This design is for a partner growth manager who shows a reward illustration to other introducers: a firm's advisors, a network's members, a fund's venture partners. It does not test SourceX's own referral earnings calculator, and it asks a different question from the maximum-payout headline study, which compares words on SourceX's landing page across site visitors. Here the object is a tool, the people are your own introducers, and the risk is a number they carry into conversations with CEOs.
Why the opening number matters more than the formula
A calculator does one honest thing: it applies the published rule, which by default is the greater of $50,000 plus 5% and 15% of Actual Supplier Payout, with no cap, on a company's first qualifying transaction, with 15% on further qualifying transactions first agreed within 12 months. Every other number on the screen is an input somebody chose. If the input box opens already filled with a payout of USD 1,000,000, the tool has chosen the first figure the reader meets: a USD 150,000 reward.
Tversky and Kahneman's 1974 paper in Science named "adjustment from an anchor" as one of three heuristics people use for a numerical prediction when a relevant value is available, and described such shortcuts as economical but prone to systematic, predictable errors (paraphrase). A pre-filled large payout looks like a relevant value. Our thesis, untested for this program, is that an introducer who sees a six-figure reward first treats it as a normal result, repeats it to companies and spends more effort on poor-fit companies in hope of the large case.
SourceX's own documents lean the same way. Section 5 of the terms says "Estimates shown in the portal or calculator are projections, not promises." Paragraph 9 of Schedule A in the sample Channel Partner Agreement adds that "Website updates, calculator changes and dashboard edits do not change this Agreement." SourceX publishes no typical payout, conversion rate or average reward, so a calculator has no honest value to open on.
What a partner's calculator may show, and what it may not
| May show | May not show |
|---|---|
| The rule as the terms state it: the greater of $50,000 plus 5% and 15% of Actual Supplier Payout, with no cap, on the first qualifying transaction | A pre-filled payout, a slider resting at the top, or any figure called typical |
| Worked cards labeled Illustrative, each tied to a section of the terms | A payment date counted from an introduction or a signature |
| A zero case: a company SourceX already knew earns nothing | Anything about what the company itself will be paid |
| One USD 50,000 floor for a parent and all its subsidiaries, on the group's first qualifying transaction | A total built by giving each group company its own floor on separate rows |
| Section 5's line that estimates are projections, not promises | Words such as "expect", "likely" or "your next reward" beside a figure |
| A link to the full program terms | Any base other than what SourceX actually pays the company; the reward is never a share of SourceX's fees |
Two rows need their reason spelled out. The company row: the reward never comes out of the company's payment. Section 4 says the reward "is paid separately by SourceX and is never deducted from the supplier's agreed compensation", and SourceX funds it from its own share, so the company is paid the same with or without a partner. A calculator that hints otherwise misleads the company as well as the introducer. The timing row: payment falls due 60 calendar days after whichever comes later, completed handoff or the day the Acquisition Partner accepts the full, validated dataset in writing as final, and a condition still open at that deadline delays payment without starting a new 60-day period. No input field knows those dates.
The balanced scenario set (Illustrative)
Illustrative. The companies are generic and every amount is invented, so read none of them as a forecast, a quote or a deal that happened. Version B shows all five cards at once, in this order, beside the empty input, so the large case is one card of five rather than the first thing on screen.
| Card | Situation | What the published rule gives | Term that decides it |
|---|---|---|---|
| 1. Known company | A 180-person insurer whose CEO had emailed SourceX the year before | $0; earlier email correspondence is a relationship, so this is not a new company | Section 3 |
| 2. Below the minimum | A 140-person distributor paid $40,000 in total | $0; below the USD 100,000 Minimum Transaction, so nothing qualifies unless SourceX agreed otherwise in writing | Section 4 |
| 3. Installments | A 260-person freight firm paid $120,000, then $160,000 more | $56,000 after the first installment; $64,000 in total after the second, since the USD 50,000 element comes once and the 5% element follows each installment | Section 4 |
| 4. Group and the floor | Two subsidiaries of one 900-person group paid $800,000 between them, $400,000 each, the second agreed within 12 months of the first | $70,000 on the first transaction; $60,000 on the second, a repeat at 15% with no second floor; $130,000 for the whole group, never two floors | Section 4; the floor once per company |
| 5. Your next introduction | Any company you have not yet introduced | Unknown; no typical amount is published | Section 5 |
Card 5 is the anchoring probe. Its only correct answer is that nobody knows, and the share of introducers who name a figure instead, above all a six-figure one, is the clearest sign the calculator teaches the wrong lesson. Cards 1 and 2 test whether people see the two ways a result falls to zero: a company SourceX already knew, and a transaction below the Minimum Transaction. Card 4 tests whether they see that the USD 50,000 floor comes once per company and never on a repeat. Whether an introduction still earns credit when a sister company of that business reached SourceX through someone else is a question the published terms leave open; SourceX settles it in duplicate review, so no card pretends to answer it.
The balanced scenario usability study
| Element | Specification |
|---|---|
| Question | Does version B reduce anchoring on the large case without reducing qualified introductions? |
| Who takes part | Introducers you are bringing into the program who have not yet seen a reward figure from you; each signs the SourceX Channel Partner Agreement before submitting any introduction |
| Version A | Your current calculator or slide, unchanged |
| Version B | Empty input, no default, the five cards beside it, Illustrative on every output, section 5's line under the result |
| Allocation | A coin flip per introducer, logged before the session; the version sticks for the whole study |
| Session | 20 minutes with one moderator: two minutes of free use, the card 5 question asked cold, then cards 1 to 4 as tasks, thinking aloud |
| Field window | 120 days from each session, which covers the 90-day attribution window plus some review time |
| Primary measure | Qualified introductions per introducer: companies they introduced that reached a verified application credited to that introduction and not closed at the cutoff |
| Where the count comes from | Dashboard stages each introducer agrees to share, copied word for word, never estimated |
| Analysis | By assigned version, whatever the introducer did afterward |
The card 5 question comes before the tasks on purpose. Asked after cards 1 to 4, it would measure what the tasks just taught rather than what the calculator taught.
Secondary measures, recorded the same way in both versions:
- Anchored answers. The share of card 5 answers that give a figure, and the share at or above the reward the pre-filled payout would produce.
- Task accuracy. Cards 1 to 4 scored against the table, with the introducer's reasoning noted.
- Poor-fit introductions. Companies closed by SourceX, using the stage words as displayed.
- Unwanted effects. An introducer quoting a calculator figure to a CEO, a CEO reading a figure as the company's own payment, and introducers who drop out after meeting the zero card.
Numbers, timing and the rules that stop the study
Set these before the first session. Eight introducers per version is our working floor for the session phase, enough to hear the same confusion twice; it is a choice, not a published standard. The field comparison needs more: compare rates only once each version has at least 20 introducers and every field window has closed, and below that report counts per version without percentages. Several partner firms may pool their results only if they use the same card set, the same session script and the same cutoff rule.
Stop or pause when any of these happens:
- A card disagrees with the terms, or SourceX publishes a new terms version: stop, correct and date the card set, and restart.
- An introducer reports presenting a calculator figure to a company as what it will get: pause both versions and fix the materials. Section 4.6 of the sample agreement bars a partner from telling a referred company that an estimate, positive interest, a signature or a data upload makes a purchase or a reward certain (paraphrase).
- A participant feels examined or pressed: they leave whenever they like, and their answers leave with them.
- One version pulls ahead on the primary measure: keep going. Stopping on a lead is how small tests produce false winners.
What each pattern of results would mean
| Pattern at the cutoff | What it suggests | What to change |
|---|---|---|
| Fewer card 5 figures under B, and no fewer qualified introductions | The cards correct the anchor at no visible cost | Move every calculator you publish to B's layout |
| Fewer card 5 figures under B, but fewer qualified introductions | Card 1's zero may discourage, or it is chance; small counts cannot tell | Keep the empty input and test the card order next, never a default |
| No difference on card 5 | Your introducers were not anchored, or nobody read the cards | Keep the simpler layout, provided it shows the rule and the zero |
| Too few qualified introductions to compare | The primary question stays open | Report counts; rest any layout change on session evidence, labeled as such |
Write down the smallest difference worth acting on before the first session, and show uncertainty as a range for the difference rather than a single number. A version that wins on task accuracy alone has not shown that it helps anyone introduce a company that fits.
Which companies the cards and the introductions are about
Every card describes a company that meets the published fit facts, because arithmetic about companies that could never be introduced teaches nothing. Those facts are a full-time headcount that reached 50 at its peak, with contractors left out; a history of records the company created itself; the rights to license them; a location and language that fit the sentence on who qualifies; and a CEO who wants the conversation, since the CEO decides for the company. Introducers ask about a company at metadata level only, meaning record types, rough years and systems, and never ask for files, samples or logins.
What stays unknown until SourceX reviews a company: whether its records match a current program, what the rights review finds and what any license would cover. A company that reached SourceX before your introducer did is a duplicate whatever a calculator shows, and only one partner can be credited per company; SourceX settles overlapping claims in duplicate review, working from dated records. An introducer who wants to weigh odds as well as arithmetic can use the expected earnings spreadsheet, which asks for their own estimates, and the onboarding experiment checks whether people can apply the terms behind the cards.
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.
Before your calculator reaches another introducer, register as a partner, check each card against who qualifies and the program terms, and make the large case one card among five rather than the first number on screen. Other test designs sit in the guides hub.