Lessons 79.1 and 79.2 borrowed trust from partners and aisles. Referrals borrow momentum from your own users — a happy person hands your work to a peer at exactly the moment it would help. That handoff only works when the shared object is genuinely useful. No incentive rescues a product nobody wants to recommend.

The vocabulary, in plain language

• Referral: one user bringing another user, through a personal recommendation rather than your advertising.
• Advocate: the happy user who refers — someone who already received value and can describe it credibly.
• Incentive: a reward for referring or accepting — credit, discount, extra storage, early access. A nudge, not the reason.
• Invite: the mechanism — a link, collaborator add, shared file, or personal message that carries the newcomer to the product.
• Recipient value: what the newcomer gets even if they never pay or reciprocate. Without it, the invite is a favor the advocate spends and the friend resents.
• Loop: the closed circle — user gets value, shares object, recipient gets value, becomes a user, shares again. An invite that dead-ends is a funnel, not a loop.

The mental model: a referral is a dinner recommendation. People share the restaurant that made their evening better, with a dish to name and a reservation link that works. Nobody recommends a menu they have not tasted because the waiter offered them a coupon.

Shareable first, incentivized second

The order matters. Organic sharing comes from the product's natural collaboration surface:

• A useful result worth forwarding — a completed brief, a before/after photo, a booked itinerary.
• A collaborator invite that improves the work — co-editing, shared review, joint plan.
• A shared object that lives between people — a research packet, playlist, template, local guide.
• A local recommendation that helps a neighbor — the tailor's care card, the shop's rush-slot list.

Use the Research Desk rule: invite a collaborator to a shared research packet only if collaboration genuinely improves the work — two analysts dividing sources, marking claims, merging notes. The packet carries its own reason to accept: the recipient sees cited evidence relevant to their question on arrival. That is recipient value before signup.

Spam is the counterfeit version: bulk "invite your contacts" prompts at signup, pre-checked contact imports, messages sent in the user's name they never wrote. It borrows the advocate's reputation and spends it. One spam wave teaches a community that your invite button cannot be trusted — and no incentive budget buys that trust back.

A reward may then help — but it cannot rescue. If the packet is worth sharing, a modest credit for both sides ("you both get one premium brief") speeds a loop already turning. If the packet is not worth sharing, the same credit buys embarrassed sends, fake accounts, and support tickets. Teach the distinction bluntly: build the shareable object first, price the incentive second, measure whether the loop turns with the incentive removed.

Design the loop, not the bribe

A working loop answers six questions in order:

1. Moment. When is the advocate happiest and most credible? After first value (Lesson 79.4), after a saved result, after a compliment — never at signup before anything good happened.

2. Object and message. What exactly travels — the packet, the template, the photo — and what words travel with it? Provide a pre-written, editable note in the advocate's voice, not marketing copy. "Here is the brief we used for the board — sources linked, steal the template" beats "Check out this amazing AI platform!"

3. Recipient benefit. What does the newcomer get in one click without creating an account or owing a favor? A readable sample, a usable template, a real answer. If the benefit requires signup first, most goodwill dies at the gate.

4. Destination. Where does the click land — the shared object itself, with context and one next action. Not the homepage, not a pricing wall.

5. Tracking. Which invite led to which activated user, honestly attributed. Keep it simple: unique link per advocate, activation defined as first value completed, weekly review.

6. Abuse. What stops gaming — self-referrals, disposable emails, credit farming? Cap rewards, require real activation on both sides, log anomalies, and publish the boundary plainly.

Run a local-business check: the tailor's "give a friend 15% off their first hem, get 15% off your next" works because the garment, fitting, and timing are already excellent. The card travels after pickup, names the garment, and lands on a booking page with real slots. Same loop, different aisle.

Exercise: write the one loop that survives without reward

Create REFERRAL-LOOP.md. If the incentive line were deleted, the flow must still make sense.

# REFERRAL-LOOP.md — [Project]

## Moment (after value, not at signup)
- Trigger event + user emotion: ___

## Sharable object + message
- Object (packet/template/result): ___
- Pre-written editable note (advocate voice): ___

## Recipient benefit (no signup required to see value)
- One-click value: ___

## Destination + next action
- Landing object URL type: ___ / one action: ___

## Tracking (simple, honest)
- Link scheme: ___ / activation definition: ___
- Weekly fields: sent / opened / activated / abused: ___

## Abuse + boundaries
- Caps, eligibility, private-feedback path: ___
- What we will never do (no spam list): ___

## Incentive (optional, last)
- Advocate gets: ___ / recipient gets: ___ / why loop works without it: ___

Worked mini-example — ByeBuy Classroom: moment is completing a first source-checked brief; object is the shareable brief page with method notes; message is "I used this checklist for Friday's memo — here is my result, blank attached"; recipient sees the full brief without login; destination offers "save your own copy" as the one action; tracking is per-learner link with activation counted on first saved brief; abuse cap is three credited invites per month with self-referral excluded; incentive is one bonus template pack — explicitly unnecessary for the share to make sense.

Finish line: a REFERRAL-LOOP.md where moment, object, message, benefit, destination, tracking, and abuse are each specified, and the loop reads credibly with the incentive removed.

Verify quickly: ask a real user, "Would you send this to a peer you respect, with your name on it, for no reward?" If they hesitate, fix the object and benefit — not the reward size.

Common failure mode: the signup-wall ambush — the recipient clicks an enthusiastic invite and meets a registration form before seeing anything useful. Every such click spends the advocate's credibility. Show value first, ask second.

Check your understanding

1. Why must the sharable object precede the incentive? 2. What makes a collaborator invite genuine versus spam? 3. How do tracking and abuse guards keep a loop honest as it scales?

Next

Referrals carry users you have already won. Lesson 79.4 builds the win itself into the product — product-led growth where first value arrives before any sales conversation, and the experience creates the next user.