Only active customers bring anyone. A loop that leaks customers leaks invites, so fix the leak before you raise the reward.
Referred customers per month is active customers, times how often each invites, times how many invites convert. Most teams work the last two. The first is usually the biggest, and it's a retention number.
referred customers per month = active customers × invites per active customer per month × invite conversion
Hold the reward and the page constant and let the active base shrink by a fifth, and referrals fall by a fifth. No change to the referral program fixes that. The Second Order was about keeping customers for a second order. In a loop, that second order is also the second month they have a reason to mention you.
A loop can't outgrow its churn.
OVRLND's membership, a monthly subscription sold through Recharge, was the loop's fuel line: members get entries every month and a draw every week. In June 2026, while the Bronco giveaway surged, 1,063 memberships started and 676 were cancelled. In May it had been 490 and 164.
ReportedThose are same-month counts, not a churn rate: many June cancels came from earlier cohorts. But the direction was plain. People subscribed to get entries for one truck and left.
The plan fixed it in an order that works for any subscription:
Why inside the cancellation? Because OVRLND's win-back email, sent after cancelling, converted at 0.25%. A save offered after someone has left is a win-back, and win-backs are the weakest tool in the box.
The same logic ran my work at Binance. A referrer's income depended on friends who kept trading, so every retention gain was also a referral gain. The first-trade milestones and the loyalty tiers from chapter 13 weren't a separate program from the referral loop. They were its fuel.
Dropbox made the same point from the other side: its referral reward was more storage, which put more files in Dropbox, which made users stay. The reward fed retention, and retention fed the next referral.
The simulator below runs the formula month by month: last month's active customers, minus churn, plus the customers you buy, plus the ones your active customers bring. Change churn and watch what happens to the referral line.
Chart active customers and referred customers by month for the last year. If referred customers held steady or grew while active customers fell, your loop runs on a few loyal inviters rather than the base, and Seed the Hard Side matters more than retention here.
This is one chapter of Close the Loop, which is free and readable in full on a single page with no form in front of it.