Why Accountability Faults Grow Back
You can redraw the lines. The culture underneath will blur them again. You can redraw the lines. The culture underneath will blur them again
Why Accountability Faults Grow Back
You can redraw the lines. The culture underneath will blur them again.
Organisations have fixed their accountability problem at least once. Someone drew a RACI. Ownership was assigned, decision rights were mapped, a governance forum was set up, and it held for a quarter or two. Then, slowly, the lines went soft again. Work drifted back to whoever would pick it up. Decisions needed more people in the room than the map said they should. And eventually, somebody proposed another clarification exercise, which everyone agreed was overdue.
This is the part of the accountability problem that gets the least attention, and it is the only part that matters. Not why accountability fails, that has been diagnosed thoroughly and correctly for decades, but why it fails again, in the same organisation, after being deliberately fixed.
The answer isn’t that the fix was badly executed. It’s that something underneath the architecture keeps rebuilding the fault.
What the fix actually addresses
Start with what a clarification exercise correctly identifies, because the diagnosis is usually sound. Accountability architecture breaks in three recognisable ways, and any competent redesign will find all three.
Ownership goes ambiguous, so that everyone feels responsible and no one is accountable. Work becomes communal property. Strong operators step in because they can, not because the structure asked them to, and over time their competence quietly becomes their obligation. The system reassigns responsibility without ever making the decision out loud.
Decision rights tangle, so that too many people can say no and too few can say yes. The organisation turns into a standing negotiation rather than a system, and speed collapses, not because people are slow, but because the structure makes speed impossible. Leaders read this as a capability gap and reach for better people, which never works because it never was a capability gap.
And incentives misalign, so that people are rewarded for avoiding risk rather than creating clarity. The structure teaches them, transaction by transaction, that ambiguity is less dangerous than ownership. So ambiguity wins, not because anyone chose it, but because it was the rational choice the architecture kept offering.
All three are real, all three are structural, and all three can be fixed on a whiteboard in an afternoon. Which raises the obvious question: if the diagnosis is this well understood and the remedy this available, why is every organisation on its third or fourth attempt?
The fault has a beneficiary
Because diffuse accountability is not a defect from everyone’s point of view. It protects people, and it protects them in ways that are entirely reasonable from where they stand.
Ambiguous ownership means no single person is exposed when something fails. A tangled decision right means nobody has to be the one who said yes. A structure where responsibility is communal is a structure where blame is communal too, which is a considerable comfort in an organisation where being wrong is expensive. The blur isn’t a failure to achieve clarity. It’s a functioning arrangement that delivers something people want.
And notice who it protects most. The clearer accountability becomes, the more exposed the people holding it are, and the people with the authority to redraw the lines are usually the people whose exposure increases most when the lines are drawn well. This isn’t cynicism about anyone’s motives. It’s the ordinary asymmetry: the cost of clarity is immediate and personal, the cost of ambiguity is deferred and shared. Anyone would find the second easier to live with.
So a clarification exercise asks an organisation to give up something it is quietly getting. The exercise wins the argument, because the argument is unanswerable. Of course, ownership should be explicit. Then, over the following months, the arrangement people were actually benefiting from reassembles itself, one reasonable exception at a time.
Culture is what keeps rebuilding it
This is the layer beneath the architecture, and it’s the one the redesign never reaches.
Culture, in any useful sense, is not the mood of a place or the values on a wall. It’s the organisation’s real answer to what pays off here: what gets rewarded, what gets tolerated, and what quietly costs you. And in an organisation where protection reliably pays off better than clarity, the architecture will keep regenerating ambiguity no matter how many times the map is redrawn. Not through anyone’s intent. Through a thousand small decisions that each make local sense: the exception granted, the owner not quite named, the decision taken to the wider group “for alignment,” the escalation nobody wanted to make.
Each of those is defensible on its own. Together, they are the culture reasserting its preference.
This is why the same organisation runs the same exercise every three years and is genuinely surprised each time. It has been treating the fault as an error to be corrected, when it is a tendency to be counteracted. Errors stay fixed. Tendencies grow back.
What it costs while it regenerates
The bill arrives in two places, and neither shows up as an accountability problem on any dashboard.
The first is capable people. They stop stepping in, because stepping in has silently become the job. They stop closing gaps they were never given the authority to close, and stop carrying weight the system was supposed to hold. What gets diagnosed as disengagement is usually something more specific: the moment a capable person stops subsidising a broken structure out of their own discretionary effort. They are not withdrawing. They are declining to keep paying for a design decision made by someone else.
The second is the standard itself. Decisions age out without being revisited. Mediocrity gets normalised because nobody owns the line that would have held it. And the organisation grows dependent on precisely the people it is steadily exhausting. A dependency that reads as strength right up until those people leave, at which point it turns out to have been the only thing holding the structure together.
What actually holds
None of this argues against the redesign. Ownership does need to be explicit, decision rights do need to sit where the knowledge is, and incentives do need to reward clarity rather than avoidance. That work is necessary and it is not what’s failing.
What’s failing is the assumption that the work is finished when the lines are drawn. If the fault regenerates because the culture beneath prefers blur, then two things must be true for a redesign to hold.
The first is that clarity has to stop being personally expensive. If naming an owner increases that person’s exposure without increasing their authority, they will find ways to un-name themselves, and so will their colleagues on their behalf. Accountability holds when the person who owns an outcome also has the means to produce it, meaning the redesign isn’t finished when responsibility is assigned, only when the matching authority is.
The second is that regeneration has to be monitored. Not audited annually, which is the same clarification exercise on a longer cycle, but noticed in the moment: the decision that drifted to the wider group, the owner who wasn’t named, the exception that made sense. Those are not administrative slips. They are the culture testing whether the new architecture will hold, and the answer it gets from the first few determines what happens to the rest.
Redesign the architecture, and you fix the current failure. Confront the preference underneath it – the quiet, entirely understandable inclination to keep responsibility blurred, because blurred is safer for whoever might otherwise be holding it – and you stop the failure from growing back.
The weak link was never the people.
And the moment competence becomes the patch, the architecture has already failed. The only question left is how long the patch holds.