Every marketing budget conversation eventually collapses into one question. How much of this should go to new bets. The wrong answer is forty percent. The wrong answer is also zero. The right answer is ten, and the difficulty is not arriving at the number, it is holding it.
Seventy percent goes to what already works. Twenty percent goes to what is showing early signal, meaning something you have run long enough to see a trend but not long enough to trust. Ten percent funds tests that could become next year's twenty. The math is boring on purpose. A framework that requires cleverness to apply will not survive a bad quarter.
Most budgets do not fail on allocation. They fail because the ten percent quietly becomes forty during a good quarter and starves the engine that paid for it.
That drift is almost always well intentioned. Revenue is ahead of plan, a founder saw something compelling at a conference, a new channel is having a moment on social, and it feels conservative to keep pouring money into the boring thing that is working. Six months later the boring thing has decayed from neglect, the novel bets have not compounded yet, and the number goes sideways in a quarter where nothing obviously went wrong.
The other failure runs the opposite direction. Teams under pressure cut the ten to zero, which feels responsible and is not. It guarantees that when the current channel saturates, and every channel saturates, you will have nothing warm to move into. The ten percent is not discretionary spend. It is the option you are buying on your own future.
One thing that makes this easier to defend. Write down at the start of each quarter which bucket every line item sits in, and share it. The split stops being a philosophy and becomes a document somebody would have to visibly edit in order to break it. That small amount of friction is most of what holds the discipline in place.