How I split a $1,000 test budget across offers (the 70/20/10 rule)
Throwing equal money at everything is how you learn nothing slowly. Here's my allocation playbook.
The setup
Monthly test budget $1,000. Three buckets of offers competing for it.
The move
— 70% to proven: campaigns already green get the bulk. Defend and scale what works.
— 20% to promising: campaigns that broke even or showed one good signal. Give them a real shot.
— 10% to wild: brand-new offers/angles with no data. Cheap lottery tickets.
Rebalance weekly. A wild that hits graduates to promising, a promising that wins graduates to proven, losers free their budget.
The numbers
The 10% wild bucket ($100) found a sweepstakes angle that became my top earner at +40%. The 70% kept the lights on. Without a dedicated wild slice I'd never have tested it, the proven stuff always felt safer.
The lesson
With no forced exploration budget, you over-invest in known winners and miss the next one. The 10% is insurance against staleness.
What I'd do differently
Never let the wild bucket roll into proven mid-week, the discipline is the point. Numbers illustrative.
Arb Files
@ArbFiles