- 1. Where Does Seventy Percent Come From?
- 2. Waiting for Certainty Is Itself a Decision
- 3. Would the Missing Information Change the Answer?
- 4. What Does Being Wrong Actually Cost Here?
- 5. Can You Test Faster Than You Can Research?
- 6. When You Genuinely Need More
- 7. How to Say “We Have Enough” Out Loud
- About the Stagnation Assassin
Seventy percent. That is the share of the information you wish you had at which most operating decisions should be made, and the remaining thirty percent is not a gap to be closed. It is a cost to be accepted, because closing it takes longer than the decision is worth and because a meaningful part of it cannot be closed at all before the decision is made.
The reason this needs saying is that nobody in a business ever argues for less information. There is no constituency for deciding sooner. Ask for two more weeks of analysis and you sound careful; say the current picture is enough and you sound reckless, and that asymmetry is a standing tax on every organization that has not deliberately countered it. The tax is paid in elapsed time, and elapsed time is not free, whatever your process reporting says about it.
1. Where Does Seventy Percent Come From?
It is a threshold, not a measurement. The method I use for this is the 70% Rule: when you judge that you hold roughly seventy percent of the information you would like, decide, and treat the deliberate acceptance of the remaining uncertainty as part of the decision rather than as a failure of preparation.
The number is doing rhetorical work as much as analytical work, and that is deliberate. You cannot actually measure the completeness of your information, because the missing part is by definition unknown. What a stated threshold does is give the room a shared vocabulary for the moment when enough is enough, which is otherwise a moment that never arrives, since there is always one more thing that could be checked.
The practical effect is to move the conversation from “are we ready” to “what specifically are we still missing, and would it change anything.” The first question has no answer. The second has three, and they are the tests in sections three, four and five.
Nor is the threshold uniform. Seventy percent is the default for the ordinary case, and sections four and six move it in both directions: down toward fifty for cheaply reversible calls, up toward as much as you can get for the small number that genuinely cannot be undone. A single threshold applied to every decision is the same error as no threshold at all, arriving from the opposite direction.
What the rule is not is a license to decide from nothing. Thirty percent is not most of the picture. A decision made on a genuinely thin understanding is a guess wearing a framework, and the point of a threshold is that it cuts in both directions.
2. Waiting for Certainty Is Itself a Decision
The central confusion this whole topic rests on is the belief that not deciding is a neutral state. It is not. While a decision is open, the business continues to operate under the status quo, and continuing under the status quo is one of the available options. Choosing it by default is still choosing it.
Frame it that way and the arithmetic changes. The question stops being “should we act or wait until we know more” and becomes “should we do A, or B, or continue doing what we are doing, given that continuing has its own cost and that cost accrues weekly.” The third option is usually the worst of the three, and it is the one that gets selected by silence.
Not deciding is not a neutral state. It selects the status quo, and the status quo is on the option list whether anyone put it there or not.
I have run turnarounds at Berkshire Hathaway, Illinois Tool Works, and Whirlpool, and spent four years owning a small industrial tank liner manufacturer that I eventually sold to its largest vendor. Across all of it, the pattern that repeats is not that businesses made bad calls on incomplete information. It is that they made no call, for a long time, on information that was already sufficient, and then made the same call anyway once the situation had deteriorated enough to remove the debate.
3. Would the Missing Information Change the Answer?
Name the specific thing you are missing, then state what you would do if it came back high and what you would do if it came back low. If the two answers are the same, you already have enough, and the analysis is reassurance rather than input.
This test disposes of more delay than any other single question I know, because most requested analysis fails it. The work is real, the analyst is capable, and the output would not have altered the outcome. Look back at the last five things your team commissioned before a decision and count how many produced a different answer than the one that would have been given beforehand. In my experience it is rarely more than one.
The discipline version is to require a stated threshold with every analysis request. One line, written before the work starts: above this number we do A, below it we do B. Requests that cannot produce that line are not decision support, and the correct response is to decide now and run the analysis afterward if it is still interesting.
There is a subtler version worth catching. Sometimes the missing information would change the answer, but not before the window closes. Information that arrives after the decision has effectively been made by default is the same as information that would not have mattered, and it should be treated identically.
4. What Does Being Wrong Actually Cost Here?
Estimate the cost of the wrong call and the cost of four more weeks of waiting, and put the two numbers side by side. Most operating decisions are partially reversible, which bounds the first number and leaves the second unbounded, and a bounded cost against an unbounded one is not a close call.
Reversibility, not importance, is the variable that should set your information threshold. A pricing move is visible within a quarter and correctable. A scheduling change is visible within a week. A hire takes months to unwind, which is why hiring genuinely deserves more deliberation than most decisions get. A plant closure is close to permanent and belongs in a different category entirely.
Sort your open decisions by reversibility and you will find the majority sit in the correctable band, where seventy percent is generous and fifty would often do. The small number in the irreversible band deserve every week you spend on them, and treating those two groups identically is the actual error most organizations make. They do not deliberate too much in general. They deliberate uniformly, which means correctly on a few and expensively on the rest.
Reversibility sets the threshold, not importance. Most organizations do not deliberate too much; they deliberate uniformly, which is correct on a few decisions and expensive on all the others.
5. Can You Test Faster Than You Can Research?
Ask whether a small version of the decision would produce better information than the analysis would, and faster. Often it does, because a live test answers the question of what actually happens rather than the question of what the data suggests might happen.
One plant instead of three. One product line instead of the portfolio. One customer, one quarter, one shift. The bounded version of a decision is available far more often than it gets used, and it converts an information problem into an execution problem, which is the better problem to have because execution produces evidence.
The objection is that a partial rollout costs more than a full one and creates two operating standards for a period. Both are true and both are usually smaller than the cost of the delay they replace. What is not usually true, and what does defeat this approach, is a decision whose economics only work at full scale. Those exist, and for them the test is genuinely unavailable.
The order matters too. Run the bounded version on the case most likely to fail rather than the one most likely to succeed. A pilot on your best plant tells you almost nothing, because your best plant makes most things work, and the result will not transfer.
The other failure mode is a test with no decision criteria, which becomes a pilot that runs indefinitely because nobody specified what result would trigger the full move. Set the criteria and the end date before it starts, in the same sentence as the approval.
6. When You Genuinely Need More
Three situations deserve the extra weeks, and being specific about them is what keeps the rest of this from being an argument for recklessness.
Irreversibility is the first. A decision that cannot be undone, or can only be undone at a cost that would threaten the business, earns whatever time it takes. Plant closures, acquisitions, exits from a segment, anything that removes a capability you cannot rebuild.
Novelty is the second. Where the organization has no precedent and no pattern to draw on, the missing thirty percent is much more likely to contain something that changes the answer, because you have no basis for judging what is in it. Familiar decisions carry accumulated pattern recognition that stands in for information. Novel ones do not.
Concentrated downside is the third. A decision where the expected outcome is good but the worst case is severe should be treated as its worst case, particularly where the worst case involves safety, a regulatory exposure, or a single customer who represents a large share of the business. Expected value is the wrong frame when the tail can end the business.
Worth naming the false version of each, because all three get claimed more often than they apply. A decision is not irreversible because unwinding it would be embarrassing. It is not novel because this particular team has not done it, if the organization has. And the downside is not concentrated because someone in the room can imagine a bad outcome. The test in each case is specific: what capability is permanently lost, what pattern genuinely does not exist, what single exposure could actually end the business.
Everything outside those three is a candidate for deciding at seventy percent, and in most companies everything outside those three is the overwhelming majority of the queue.
7. How to Say “We Have Enough” Out Loud
Say what you are missing, say why it would not change the answer, name the cost of being wrong, and state the date you will check whether you were. Four sentences. That is what makes the position defensible in a room where the opposite position never has to defend itself.
The last sentence is the one that does the work. A decision made at seventy percent with a scheduled review date is a fundamentally different proposition from the same decision made and never revisited, and offering the review date converts the conversation from a debate about certainty into an agreement about how you will find out. It also creates the record that makes the decision improvable later, which is the thing most organizations lack entirely.
Then hold the date. A review date that passes without a review teaches everyone that the commitment was rhetorical, and the next time you say you have enough, the sentence carries nothing.
One thing to prepare for: the first few times someone says this out loud in a room that is not used to it, the reception will be cool. The position has no social protection, which is the whole reason it needs a script. What changes the room is not the argument, it is the second or third instance of a decision made at seventy percent, reviewed on the stated date, and found to have been right, with the elapsed time visibly shorter than it would have been. Two of those and the sentence starts carrying weight on its own.
Where these decisions get made, and on what rhythm, is a structural question that sits in the operating cadence. Why the waiting costs more than most leaders estimate is set out in what a slow decision actually costs, and the specific places the waiting happens are covered in the five places decisions get stuck.
About the Stagnation Assassin
Todd Hagopian is a Fortune 500 transformation executive who has generated $3B in shareholder value across Berkshire Hathaway, Illinois Tool Works, Whirlpool, and JBT Marel, where he serves as Global VP of Product Strategy and Marketing. Known as The Stagnation Assassin, he is the author of two published books: The Unfair Advantage: Weaponizing the Hypomanic Toolbox and Stagnation Assassin: The Anti-Consultant Manifesto. His blog is published in 15+ languages and read by operators worldwide. Bring him to your stage via his speaking page or connect with him on LinkedIn.
Answer this one yourself: take the open decision that has been open longest, name the one thing you are still waiting on, and write down what you would do if it came back high and what you would do if it came back low. If the two are the same, you have had enough information for a while. If you want a second read on a decision that is stuck, send me the two answers.