Work & thinkingApproachAllocation
70 / 20 / 10
A starting point for dividing a budget between what works today and what might work next. The split should give the team room to learn while keeping the business’s current needs in view.
- Problem
- How do you fund new ideas when the proven work has the clearest return?
- Principle
- Give current performance and future learning their own place in the budget.
- Allocation
- 70 · 20 · 10Proven work · promising ideas · new experiments
Give each part of the budget a job.
The 70% supports work that has earned continued investment. The 20% develops promising ideas that need more evidence. The 10% makes room for new experiments whose outcomes are less certain.
Those percentages are a starting point for a conversation. The useful question is whether the budget gives each kind of work enough room to do its job.
Interactive example
Try a budget split over four quarters.
Move the sliders to see how the allocation changes the result under this model’s assumptions. The example assumes that proven work becomes less efficient over time and that some new ideas develop into future sources of return.
This is an illustration, not a forecast or evidence that one split will always win.
This split puts 70% into proven work, 20% into promising ideas and 10% into new experiments.
Modeled return this quarter
$1.09
modeled return per $1
Over four quarters
Your splitAll proven
These are illustrative assumptions, not client results. In this model, proven work starts at about $1.19 per $1 and loses 10% of its efficiency each quarter. Promising ideas succeed 35% of the time and, when they do, become the next quarter’s proven work. Successful experiments become promising ideas, then proven work two quarters later. The results depend on these assumptions.
Protect room for an honest result.
When an experiment fails, it’s easy to move its budget back to familiar work. Do that often enough and the testing budget disappears.
The team needs room to report that an idea didn’t work. Otherwise it learns to propose only what already looks safe. Protecting the space to test means being willing to hear the result, including one that gives us a reason to stop pursuing an idea.
Recommendations