Why most turnarounds fail, and why an honest instrument needs a category the repair industry doesn’t have. The Great Homecoming research programme · June 2026
When a system is failing — a company, an institution, a public body — the universal instinct is to intervene: new leadership, a restructuring, fresh capital, a culture programme. We treat repair instead as a matching problem with hard gates. The repair must match the way the system actually failed, not the way failures are usually fixed. The system’s stage of hardening decides which repairs can still land at all. Who may apply the repair is usually decided one level above the system, not inside it. And order matters: restore what the system is for before rebuilding what it can do. Most distinctively: some systems are past repair, and an honest diagnosis says so out loud. Turnaround practice has no category for that call — every diagnosis it produces ends in an engagement. That missing category, we hold, is what separates diagnosis from sales.
Look at the structure of the turnaround industry rather than its results, and one feature stands out: the people who diagnose failing organisations are the people paid to repair them. The incentive does not need to corrupt anyone consciously; it only needs to shape the menu. A diagnostic vocabulary in which every condition is treatable will, under that incentive, emerge on its own — and it has: rich taxonomies of what went wrong and which levers to pull, and almost nothing on the question that should come first: is this system still reachable by any lever at all?
We make that the first question. Repairability is a property of the system being repaired — it varies, it declines, and at a certain point it reaches zero. An instrument that cannot return that reading is not measuring; it is marketing.
Systems do not fail in one generic way, so there is no generic fix. In our framework a system goes blind by one of three routes (each set out in the companion piece, “The Three Ways Systems Go Blind”). In the first, its incoming information is bent: bad news is softened and metrics are gamed until the dashboard is full, current, and wrong. In the second, the system seals — it stops taking in the outside and consults only itself, so every decision runs on internal input. In the third, the system’s own structure becomes the screen: the org chart and the processes are all dutifully followed, every box reports green, and the failure lives in the gap that no box owns.
Each route calls for a different repair. A bent signal needs the channels that carry bad news re-opened. A sealed system needs outside reality forced back into its deliberations. A structural veil needs the purpose the structure was built to serve renewed, and the hollowed mechanisms rebuilt.
A repair aimed at the wrong route is not merely wasted — it is consumed by the condition. Hand a reorganisation to a leadership that has sealed itself in, and the reorganisation becomes one more instrument of the sealing: the boxes move, the loyalists land on top. The intervention’s energy is real; the condition metabolises it.
Whichever route the blindness enters by, it hardens in stages, and the stages gate the repairs. The mechanism is simple: every error a system declines to correct leaves a residue, and the residue weakens its response to the next correction. Correction capacity is not a constant; it is spent.
Early, the residue is thin: bad news still lands, and ordinary good management still works — this is where the turnaround industry’s genuine success stories live. In the middle stage the system half-knows: corrections are acknowledged and then dissolved, and repair now needs force behind it — changed incentives, outside adjudication, consequences that do not depend on the patient’s cooperation. Whether repair reliably works here is exactly what our forward tests are for. At the late stage the loop is closed: correction attempts are absorbed without effect, reinterpreted into confirmation of what the system already believed. The same intervention that saves a system at the first stage is shrugged off at the last. The intervention did not change. The reachability did.
Most repair that works does not come from inside the failing system, and we hold that this is structural: a system whose correction loop is damaged cannot use that loop to fix itself. The repair has to be applied from the level above — the owner, the court, the regulator, the parent institution; for a person, the family or community around them. Two conditions follow. The level above must itself be healthy, or its intervention inherits the same blindness. And it must adjudicate, not appease: a rescue that re-funds the system without ruling on the underlying condition extends the runway and thickens the mask.
This carries one hard asymmetry. Systems at the top of their hierarchy — a dominant institution, a whole civilisation — have no level above to reach in and adjudicate. For them only self-renewal remains: re-opening their own correction loop from the inside.
That is not a mystical act, and there is one opening that keeps it possible — a leader who can integrate. The self-renewal of a system at the top of its hierarchy almost always runs through such a person: the kind described in the companion piece, “Choosing Leaders Who Can Integrate,” who re-opens the channels that carry bad news, forces outside reality back in, and re-points the system at what it was for. A system with no level above is not without an exit; its exit is internal, and a leader is usually how it is taken.
But the opening is conditional, and that is the point. Self-renewal is recorded in history as rare and never accidental, and it is available only from a stage where the loop is still partly open. Even an extraordinary leader cannot adjudicate a system that has already sealed past reach — the same intervention that would have renewed it earlier is now absorbed and reinterpreted like any other. The agent of renewal exists; the window in which they can act expires.
There is also an ordering rule, and it is the one funders most often get backwards: restore what the system is oriented toward before rebuilding what it can do. Capability rebuilt on a misoriented system produces a more capable wrong thing — better execution of the behaviour that was killing it. Fresh capital injected into a hollow organisation does not buy life; it buys runway, and runway spent masked makes the eventual reckoning larger (the mechanism is in “The Two Clocks”). First the direction, then the muscle. Reverse the order and the repair funds the disease.
If a system reads as late-stage — corrections absorbed without effect, the loop closed — the honest finding is that it is past repair: no available intervention will still reach it. That finding does not end the work; it changes what the work is. Three kinds of real work remain, and none of them is repair.
Make the decline visible. Put on the record, in advance, what the dashboards are hiding — a clear reading of the true condition while the surface still looks healthy. This will not save the sealed system. But it tells everyone still depending on it where things actually stand, and it can move the level above to act while acting still helps.
Save what can be saved. Carry out the parts the condition has not yet reached — the people, the knowledge, the units that still work. This is triage, not disloyalty: protecting what is sound before the failure spreads into it.
Prepare what comes next. In our framework a collapse is never simple death; it is reorganisation at a lower level — the system reconstitutes in some smaller or simpler form. What that successor looks like depends heavily on what was preserved and thought through before the end, which makes preparing the rebuild present-tense work, not a post-mortem.
None of these three is something a firm paid to sell a turnaround can offer — there is no fee at the end of them. An instrument paid for the honest reading rather than the cure can. That, more than any single technique, is the difference we are claiming.
The Great Homecoming is an independent research programme on why systems cohere or fragment. Companion pieces: “Why Systems Don’t Die From Shocks,” “The Two Clocks,” “The Three Ways Systems Go Blind.” The instrument described is under live forward test; we describe it that way until its register adjudicates. Contact: Wim Van Laere.