Week 1 — Audit
List every recurring discount you offer. For each, write down: who converts on it, what the trigger event is, and what the payout mechanic looks like. Most will fail this test — they have no trigger and no ceiling.
Week 2 — Design one outcome incentive
Pick one campaign. Rewrite it as trigger + payout + hedge + ceiling. Test it against your existing discount as a controlled parallel.
Week 3 — Watch the ROAS
By day 10, realized ROAS on the outcome campaign will be visibly higher — because you're not paying non-triggering customers. Wait for the data before scaling.
Week 4 — Retire the discount
Once outcome ROAS clears discount ROAS by a comfortable margin, sunset the discount. Reallocate the freed budget to a second outcome campaign. Repeat.