# Mature company data referral study: a design | SourceX

Canonical: https://refer.sourcex.si/guides/do-mature-businesses-respond-differently-to-ai-data-opportunities
Last updated: 2026-11-04

Nobody knows yet, and this page publishes no results. Two plausible stories point in opposite directions: older companies hold longer runs of their own records, while a Census Bureau working paper found AI use more common among younger firms once size and industry were controlled. The study below would settle the question for introductions made through partner networks. It fixes each company's age, leadership generation and record span before the CEO is asked, follows every answer, silence included, and compares bands only after the cohort closes.

## Two reasons maturity could matter, pulling in opposite directions

A research director who assumes "mature companies are the best introductions" is assuming two things at once: that maturity raises fit, and that mature CEOs say yes more often. The first follows partly from SourceX's screen. The second has never been measured.

| Statement | Standing | Basis |
|---|---|---|
| SourceX screens for roughly 50 or more full-time employees at peak and several years in operation, and some programs set higher thresholds | Program fact | Who-qualifies page |
| In Census survey data, firm-weighted AI use fell with firm age, and younger firms were more likely to use AI with size and industry controlled | Known, about AI use, not licensing | Census working paper CES-WP-24-16 (paraphrase) |
| Older companies agree to a licensing conversation more often, because they hold longer records and steadier operations | Hypothesis H1 | Unmeasured |
| Older companies agree less often, because they are further from AI work and slower to take up an unfamiliar idea | Hypothesis H2 | Unmeasured |
| Whether any company is accepted, licenses records or produces a reward | Conditional | SourceX review and the company's own signed agreement |

The Census finding concerns whether firms use AI in their operations. It says nothing about licensing records to AI developers, and the study does not treat it as if it did; it is the reason H2 deserves a fair test.

Neighboring pages answer other questions. The [age-and-ownership worksheet](/guides/how-to-find-mature-businesses-in-an-early-stage-fund-s-network) helps a fund find mature companies; the [stage-normalized scorecard](/guides/how-to-compare-referral-candidates-across-different-growth-stages) compares candidates; the [operating-history study](/guides/how-much-operating-history-sits-outside-venture-portfolios) counts how many mature companies sit outside a portfolio. This one asks what happens after the ask.

## How maturity is defined before anyone is asked

Maturity is recorded on the day the company enters the cohort, before its CEO hears anything, and is never edited after an answer arrives. Each variable has a source and a date.

| Variable | Bands | Recorded from |
|---|---|---|
| Company age | 5 to 9 years, 10 to 24 years, 25 years or more; younger companies logged separately | The year operations began, from the owner's knowledge or a public registry, with the source noted |
| Leadership generation | Founder is CEO; successor CEO from inside; successor CEO from outside or an owner group | The relationship owner |
| Record span | Years covered by the company's own records in its main system: under 5, 5 to 14, 15 or more, unknown | The relationship owner, from conversation with the CEO, never from files |
| Peak full-time headcount (covariate) | 50 to 199, 200 to 999, 1,000 or more | The relationship owner or a public filing |
| Relationship age (covariate) | Under 2 years, 2 to 9, 10 or more | The relationship owner |
| Path (covariate) | Owner asks the CEO directly; through one other person | The relationship owner |

**Why age needs a rule.** Registry dates lie when a company reincorporates, merges or is bought. The Census Bureau's Business Dynamics Statistics handle this by giving a firm an initial age from the age of its oldest establishment at birth and then adding the years since, so mergers and divestitures do not make firm age jump (paraphrase). The study follows the same idea: age runs from when the operating business began, and a company formed by an acquisition takes the age of the oldest business it absorbed. The [age traps](/guides/how-to-find-mature-businesses-in-an-early-stage-fund-s-network) on the worksheet page show where registry dates mislead.

## The maturity-based response study

| Protocol element | Rule |
|---|---|
| Frame | Every company a participating network plans to ask during a six-month entry window, listed before the ask |
| Unit | The company, grouped by parent; one row however many owners know it |
| Inclusion | The owner believes the company passes the published screen and the CEO has not been asked before by this network |
| Exposure | Company age band, fixed at entry; leadership generation as a second exposure |
| Primary outcome | The CEO's answer within 60 days of a dated ask: agreed to a conversation, asked to wait for a named date, declined, or no answer |
| Secondary outcomes | Days to answer, with unanswered cases censored at day 60; the CEO's first question; whether a verified application follows, from the partner's own portal status |
| Consent | Each owner agrees before rows are compiled; no company is contacted for the study; nobody handles or reviews company records |
| Anonymization | Companies and owners are coded; outputs show bands only; any cell under five is merged with its neighbor |
| Pre-registration | This protocol is dated before the window opens; any later change is logged with its date and reason |

**The first-question codebook.** When a CEO replies, two coders who cannot see the age band classify the first question asked. The codes are fixed now: effort (who at the company would do the work), people in the records (customers or staff), money (what the company would receive), control (who decides scope and can stop it), exposure (who else would know), and none (the CEO answered without a question). Agreement between coders is reported before any comparison by band. The codebook asks what a CEO wants to know first, which is a different question from why a message went unanswered; that one belongs to the [unanswered-offer study](/guides/why-business-owners-ignore-referral-offers).

**What a counted company has to show.** The cohort holds companies the owner believes pass SourceX's screen: 50 or more full-time employees at peak with contractors left out, records the company created itself, the rights to license them, and a CEO who decides whether to talk. On location, SourceX publishes this sentence: 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. Relevance to a current program is decided only in SourceX's review.

## Every silence and refusal stays in the count

| Disposition | Where it is counted | Why it matters |
|---|---|---|
| Owner chose not to ask | Reported by age band, outside the outcome denominator | If owners skip young or old companies, the cohort is selected before any CEO answers |
| Agreed to a conversation | Outcome denominator | The primary event |
| Asked to wait for a named date | Outcome denominator, its own row | Neither a yes nor a no; merging it with either would bias the bands |
| Declined | Outcome denominator | Kept in full |
| No answer by day 60 | Outcome denominator, read two ways | Once as a no, once excluded, and both shares printed |
| CEO had already been in touch with SourceX | Reported separately | A duplicate under terms section 3, whatever the band |

Credit, review and payment play no part in the primary outcome. A company that agrees and later proves a poor fit still counts as a CEO who agreed, because the study measures response to the opportunity, not SourceX's decision.

## The comparison rules, set before the first ask

**Primary comparison.** The share of asked CEOs who agreed, in the oldest band against the youngest, with a 95% interval for each band and for the difference. The secondary comparison does the same for leadership generation.

**Adjusted comparison.** A model of agreement on age band with relationship age, path, peak headcount band, sector band and network as covariates. Relationship age matters most: owners tend to have known older companies longer, and a long relationship may raise agreement on its own.

**Time to answer.** Days from ask to answer by band, with unanswered cases censored at day 60, so slow answers are not mistaken for refusals.

**Sample arithmetic (Illustrative).** To estimate one band's agreement share within plus or minus 10 points at 95% confidence when the share is near one half, the standard formula gives about 96 asked CEOs per band (1.96 squared, times 0.5 times 0.5, divided by 0.1 squared). Three bands need about 290. The one half is an assumption, not a finding; a smaller cohort reports counts and intervals and draws no comparison.

**Descriptive, not causal.** Maturity cannot be assigned, so any difference is an association. The report says so in its first paragraph.

## The age-band results table, empty until the cohort closes

**Results: not published.** No cohort has been opened, and no number on this page is a finding. The report will carry the protocol date, the networks taking part, the entry window, dispositions by band, the table below with intervals, the adjusted estimates, coder agreement and every dated departure from this plan.

| Age band | Asked | Agreed | Asked to wait | Declined | No answer |
|---|---|---|---|---|---|
| 5 to 9 years | Not published | Not published | Not published | Not published | Not published |
| 10 to 24 years | Not published | Not published | Not published | Not published | Not published |
| 25 years or more | Not published | Not published | Not published | Not published | Not published |

## Results that would overturn either maturity hypothesis

**Counterexamples the report must print.** If older companies agree more often but the gap vanishes once relationship age is in the model, maturity was standing in for familiarity. If the oldest band agrees less often but asks to wait more often, the difference is timing, not refusal. If results differ sharply by network, a pooled figure would mislead and the report shows networks separately.

- **Survivorship.** Every company in the oldest band survived long enough to get there; that alone may make its CEOs different.
- **Selection by owners.** Owners decide whom to ask. The "chose not to ask" row shows it but cannot remove it.
- **Age errors.** A registry date can understate an operating history by decades, and the rule above reduces that error without removing it.
- **One period.** Attitudes to AI change quickly; a cohort read in one six-month window describes that window.
- **Borrowed evidence.** The Census paper measures AI use, a different behavior from agreeing to discuss a data license.

**Methods note.** The study measures CEO answers to dated asks made through participating networks, against maturity variables fixed before each ask, from owner logs, public registries and each partner's own portal statuses. It does not measure record quality, demand for any company's records, payouts or rewards, and it contacts no company for research purposes.

## 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.

Until the cohort reports, treat maturity as a fit fact rather than a prediction of the CEO's answer: [register](/register) as a partner, check each company against [who qualifies](/who-qualifies), read the [program terms](/terms) and browse the other study designs in the [guides library](/guides).

Answering this question is network research; evaluating SourceX as a company is a different decision, and investors who want to make it 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://refer.sourcex.si/terms
- https://refer.sourcex.si/who-qualifies
- https://www.census.gov/programs-surveys/bds/documentation/methodology.html
- https://www.census.gov/hfp/btos/downloads/CES-WP-24-16.pdf
- https://sourcex.si/investors
