The Frozen Middle Is a Symptom, Not a Cause
We have built a system in which the same layer is unseen in success and blamed in failure, and then we are surprised that it behaves defensively.
The manager everyone agreed was the problem
Every stalled transformation I have sat inside eventually produces the same moment. It arrives late, usually in a closed session after the formal review, once the slides are down and the sponsors are tired. Someone — often the most senior person in the room — says a version of the same sentence: the strategy is sound, the technology works, the business case still holds; the problem is the layer in the middle. Heads nod. The frozen middle. The permafrost. The clay layer that the warmth of the vision somehow cannot melt. A name is found, a workstream is chartered to “drive adoption and address resistance,” and the meeting breaks up faintly relieved, because a problem with a name feels like a problem half-solved.
I have nodded along to that sentence more times than I would now care to admit. I have stopped. Not because the middle is innocent, and not because transformations do not stall there — they do, reliably, which is exactly what makes the diagnosis so tempting. I stopped because the story we tell about why they stall there is almost always wrong, and wrong in a particular way: it points the enquiry downward, away from the people who designed the programme, and it guarantees that the next one will freeze in precisely the same place.
A diagnosis of convenience
The appeal of the frozen-middle story has never been that it is well supported. It is that it is comfortable for everyone senior enough to commission a transformation in the first place. It locates the fault one layer below the point of decision. It converts a structural problem — how the change was designed, sequenced and resourced — into a behavioural one, a matter of mindset and will. And behavioural problems are wonderfully outsourceable: they can be handed to a change-management workstream, a communications plan, a set of “leadership behaviours” cascaded down the line. None of that requires the design to be reopened.
This is why the diagnosis survives its own failure rate. When the adoption workstream does not fix the “resistance,” the conclusion is rarely that the diagnosis was mistaken. It is that the middle was even more frozen than feared. The story is unfalsifiable by construction, and an unfalsifiable story is not an analysis. It is an alibi.
What the middle is actually for
To see why the blame lands in the wrong place, it helps to be precise about what that layer does. A transformation is authored in the language of outcomes — synergies, a target operating model, improved service levels, a cost line that comes down. It has to be executed in the language of Tuesday: who covers the desk while two people are in training, which customer gets called back today, what breaks if we cut over this process this week rather than next. Those are two genuinely different languages, and the middle manager is the only person in the organisation who is fluent in both. The executive holds the intent but not the operational detail; the front line holds the detail but not the intent. The middle holds both at once, and spends its days converting one into the other.
A transformation is written in the language of outcomes and executed in the language of Tuesday. The middle is the only place those two languages are reconciled — which is precisely why, when the reconciliation fails, the middle is where the failure first becomes visible. Visible is not the same as responsible.
Call it the translation layer. When it works, it is invisible; strategy simply appears to become operational reality, and the credit flows upward. When it fails, it is the most visible thing in the building, because everything above it and below it seizes at once. We have built a system in which the same layer is unseen in success and blamed in failure, and then we are surprised that it behaves defensively.
The version of the story that is true
I want to be honest about the other side of this, because the frozen-middle narrative did not appear out of nothing, and a perspective that only flatters the middle would be as useless as the one that only blames it.
Some middle managers genuinely do obstruct. I have watched a capable, well-regarded manager quietly starve an initiative he had privately decided was foolish — attending every meeting, agreeing every action, and ensuring through a hundred small omissions that nothing actually moved. Information gets hoarded because information is leverage. Headcount gets protected because status and security are measured in headcount. A director who will rotate to another division in eighteen months can, with patience, simply be outlasted. This is real, it is sometimes cynical, and pretending otherwise would insult everyone who has had to manage around it.
The steelman goes further than individual cases. Perhaps the layer really is, on average, the conservative one. It sits closest to the operational risk. It is personally exposed if service falls over on a Monday morning. It has been rewarded for years — in bonus, in reputation, in a quiet life — for keeping things stable. On that reading, caution is simply what that position selects for, and what looks like resistance from the top is only prudence seen from above.
I take that seriously. And it points to the test that settles the matter. If resistance were a property of the people — a mindset, a generational trait, a failure of calibre — then it would persist when the circumstances changed. It does not. Change the span of control, change what the manager is measured on, give them real authority over sequencing, and the very same managers who were “frozen” last quarter become the engine of the change this quarter. A characteristic that dissolves the moment you alter the design was never a characteristic of the person. Genuine obstruction exists, and it is a minority of what we see. The bulk of what we label resistance is something duller and more damning: a rational response to an incentive structure we built, and then blamed people for inhabiting.
How we manufacture the bottleneck
If the middle is not frozen by nature, how do we reliably freeze it? Three mechanisms do most of the work, and they are all decisions made well above the layer that gets the blame.
- Initiative collision. We count change at the portfolio level, where it looks manageable, and never at the point where it actually lands.
- Accountability without authority. We make the manager answerable for adoption while removing every lever that would let them shape how the change arrives.
- Measuring stability while paying to destroy it. We keep the manager’s targets pinned to the service continuity that the transformation, by design, disrupts.
The first is the one I would put a number against, because it is the one leaders almost never see. Consider a single operational team — twelve people, one manager. Across one quarter, that manager is named the local lead for seven concurrent initiatives: a system migration, a redesign of the service model, a cost-reduction target, an offshoring transition, two process-standardisation efforts run out of different functions, and a compliance remediation. Each has a different director as sponsor. Each has its own steering meeting, its own status report, its own definition of “on track.” And each business case quietly assumes that it has first call on the same two or three genuinely capable people on that team — the ones who can actually be trusted with a change.
No one ever decided that this team should absorb seven simultaneous changes. It is simply the sum of seven separate, individually reasonable decisions, none of which looked at the others. The manager’s real job — unassigned, unrewarded, and invisible on every plan — becomes triage: deciding which director’s initiative quietly slips this month so that the desk stays covered and the other six do not collapse. From inside any one of those seven steering committees, that triage is indistinguishable from resistance. The director sees only that their initiative is the one being slowed, by this manager, again.
Now the before-and-after. When one organisation stopped counting change at the portfolio level and started counting it at the point of collision — asking not “how many initiatives are in flight” but “how many concurrent changes are landing on each front-line team” — the number that mattered turned out not to be the fifty-odd in the portfolio. It was the seven hitting that team. They cut it to two, sequenced them, and gave the manager explicit authority to hold the queue and refuse a third. The adoption problem for those two initiatives ran ahead of schedule. Nothing about the managers had changed. The load had, and the authority to manage it had.
| The blame diagnosis | The design diagnosis |
|---|---|
| What is wrong: the people in the middle | What is wrong: the load and the authority we handed them |
| Who owns it: the managers, individually | Who owns it: whoever shapes the portfolio and the incentives |
| The intervention: drive adoption, address mindset | The intervention: sequence the change, match authority to accountability |
| Next time: it freezes in the same place | Next time: the layer becomes the instrument of delivery |
What follows if we take this seriously
The point of retiring the frozen-middle story is not to replace one slogan (“the middle is the blocker”) with a warmer one (“invest in your middle managers”). Sloganising is how we got here. The point is to move the enquiry to where the leverage actually is.
Three disciplines follow directly. First, count change where it collides, not where it is commissioned; the only initiative number that predicts delivery is the one landing on a single team. Second, make authority track accountability — if a manager is answerable for whether a change sticks, they must hold real discretion over how and when it arrives, including the standing to say “not yet.” Third, stop measuring people on the stability you are paying to disrupt; a manager scored on this quarter’s service continuity will protect it, and should not then be called frozen for doing the job we set them.
“Resistance that dissolves the moment you change the design was never resistance. It was arithmetic.”
None of this is soft. It is more demanding than the blame diagnosis, not less, because it puts the work back on the people who hold the design — the sponsors, the portfolio, the incentive structure — rather than dispatching it downward to a workstream. That is exactly why the frozen-middle story endures: it is the cheaper of the two explanations, and it always will be.
So the next time the room goes quiet and someone reaches for the sentence about the layer in the middle, the more useful move is not to nod. It is to ask a plainer question. What, precisely, did we ask this layer to hold — how many changes at once, with how much authority to shape them, measured against what? The answer is almost never “the middle is frozen.” The answer is usually that we designed the freeze, chartered it across seven steering committees, and then looked for somewhere lower down to put the blame.