Only 55% of middle managers can name one of their company's top five priorities. Strategy does not fail in the document. It fails at the handover.

First, clear away the folklore

The claim that 70 percent, or 90 percent, of strategies fail is repeated in books, keynotes and proposals, usually without a source, occasionally with a source that itself has no source. The most careful examination of the question is by Cândido and Santos in the Journal of Management & Organization (2015). They traced the published failure-rate estimates and found them ranging from 28 percent to 90 percent, resting on evidence they describe as outdated, fragmentary, fragile or just absent. Their conclusion: 90 percent is the upper limit of the estimates, and likely an overestimation. Survey houses have produced their own figures, Bridges Business Consultancy's 2016 implementation survey put failure at 67 percent, but the honest summary is that nobody has a reliable single number, and anyone quoting one with confidence is selling something.

That does not mean the problem is imaginary. It means the interesting question is not how often strategies fail but where.

What the evidence actually shows

The most substantial study of execution remains the Harvard Business Review research by Donald Sull, Rebecca Homkes and Charles Sull (March 2015), built on surveys of nearly 8,000 managers in more than 250 companies over nine years. Its findings are specific rather than dramatic, and they all point at the same gap.

Only 55 percent of middle managers could name even one of their company's top five priorities. Read that again: in companies that had a strategy, nearly half of the layer responsible for delivering it could not name any part of it. Execution also turned out to be horizontal, not vertical: 84 percent of managers said they could rely on their boss and direct reports all the time, but only 9 percent said the same of colleagues in other functions, and 30 percent named failure to coordinate across units as the single greatest execution challenge. The chain of command works. The handoffs between departments, where any real strategy actually lives, do not.

The same research explains why plans age so badly once handed over. Eighty percent of managers said their companies fail to exit declining businesses or kill failing initiatives quickly enough, and only 20 percent said their organisation is good at shifting people across units to support strategic priorities. A strategy is a set of resource movements, and the organisations meant to execute one mostly report that they cannot move resources. The plan assumes agility that the delivery machine does not have, and nobody is left in the room to renegotiate the plan against reality.

None of this is a document problem. Analysis can be right, the market read can be right, the plan can be internally coherent, and the strategy still dies in the space between the report and the Monday-morning decisions of people who were not in the room when it was written.

Why the handover is the weak point

A strategy document compresses months of thinking into fifty pages, and the compression is lossy. The reasoning behind each choice, the options rejected and why, the assumptions that would change the answer if they broke: most of that lives in the heads of the people who did the work. When they leave at the handover, the organisation keeps the conclusions and loses the reasoning. Six weeks later reality diverges from an assumption, as it always does, and no one left in the building knows which parts of the plan flex and which are load-bearing.

The second weakness is accountability. The authors of a plan that stops at the report are accountable for the quality of the document, which is assessed on the day it is delivered, before any of it has been tested. Every incentive points toward an impressive document rather than a workable one. And the cross-functional coordination that the HBR data identifies as the binding constraint belongs to nobody: each function executes its chapter, and the strategy fails in the seams between chapters.

What changes when the authors stay

We built Silvengate on the other model: the people who do the diagnostic and write the plan stay to run it. Three things change, and each maps to a failure mode in the data.

The reasoning stays in the building

When an assumption breaks, the people who made it are present, recognise it early and adjust the plan rather than watching it be quietly abandoned. The plan becomes a living argument instead of a fixed artefact.

Accountability moves from the document to the number

A team that must execute its own recommendations writes different recommendations: fewer, sequenced, sized to the organisation's real capacity, each with a measurable claim attached. The temptation to recommend everything disappears the moment you have to do everything.

The seams get an owner

Cross-functional coordination, the 9 percent problem, stops being everyone's job and nobody's. Someone whose only role is the strategy sits across the functions, holds the sequence, and chases the handoffs that the org chart does not cover.

What to demand of any strategy work

Whoever you work with, hold the work to the standard the evidence implies. Ask how the plan reaches the 45 percent of managers who will otherwise never absorb it. Ask who owns the cross-functional handoffs by name. Ask which assumptions, if broken, change the answer, and who is watching them. And ask what the authors are accountable for after delivery day. A plan whose writers have reason to care what the numbers say in month six is a different, and shorter, document than one assessed on the day it is handed over.