The famous rule for splitting an innovation budget comes with a second number that almost nobody quotes, and the second number is the interesting one.
It also does not apply to most small businesses, for a reason worth understanding before you divide anything.
What the 70-20-10 rule actually says
Bansi Nagji and Geoff Tuff, both Deloitte partners at the time, published the innovation ambition matrix in Harvard Business Review in 2012. It sorts projects into three kinds: core work that improves what you already sell to the customers you already have, adjacent work that pushes an existing strength into a nearby market, and transformational work that builds something new for people who are not your customers yet.
The allocation everyone remembers is that successful firms put “70 percent of investments in core projects, 20 percent in adjacent projects, and 10 percent in transformational.” The part that gets dropped is the return side, where the ratio flips: “10 percent of returns are from core projects, 20 percent from adjacent, and 70 percent from transformational.” (deloitte.com)
Read those two sentences together and the rule stops being a budget instruction. It is a statement about averages across a lot of projects run at the same time. Seventy percent of returns arriving from the transformational slice only makes sense if you are running enough transformational bets that a few can pay for the many that fail. That is a portfolio.
A six-person shop is not running a portfolio. It is running one thing at a time and hoping it works. Applying a portfolio ratio to a single bet produces a number with no meaning attached: 10 percent of a 60,000 dollar budget is 6,000 dollars, but 6,000 dollars is not a diversified transformational allocation, it is one attempt that either lands or does not.
Size the bet by what you can lose, not by a percentage
The better question is not what share of the budget a new idea deserves. It is how much cash can leave the business and not come back without changing anything that matters.
There is a good measure for this already. The JPMorgan Chase Institute studied 597,000 small businesses across 470 million transactions and found that the median small business holds 27 cash buffer days in reserve, meaning 27 days of normal outflows sitting in the account. (jpmorganchase.com) That is the number a new bet is actually spending, and it is far more honest than a percentage.
The arithmetic takes a minute. Take your average monthly outflow, everything that leaves the account in a normal month, and divide by 30 to get a daily figure. A business with 40,000 dollars of monthly outflow is spending about 1,333 dollars a day, so 27 buffer days is roughly 36,000 dollars of cash on hand.
Now price the idea in the same units. A 9,000 dollar bet is not 15 percent of anything useful. It is close to seven buffer days. Spending it moves the business from 27 days of cover to about 20. That is the decision, stated in the only terms that will matter if the bet fails and a slow month lands on top of it.
Most owners find the cap is smaller than the percentage rule suggested and considerably more real. If the honest answer is that you cannot lose more than three buffer days, then the plan is a three-buffer-day experiment, and the idea has to be cut down to that size or wait. Sizing the target first is the same discipline that makes a three-year goal survive its own arithmetic.
Write the stop rule before the money moves
This is the part I would argue for hardest, and it is judgment rather than a finding.
The expensive failure in a small business is almost never choosing the wrong idea. It is funding the wrong idea for fourteen months because no one ever agreed what failure would look like. The bet does not die, it just quietly keeps costing money, and every month that passes makes stopping feel more like admitting the previous months were wasted.
The fix is boring and it works: decide the ending before the beginning. Before any money moves, write three things down in plain language.
First, the number that would make you continue. Not “it seems promising,” but a figure you could read off an invoice or a booking calendar. Eleven paid orders. Four repeat customers. One signed contract over 5,000 dollars.
Second, the date you check it. Put a real date on it, far enough out to be fair to the idea and close enough that the loss is still the size you agreed to. Ninety days suits most things a small business tries.
Third, what happens if the number is not there. This is the one people skip, and skipping it is why the other two do not save anyone. Write the actual sentence: “if we have fewer than eleven paid orders by 15 October, we stop spending on this and go back to the core work.” An unwritten stop rule is not a stop rule. It is an intention, and intentions lose arguments to sunk costs.
The rule also has to be written before you are emotionally invested, which is precisely why it has to be written before the money moves. Nobody has ever set a fair kill condition in month nine.
Where the cap and the date live
None of this needs new software, and a spreadsheet with a calendar reminder covers the arithmetic. What it does need is for the cap, the review date, and the stop condition to sit somewhere a second person can see them, because the whole point is to remove the decision from your mood on the day. Prices verified on the vendors’ own pricing pages today.
Trello is free for up to 10 collaborators per Workspace, which is enough to hold one card per bet with the cap in the description and a hard due date on the review. Standard is 5 dollars per user per month billed annually or 6 dollars billed monthly, and Premium is 10 dollars annually or 12.50 dollars monthly. (trello.com/pricing)
Asana covers the same ground with stronger reporting if several people need to see the review dates at once. Its Starter plan is 10.99 dollars per user per month billed annually or 13.49 dollars billed monthly, and Advanced is 24.99 dollars annually or 30.49 dollars monthly. (asana.com/pricing)
If the money for the bet has to come out of existing costs rather than spare cash, the renewal calendar in a playbook on protecting margin when costs rise is a better place to find it than a general cost-cutting pass.
The prompt that produced the original version of this article, the Innovation Budget Allocator, is built to walk a team through scoring several ideas against each other before the cap is set, and the wider library at BusinessPrompter.com covers the neighboring planning decisions.
What the cap does not decide
A spending cap answers how much, never whether. It cannot tell you that an idea is worth wanting, and it will happily approve a well-sized bet on something that does not deserve the attention. That judgment sits upstream, in whatever you have decided the business is actually for, which is why a vision statement is only useful when it can argue against a real opportunity.
There is also a cheap way to fund innovation that owners reach for when cash is tight, which is to run the experiment on top of everyone’s existing week. It is not free. It is paid for in the hours of people who were already fully booked, and it usually fails on execution rather than on premise, which teaches you nothing about whether the idea was any good. If a bet is worth making, it is worth buying the time to make it properly, and the capacity a new line of work needs is the argument for the next hire rather than a test of how far the current team stretches.
Do this before the next planning conversation
Work out your buffer days first: cash in the account, divided by average daily outflow. Then write the cap for the next idea as a number of those days rather than a share of anything. Two figures, one line, before anyone starts advocating for a favorite project. The conversation that follows is a different and much shorter one.
Frequently Asked Questions
Does the 70-20-10 split ever make sense for a small business?
Only when you genuinely have several projects running at once. The ratio comes with an inverse return pattern, where “10 percent of returns are from core projects, 20 percent from adjacent, and 70 percent from transformational,” and that pattern is an average across many bets rather than a prediction about any one of them. (deloitte.com) With one bet at a time you have a queue, so sequencing and size matter more than the split.
How do I work out my cash buffer days?
Divide the cash in the business account by your average daily outflow, which is a normal month’s total outgoings divided by 30. The JPMorgan Chase Institute found the median small business holds 27 cash buffer days across a study of 597,000 businesses. (jpmorganchase.com) Comparing your own figure to that gives you a sense of how much room a new bet is really eating.
What if the review date arrives and the result is borderline?
Borderline is the case the stop rule exists for, and the honest answer is that borderline counts as a miss, because the number was set when you were thinking clearly. If you want a second run, treat it as a new decision with a new cap and a new date rather than an extension, and write both down again. Rolling a deadline forward without resetting the cap is how a bounded experiment turns into an open commitment.
Do I need paid software to track an innovation budget?
No. A spreadsheet and a recurring calendar entry hold the cap and the review date perfectly well. Trello’s free tier covers up to 10 collaborators per Workspace if you want the review date visible to the team, with Standard at 5 dollars per user per month billed annually. (trello.com/pricing) The tool matters much less than the date being visible to someone other than you.
