are uninterested or worried gets the same space as any finding the partnership would welcome.

Keep the claims descriptive. An opt-in survey of one council's members describes those who answered, so write "of the 140 members who answered," never "the industry." For measuring the partnership's value without counting every member, see [the opt-in cohort dashboard](/guides/how-to-measure-a-partner-benefit-without-counting-every-member).

## What the report may say about SourceX

If SourceX appears at all, give it one dated paragraph built from its published pages, and set the council's disclosure inside it. An Illustrative version:

"SourceX licenses business records straight from the companies that created them, for training and evaluating AI. Each company's CEO decides whether to explore it, and fit is decided in SourceX's own review. 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. The council is a SourceX referral partner and could receive a reward from SourceX if a member it introduces completes a paid licensing deal; SourceX's terms say that reward 'is paid separately by SourceX and is never deducted from the supplier's agreed compensation.' Nothing in these findings indicates whether any member would qualify."

Use the partner sentence only with SourceX's written authorization, as section 4.5 of the sample agreement requires. Never add a count of members who "could fit." SourceX's screening page tells promoters never to list companies as eligible, and a total is a list with the names removed.

## The board's publication decision, fixed before fielding

| Condition on the lock date | Decision |
|---|---|
| At least 30 answers, the reporting floor met, pilot misreadings fixed | Publish with the plan attached |
| Between 10 and 29 answers | Publish overall counts only, no breakdowns |
| Fewer than 10 answers, or any table that could point to a member | Do not publish; tell respondents why |
| A misreading the pilot missed surfaces after fielding | Publish with the problem stated beside the affected question |

Explaining the outcome to members afterward fits the council's existing feedback cycle; [the member feedback loop](/guides/the-member-feedback-loop-that-improves-referral-fit) shows one that records nothing about any member's company.

## Where a member's own interest goes

The survey never routes anyone. A member CEO who wants to explore SourceX contacts the council's named person or applies to SourceX directly, and their answers stay where they were. The council collects no member's records, systems or headcount for SourceX: each introduction concerns one company whose CEO agreed to it, submitted by whoever signed the Channel Partner Agreement, whether the council itself or a person. SourceX's published screen looks for at least 50 full-time staff at a company's largest point, contractors not counted, years of records the company produced, clear rights to license those records and a CEO keen to talk; rights and relevance are settled in SourceX's review. Credit is decided from dated records under the terms, and one company is credited to one partner at most. How the council keeps its list private while offering the program is covered in [offering a referral program without selling your member list](/guides/how-to-offer-a-referral-program-without-selling-your-member-list), and how to explain why it chose this program in [the selection criteria disclosure sheet](/guides/when-members-ask-why-the-association-chose-this-program).

## Where a council's findings can mislead

- **Who answers.** Members interested in the topic answer more readily, so interest will look higher than it is.
- **The council's own stake.** A partner council has a reason to like certain results; publishing the plan first is the check.
- **Partner status colors answers.** Members who know the council is a SourceX partner may answer more warily or more politely.
- **Small industries identify.** In a sector with six large firms, even a "large members" row can point to one.
- **The topic moves.** Views gathered this quarter may read differently a year on; date the report prominently.

## 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 the council will also introduce members whose CEOs ask for it, [register](/register) the signing party, keep [who qualifies](/who-qualifies) beside the SourceX paragraph above and read the [program terms](/terms) before the report goes to print.

## Sources

- https://refer.sourcex.si/terms
- https://refer.sourcex.si/who-qualifies
- https://aapor.org/wp-content/uploads/2022/12/AAPOR-2020-Code_FINAL_APPROVED.pdf
- https://sourcex.si/documents/partner/SourceX_Channel_Partner_Agreement_Sample.pdf
