The Decision-to-Done Gap: The Most Expensive Problem in Executive Work

The Decision Nobody Carried

A leadership team makes a call on a Tuesday. The room agrees. Four or five people leave with something to do, and everyone who walks out believes it's handled.

Three weeks later the work turns up somewhere else, or nowhere. The deadline to fix it has already passed. Somebody goes looking for the person who dropped it, and the honest answer is that nobody did. The people were good. The meeting ran on schedule, and the document captured every decision the team asked it to. The plan came apart anyway, and no one can say exactly when it stopped being true.

For years I described that kind of loss as a workload problem. I blamed too many meetings and too much email. That explanation has the advantage of being visibly true, and I believed a version of it for most of my career. The cost sits somewhere more specific. It sits in the distance between the room where the leadership team made the call and the weeks after it, where the doing was supposed to happen.

That distance belongs to nobody.

We decided. The organization did something else.

What the First Essay Named

The first essay in this series ended by naming context blindness as "the condition the category exists to end: the structural gap between what the work stack captures and what the executive needs to know."

It also named the responsibility the category owns: "whether the decisions, commitments, and goals a leader sets in motion carry through."

Read those two sentences together and they don't quite fit. One names a failure that happens on the way into a decision, when the leader lacks the assembled picture the moment requires. The other claims territory that only opens after the call is made, when the decision has to survive contact with everything else. Both are right. The first essay left the relationship between them unexplained, and that relationship is the subject of this one.

They are two stages of a single distance. Naming only one of them is what has made this problem so hard to see. Attend only to the failure before the decision and the follow-through breakdowns look like separate, unrelated events. Attend only to the failure after it and the preparation cost looks like a personal habit. Writers have described the two failures separately for decades, because nothing named the thing they're two halves of.

The Gap Has a Shape

Every consequential call a leader faces has to travel a long way to become real. It travels first to a clear decision, then all the way to done. Most calls don't survive the trip.

That whole distance is the decision-to-done gap, and it's the most expensive problem in executive work. It costs a company a week at a time, hundreds of times a year, in decisions that were correct on the day the leadership team made them.

The gap opens in two distinct stages, with a different failure mode in each. The decision sits between them as the hinge.

Before the decision is the clarity gap. The understanding the call requires is scattered. The relevant email is in an inbox, the last decision on the subject is in a deck, the commitment someone made about it is in a meeting transcript nobody reads, and the assembled picture that would turn all of it into a basis for the next call exists nowhere. The executive rebuilds that picture by hand, under time pressure, before every decision that matters, and absorbs the cost of every reconstruction that comes up short.

The enemy in this stage is context blindness. The information exists somewhere. Nobody has assembled it. A chief of staff at an enterprise cybersecurity company gave us the cleanest statement of it I've heard: "Executives today don't suffer from lack of information. They suffer from fragmented context."

A chief customer officer at a business-process-automation company put a scene to it. He'd walked into a customer meeting that day without the background he meant to have: "I would have liked to have known more about that customer. I could have gotten it pretty easily. I could have probably been more specific with the CSM... but there was maybe miscommunication, so he didn't send it over. And so then I was like, hmm, looking up the person on LinkedIn while I'm on the call." The briefing existed and was easy to get. It just sat unbuilt, so he pieced together what he could in the seconds he had, live, in front of the customer.

A decision made on a thin picture is already partway into the gap before anybody leaves the room. The leader who can't reconstruct why the team made the last call will decide the next one on less.

After the decision is the execution gap. The leader makes the call, the room agrees, and the energy behind it starts draining that afternoon. Weeks later the decision is half executed and half forgotten, with pieces of it quietly reopened, and nobody can say exactly when it stopped being true.

The enemy in this stage is decision decay. A COO at a defense and aerospace technology firm described his version of it in three sentences. "It was a partnership. The decision had been made, but there was no follow-through to implement it. We went four days from decision to where the relationship owner picked up the ball." His company cared about that partnership, and the decision sat for four days with nobody driving it. Nothing went wrong in the meeting. Nothing went wrong with the choice. The decision lost altitude until somebody happened to notice it needed a driver.

Some decisions should die, and killing one isn't decay. The bigger corporate sin is often rigidity, holding a plan past the point where conditions justify it, and a reversed decision is still a decision, because somebody resolved it. Decision decay is what happens when nobody resolves it either way. Nobody executed the call and nobody killed it, so the organization keeps acting on something that no longer holds.

The two stages compound. A decision that decayed last quarter can leave its reasoning unavailable to the leader making the related call this quarter, which widens the clarity gap on the next decision, which produces a weaker call, which decays faster. Context blindness sets how wide the gap opens. Decision decay sets how long it stays open. The loss compounds in a direction most companies never look in, because each individual instance is small enough to absorb.

Follow-through is the word leaders reach for, though it names only the second half of the distance, the getting-done after the call. Every executive I talk to lands on some version of it within about ten minutes, and almost none of them has a system responsible for either half.

What Falls Into the Gap

Three things fall in, and each one fails in its own recognizable way.

Decisions have no system of record that fills itself. The other important objects in a company at least have a home the work is expected to reach. Customers live in the CRM, money lives in the general ledger, and code lives in version control, and each of those systems has a team whose job includes keeping it current. Decision logs exist, in the form of architecture decision records and board minutes, and each of them holds only the decisions somebody remembered to write down. Left to that, decisions live in a person's head, in a deck nobody reopens, or in a memo written by hand, filed once, and never read again. What the room decided survives as a one-line conclusion at best. What it ruled out, who disagreed, and what evidence carried the day are gone within the week. The organization keeps the answer and loses the reasoning, which is the part that would have let anyone judge later whether the answer still holds.

The partial-execution version of this is the one leaders describe most often. A CFO at a construction contractor told us about a decision his leadership team made and then watched drift: "One example is we decided to formally redo position descriptions with levels and pay bands, but so far only one department lead has completed the task. Incomplete banding has effects on HR and recruiting." One lead finished. The leadership team made that decision once, for everyone, and it landed in one place out of several. A system holding that decision would have known which leads were still outstanding, and would have said so in week two.

Commitments scatter and slip. A decision creates obligations, and those obligations spread across task systems, calendars, project trackers, and threads, in whatever form each owner happens to prefer. A promise has a counterparty, someone who notices when it doesn't arrive. A task carries an assignee, and it rarely carries that counterparty or the decision it came from. That difference is where follow-through breaks. Most of what a decision produces gets recorded, when anyone records it at all, in the form that keeps the assignee and loses both the counterparty and the reason.

A VP of sales at a technical staffing firm gave us the sharpest version of that failure. His team was building a proposal with two product specialists, and the work overlapped. "Pull through was delayed because we couldn't remember who committed to the overlapping duties. We all thought someone else committed to the part while we each worked on our own parts." Three capable people shared one piece of work, and no record showed who took it. The commitment existed. Ownership of it didn't survive the conversation that created it.

Goals disconnect from the work. The target lives in one system and the work meant to advance it lives in five others, so a person maintains the connection by hand, on a cadence nobody has time for. Leaders type goals in during planning, review them twice, and abandon them by the middle of the quarter. Nothing in the stack scores today's activity against the goal it was supposed to serve, so drift accumulates silently until a quarterly review makes it visible, months after it would have been cheap to correct.

In every case the organization decided something and then had no system accountable for whether the decision carried through.

Five Ways It Shows Up

The decision-to-done gap is invisible as a category, which is part of why it has gone unowned. Those three object-failures are the mechanism. Here is how they reach the executive's own week, as five distinct failures: three you can place on one side of the decision or the other, one that is the two stages compounding, and one that is the tax beneath them all. An executive recognizes every one of them on sight. Almost nobody connects them to a single cause.

Walking in cold is the clarity gap at the moment it bites. Every meeting starts with the leader catching up. The first ten minutes go to reassembling who these people are, what the group agreed last time, and what has moved since. A leader who walks in cold reacts to the room. A leader who walks in oriented shapes it. The difference compounds across a calendar with six meetings before noon, each one requiring a different context loaded from scratch in the gaps between them. The first essay called the cost of that manual reassembly the largest ongoing tax on senior leadership, and tax is the right word, because a tax is what you pay continuously and on schedule.

Follow-through gaps are what decision decay feels like from the leader's chair, and they cost differently. The leader delegates the work and it disappears. The commitment scatters into tools and calendars and threads, and nobody watches whether it happens. The leader finds out it didn't when the deadline is already gone, which is what makes this failure so expensive: the cost lands after the window to fix it has closed. Walking in cold is the tax you pay every week. This is the loss you discover once, too late, and can't recover. The scale of it shows up in the research on commitment reliability. Donald Sull and Charles Spinosa's work on promise-based management found that managers, asked what share of the commitments made to them they can actually rely on, typically answer about half. That number is self-reported, never measured, and it still means a leader planning a quarter is stacking work on promises the people making them treat as roughly coin flips. A CEO at a health-tech startup put the daily version of it plainly: "Even with senior hires, when a list of follow-up actions come, the ball gets dropped, which is frustrating."

Decision amnesia is what the two stages look like when they compound. A settled question comes back to the table three months later as though the first conversation never happened. The team relitigates what it already worked out, spends judgment it should have banked, and sometimes reverses a good decision because nobody can reconstruct why the room made it. The expensive part is rarely the hour in the room. The team makes the second decision with less context than the first, which means the organization is making progressively worse calls about the same question over time.

Priority drift is decision decay at the level of the quarter. The strategic work loses ground to the reactive work, one urgent interruption at a time. Each interruption looks harmless on its own, which is exactly how the damage accumulates without anyone deciding to allow it. A quarter that started with three priorities ends with the leader having spent their attention on forty things that felt urgent and a handful that mattered. The goals didn't change. Nothing in the stack held them steady against the daily tide of what felt pressing.

Cognitive load is the tax under the other four. It's the mental cost of holding the picture together by hand, plus the background hum of suspecting that something important is slipping and not knowing where. Leaders carry that load home. They describe it as stress and they blame themselves for it, because nobody has told them it's the predictable output of a system nobody built to carry a decision from the room to the result.

Those five failures resolve into two stages of one gap. An executive who reads that list recognizes every item, because every item is a Tuesday.

AI Widens the Gap

Every function in the company now has a tool that makes its work faster. Engineering has coding agents, sales has revenue intelligence, marketing has generation tools, and legal has research assistants. Each of those tools moves its function's work from doing toward deciding. The output surfaces more options and trade-offs than the people below can resolve, so the deciding escalates to the layer already carrying the decision load. More calls reach the top, faster, with shorter windows to choose between them. Every one of those additional calls enters the same interval where nothing holds it.

The volume of decided work has gone up. The layer accountable for carrying it forward hasn't appeared. The distance between those two lines is where the cost of the next few years accumulates, and it widens every quarter that a company adds capacity to decide without adding capacity to follow through.

The Symptoms Everyone Treats Instead

Now place those five failures against the problems the industry sells against. The industry sells against meeting overload, email volume, app switching, tool sprawl, and AI fatigue. Each has a product category built to relieve it, and each is real enough that the category sells.

Each is also downstream of the decision-to-done gap.

Consider meeting overload. A large share of the meetings on an executive's calendar exist to re-establish state that lapsed since the last meeting. The status meeting is there to answer a question a maintained system would answer on its own. Alignment meetings happen because the alignment from the previous decision started dissolving the moment that room emptied. The follow-up chases commitments no system was tracking. Cut those meetings without closing the gap and the executive loses the one mechanism that was narrowing it, then schedules two smaller meetings to make up for it.

App switching tells the same story. The executive opens seven tools before a single decision, because the record of what the company already decided, promised, and targeted is spread across all seven. The switching is reconstruction, performed manually, one tab at a time. A better-designed tool reduces the friction of each switch. It leaves the number of switches alone, because that number is set by how scattered the record is, and the record gets more scattered with every tool the company adopts.

AI fatigue is the newest symptom and the most revealing. Leaders are tired of AI tools that answer the question they typed and forget the conversation the moment it ends. The fatigue comes from a specific mismatch. The executive wants a system that already knows what the team decided last month and what's owed this week. The tool offers a system that responds to prompts. Every session starts over, so the executive hands the context into the prompt by hand, every time, which is the exact labor they were trying to escape. A tool you have to tell what the team decided can't be accountable for whether the decision carries through.

These failures persist because existing systems never carried responsibility for what happens to a decision, either before the room makes it or after. The market is saturated with tools. The market has no system that owns the interval. Every category sells a sharper instrument for capturing a slice of the work. Not one of them takes responsibility for driving the decided thing to completion.

Why Nobody Closed It

There's a tempting counterargument. If meeting overload, app switching, and AI fatigue are the symptoms, then a company that relieves enough of them eventually closes the gap underneath. Stack enough point solutions and the distance disappears on its own.

I ran enterprise software companies for two decades, and I watched that exact bet fail across category after category. It fails for a structural reason. A tool built to capture a slice of the work is architecturally committed to that slice. The CRM owns the deal record, the calendar owns the schedule, the document tool owns the artifact, the chat app owns the thread. The task manager owns a list of items that may not preserve the decisions that produced them. Each one is excellent at its slice, and none of them was built to carry a decision across all five tools, because carrying it was never any single vendor's job.

The management literature saw the problem long before the software industry did, and it kept reaching the same verdict: the failure is structural. Peter Drucker named it in 1967. A decision, he wrote, isn't made until carrying it out has become "someone's work assignment and responsibility." Until then, "there are only good intentions." Three decades later Jeffrey Pfeffer and Robert Sutton, comparing companies that turn knowledge into action against companies that can't, concluded that the difference "is not that one set of firms is populated by smarter, better, or nicer people than the other." What separated them was the systems and the day-to-day management practices. Donald Sull, Rebecca Homkes, and Charles Sull later put a number on where the structure breaks. In a study of nearly 8,000 managers, 84% said they could rely on their boss and their direct reports all or most of the time, and only 9% said they could always rely on colleagues in other units. The chain of command holds. The seams between units, where almost every consequential decision has to travel, are where reliance falls off a cliff.

Then look at what that same literature prescribed. Larry Bossidy and Ram Charan's mechanism for follow-through was Bossidy's own follow-up letter to each leader after a review, restating what the room agreed. Andy Grove's objectives and key results, carried from Intel to Google, are a list of goals someone rescores and resets by hand every quarter. Robert Kaplan and David Norton, needing something to bridge strategy and daily work, proposed an Office of Strategy Management, which solves a missing-layer problem by hiring people to be the missing layer. The 4 Disciplines of Execution names the enemy perfectly, "the real enemy of execution is your day job," and then asks a team to keep a scoreboard current forever.

The diagnosis is structural. The prescription is personal. When the prescription is personal, the failure lands on the person, whatever the diagnosis said.

Those authors weren't careless. No one could build the instrument they needed. Someone tried. Fernando Flores and Terry Winograd shipped The Coordinator in the mid-1980s, a system that modeled work as requests and promises and tracked each one to completion. It never caught on, because it made people classify their own sentences before they could send them, and a system that files your speech into categories reads as a system of discipline whichever way its makers sell it. One team kept using it only because management insisted; when a single frustrated user threw the software out of his office, the group dropped it within six weeks. The failure ran deeper than one product. Jonathan Grudin, cataloging why this whole generation of software fell over, found the reason was structural: the people who have to do the extra work of capture are rarely the ones who reap the benefit, so left to their own discretion, they don't do it. The capture had to happen without the speaker lifting a finger, and in that era that was impossible.

What changed is the reading. A system can now watch the work as it happens, the mail, the meetings, the documents, and form the decisions, commitments, and goals out of it without anyone stopping to classify a sentence. The capture Grudin called fatal, the kind that costs the person doing it, is the cost that finally dropped to zero.

The chief of staff is the clearest evidence of what happened next. Every company that hired one was correctly diagnosing a missing layer and paying a person to be it. The role is a human patch on a structural gap, and the gap keeps widening underneath the patch, because the work doesn't scale to one person. A system that carries that mechanical load turns the chief of staff from a stopgap into the one doing the judgment work no system can.

This Is Not a Performance Problem

The most damaging thing about the decision-to-done gap is how personal it feels.

When work you assigned goes quietly undone, you assume you should have followed up harder. When a decision you made comes back to the table for the third time, you assume you failed to communicate it. When you walk into a meeting unprepared, you assume you should have found the half hour. The whole experience arrives as a verdict on you. You work constantly, the work you delegated still isn't done, and the only explanation on offer is that you aren't on top of it.

That verdict is wrong, and it's wrong for a reason you can verify. Hand the same job to the most disciplined operator you know, give them the same stack, and they hit the same wall, because the wall is structural. The tools record activity. Not one of them owns whether the thing you decided actually happened. You've been doing, by hand and at the edge of your capacity, a job that should belong to a system. Doing it imperfectly is what carrying a system-sized responsibility with human-sized memory produces.

None of that is absolution. Some share of the failures in any organization is genuinely blameworthy, and a leader still owes those conversations. The trouble is that you can't find that share while you're mistaking the rest for it, and no organization has ever had the instrument that separates them.

The cost of the arrangement is real, and it's large enough to measure. In McKinsey's 2019 survey on decision-making, 61% of respondents said at least half the time they spend making decisions is ineffective. Most of the people who decide for a living, in other words, call half of that work wasted. Deciding is no sliver of the calendar either: McKinsey estimates that managers at a typical Fortune 500 company spend about 37% of their time on it. My own read is that a good share of the waste lands on both sides of the decision: in calls made on a picture the leader had to rebuild by hand, and in calls remade because the first one never arrived anywhere.

That cost stays off every line item, which is why it has gone unmeasured. It shows up as weeks of engineering spent on work a team already agreed to cut, a partnership that sat for four days with nobody driving it, a pay-band redesign that one department out of several finished, a quarter that drifted off its goals, and a capable leader who goes home convinced they're the problem. A later essay in this series puts a harder number on it, drawn from original research. This one names the cause honestly, so the number lands on the right target.

The Category That Owns It

Closing a problem this structural takes a system whose entire job is the responsibility every existing tool leaves on the floor.

That system is a category, and the category has a name. Executive Intelligence is the category of system responsible for making sure an executive's decisions, commitments, and goals carry through. The name sits one word from business intelligence, and the difference between them is the whole job: business intelligence reports on the state of the business, and Executive Intelligence owns whether the executive's decisions actually reach done. It sits above the work stack, one level up from every tool inside it. Decisions, commitments, and goals become first-class objects, formed from the work as it happens, with no template for anyone to remember to open. The system holds the picture together continuously, so the leader walks into the call with the context already assembled, and it keeps pressure on every commitment a decision creates until the work is done.

Closing one stage without the other leaves the gap open. A system that assembles context and then forgets what the room decided has closed the clarity gap and left the execution gap wide. A system that tracks commitments it was handed, without ever seeing the decision that produced them, has done the reverse. The decision is the object that crosses both stages, which is why one system has to hold it.

The Executive Intelligence Papers is a series of seven essays by the founders of Brief on the formation of a new category of software.