# Referral follow-up timing test: a design | SourceX

Canonical: https://refer.sourcex.si/guides/a-follow-up-timing-test-that-respects-founder-preferences
Last updated: 2026-11-04

Test timing only inside the window each founder gives you. When a founder agrees to hear about SourceX, ask once when a single check-in would be welcome and which weeks to avoid. Founders who name a date get that date. Only founders who leave the timing to you are assigned, from a list drawn before anyone enrolls, to an early or a late check-in inside their own window. Read every company on one clock, day 120 after the yes, so the late group is never judged before its time is up. No results exist yet.

The test never changes how many messages anyone receives. Every founder in it gets one check-in at most, and none at all if they said so.

## Timing is its own question, separate from how many reminders

A Wave 8 design already asks [whether more follow-ups improve referral quality](/guides/do-more-follow-ups-improve-referral-quality), holding timing roughly fixed and varying the count. This design holds the count at one and varies only the day. The two cannot be combined in one cohort without making both unreadable.

The hypothesis is written as the dull answer: among founders who leave timing to the partner, an early check-in and a late one produce the same share of qualified next steps by day 120, with no difference in stop requests. Anything else has to beat that.

There is a pull toward early check-ins that has nothing to do with founders. Under section 2 of the terms, credit goes to the first valid introduction only when the company submits a verified application inside the 90-day window, and the window does not stretch. A partner thinking about credit wants the check-in early. This design ignores credit entirely, and the reason is simple: the application is the company's own act, on the company's own schedule.

| Term | Meaning in this design |
|---|---|
| Founder's yes | The written message in which the company's CEO agrees to hear about SourceX; it starts the clock at day 0 |
| Window | The earliest and latest day the founder said a check-in would be welcome |
| Avoid weeks | Weeks the founder asked you to stay away from, recorded in their words, with no reason requested |
| Assigned founder | A founder who left the timing open and was placed in arm E or arm L by the advance list |
| Qualified next step | The company's own verified application inside the 90-day window, still open or further along when its dashboard stage is read on day 120 |

## The window question, asked once with the thank-you

The question goes in the message that thanks the founder for the yes, and the answer is kept word for word.

| Element | Wording or rule (Illustrative) |
|---|---|
| The question | "If nothing has moved on your side, when would one check-in from me be welcome, and are there weeks I should stay clear of?" |
| A named day | That day is used, give or take a working day; the row is observed, never assigned |
| A range | The range becomes the window, minus any avoid weeks; the founder can be assigned |
| "Whenever suits you" or no answer | A default window of day 7 to day 42, a setting fixed before enrollment; the founder can be assigned |
| "Please don't check in" | No check-in, ever; the row is observed |
| Study notice | "I note the dates of my check-ins to learn how to time them better. Say the word and your company's dates stay out of it." |
| Never asked or recorded | Why a week is busy, what is happening at the company, any figure from its records |

A founder who opts out of the study still gets exactly the check-in they asked for; only their dates leave the dataset. A founder who asks for more time with a named condition is not in this test at all: the [requested-delay follow-up](/guides/a-referral-follow-up-after-a-company-has-asked-for-more-time) governs that row.

## Early third or late third: the consent-bounded timing experiment

For every assigned founder, take the window, remove the avoid weeks and split the days that remain into thirds. Arm E's check-in lands at the middle of the first third; arm L's at the middle of the last third, moved to the nearest working day. The middle third goes unused, which keeps the two arms clearly apart.

| Group | Who sets the day | Check-ins | Role in the analysis |
|---|---|---|---|
| Named a day | The founder | One, on that day | Described, never compared |
| Asked for none | The founder | Zero | Described, never compared |
| Arm E | The advance list, inside the founder's window | One, early third | Compared with arm L |
| Arm L | The advance list, inside the founder's window | One, late third | Compared with arm E |

The list is drawn before the first founder enrolls, in blocks for each partner and for two window lengths: three weeks or shorter, and longer. Partner skill and window length both affect outcomes, and blocking stops either from piling up in one arm. NIST's statistics handbook explains blocking as holding a nuisance factor fixed within groups so the factor under study varies inside each, and it puts the rule of thumb as "Block what you can, randomize what you cannot."

Three rules handle the awkward rows. If the company applies before its assigned day, the check-in is cancelled and the row stays in its arm, so arms are compared as assigned, not as delivered. If the founder adds an avoid week later, the check-in moves to the nearest allowed day in the same third and the move is logged. If the assigned day falls after day 90, it stays where it is.

## One clock for everyone: why the read is day 120, not two weeks after the check-in

Reading each row a fixed number of days after its check-in looks fair and is not. A founder checked in on day 12 would be read on day 26; one checked in on day 38, on day 52. The late row has had twice as long since the yes for the company to act, so the comparison would partly measure elapsed time instead of timing. Reliability engineers have a word for outcomes cut off by the end of observation. The NIST handbook, paraphrased, calls data right censored when a test stops at a fixed time and anything that would have happened later is missing.

So every row is read at day 120 after its own yes, past the 90-day window, whenever its check-in went out. The measures are counted the same way in both arms.

- **Primary:** qualified next steps per assigned founder, arm L against arm E.
- **Timing fit:** when a row closes, founders who replied at least once get one optional question: did the check-in come too early, about right or too late? Answers are reported by arm, with the share who did not answer.
- **Harm:** stop requests and complaints per assigned founder, and any check-in that landed inside an avoid week, counted as a breach.
- **Descriptive only:** days from the yes to the application, for companies that applied.

Partners copy dates and dashboard stage words, nothing more. No company records or figures enter the study, and nobody asks a company anything for the study's sake. Whether a company's records suit a current program is settled in SourceX's review, untouched by when anyone wrote.

## Floor, stop rules and the reading date

- **Enrollment:** two quarters of founder yeses from consenting partners who have signed the Channel Partner Agreement.
- **Floor:** no comparison is read unless each arm holds at least 30 assigned founders; below that, counts are reported and nothing is concluded. The number is a setting written into the plan, not a calculated sample size.
- **Breach stop:** a check-in outside the window or inside an avoid week pauses enrollment until the cause is fixed.
- **Harm stop:** if one arm's stop requests exceed the other's by three or more at a monthly check, that arm closes and its remaining founders get a middle-third check-in.
- **Program change:** SourceX's programs can change their criteria; a dated change is reported in the results, never absorbed into them.
- **Reading date:** when the last enrolled founder reaches day 120.

## Four founders through the rules (Illustrative)

Illustrative. Generic companies, invented answers, no outcome implied for anyone.

| Founder | What they said at the yes | Group and window | Check-in | How the row ends |
|---|---|---|---|---|
| CEO of a 140-person claims administrator | "Not before the 15th, and skip the last week of the month, that's our close" | Assigned; day 15 to 48, close week removed | Arm L, day 44 | Applied on day 51; read on day 120 like every other row |
| CEO of a 300-person distributor | "Whenever suits you" | Assigned; default day 7 to 42 | Arm E, planned for day 12 | Applied on day 9; check-in cancelled, row stays in arm E |
| CEO of a 90-person engineering firm | "Try me on the 3rd of next month" | Named a day; observed | That day | No reply by day 90; closed as no reply |
| CEO of a 210-person staffing firm | "Please don't chase, I'll come back to you" | Asked for none; observed | None | Declined in writing on day 30; nothing further sent |

The second row is the one teams get wrong. Dropping it because no check-in was sent would remove an early-arm success and bias the comparison toward arm L.

## Reading the comparison

If the arms tie, the practical rule is to let founders choose and use the middle of the window when they will not. If arm L does better with no rise in harm, a later default makes sense for founders who delegate, and never for those who named a day. If one arm draws more stop requests, that arm is dropped whatever its applications. A result inside the assigned group says nothing about founders who chose their own timing, and it never justifies a second message. For timing set by events rather than founders, see [the event follow-up that records what the founder requested](/guides/the-event-follow-up-that-records-what-the-founder-requested) and [why a fundraising sprint may be the wrong window](/guides/why-a-fundraising-sprint-may-be-the-wrong-referral-window).

## Where a timing comparison can mislead

- **Who delegates.** Founders who leave timing open may be busier, or less interested, than those who pick a day; the result describes them alone.
- **Uneven windows.** Avoid weeks shrink some windows more than others, so arm L's day can sit close to arm E's in a short window; window length is blocked for that reason.
- **Calendar drift.** A cohort that enrolls before a holiday season meets different weeks than one that enrolls after it; enrollment month is reported by arm.
- **Small numbers.** Even at the floor, a few applications either way move a share; counts are printed beside every share.
- **Hindsight in the timing answer.** Founders who applied may call any timing "about right", so timing fit is read next to outcomes, never instead of them.
- **Stage lag.** Dashboard stages can update after day 120; the report states the date each stage was read.

**Method note.** The design rests on SourceX's terms, how-it-works and who-qualifies pages, checked 2026-10-28, and on the NIST handbook's treatment of blocking and censoring. It measures applications, stages, timing answers and stop requests per assigned founder; it does not measure demand, record quality, credit or any payout, and every setting named here is a choice for the plan, not a finding.

## 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.

The window question works before any study does: [register](/register), sign the Channel Partner Agreement before you submit anyone, check each company against [who qualifies](/who-qualifies) and the [program terms](/terms), and find related designs in the [guides hub](/guides).

Timing a check-in is a referral question. Readers weighing SourceX as an investment are asking a different one, and can ask for investor materials on 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/how-it-works
- https://refer.sourcex.si/who-qualifies
- https://www.itl.nist.gov/div898/handbook/pri/section3/pri332.htm
- https://www.itl.nist.gov/div898/handbook/apr/section1/apr131.htm
- https://sourcex.si/investors
