War-game a referral program before reward tiers, eligibility rules, and attribution windows harden into company policy.
Referral programs fail when self-referrals clear the same as real ones, when double-counting across partners and affiliates is ignored, when reward cost outruns lifetime value for soft segments, and when support cannot explain clawbacks after a refund. A polished invite flow is not evidence.
Freeze the package
State who can refer and who can be referred, the reward form and timing, the attribution window, the fraud checks you will actually run, and the success metric after thirty and ninety days. Attach margin after rewards, refund and chargeback rules, partner-channel collisions, and support capacity for disputes. If growth, finance, and legal disagree on clawback language, reconcile before the seating pass.
Name the decision you will make if the war game finds nothing new, and the kill criteria if reward cost destroys margin without changing qualified pipeline.
Seats
- Referred buyer. What feels like bait after the reward clears.
- Finance. Margin after rewards, cash timing, and fake growth.
- Fraud and risk. Self-referral, coupon stacking, and refund games.
- Partners and affiliates. Which paths collide or dual-count.
- Skeptic. The claim that looks strongest and is least sourced.
Optional counsel seat if regulated marketing or gambling-adjacent rules bind the language. Give every seat the same pack.
Run the loop
Feed Pingpong the program memo and exhibits. First pass steelmans the ladder. Later passes attack from the seats above. Final pass turns surviving objections into narrower eligibility, clearer clawbacks, or a hold. Delete invented viral coefficients and dual-counted partner credits.
Force a day-after narrative: what happens if a cohort refunds after rewards pay, if a partner demands the same terms, or if support volume spikes on disputed attributions. If those stories are stronger than your fraud and margin plan, fix the package before the program ships. Separate true new logos from recycled accounts and trial-hoppers. If blended dashboards hide a soft segment, ask finance and risk seats to attack until each path has owners and kill criteria.
Ask every seat to mark which fraud checks are optional in practice. Optional checks that never run still appear in the flowchart. If risk skips device or payment fingerprinting under load, finance should treat that skip as a design failure, not a staffing gap. Write the mandatory checks into the package, then attack whether capacity can actually clear them.
When the program leans on a standing cash reward, force finance to price the next cohort that will farm the same deal without genuine intent. Write eligibility, clawbacks, and exception owners into the package, then ask the finance and risk seats to attack those rules. A program that only works by forever matching the richest invite is margin erosion with better reporting. Related: war-game a growth experiment, war-game a discount policy, war-game a packaging change, stress-test a billing change, and the war-game decisions hub. Process: how to run a Pingpong.