Only 11% of managers believe all their company's strategic priorities have the financial and human resources needed for success.
Why it matters when the plan changes
Strategy documents are written as if understanding were the hard part. It rarely is. The hard part is that five specific things must be true at once: people know which strategy applies, what to choose and drop, what their own contribution is, who decides and owns the outcome, and which interfaces the work runs through. Each is observable, and each absence can be forecast.
The tension is that the five are not equally weak. Direction is usually the best served, because it is what leaders spend their communication on. Priority is often thinner: Sull, Homkes and Charles Sull's 2015 Harvard Business Review research found only 11% of managers believe all their company's priorities carry the resources needed. Dependency is usually the worst, since it sits between owners and belongs to nobody; the same research found only 9% can rely on colleagues outside their own unit all the time. Averaging hides what matters, that a plan fails at its weakest clarity, not its mean.
In practice
A leadership team scores itself well on direction after a strong town hall and treats the translation as finished. Asked separately which two of eleven initiatives they would drop, no two members give the same answer. Direction was clear and priority was not. The plan proceeded on the strength of the clarity that was easiest to demonstrate.
Evidence
Most managers do not believe their stated strategic priorities are actually resourced.
Donald Sull, Rebecca Homkes and Charles Sull, Why Strategy Execution Unravels and What to Do About It, Harvard Business Review (2015)Reliance across units is far lower than reliance within a line, which is why dependency is the clarity that fails first.
Donald Sull, Rebecca Homkes and Charles Sull, Why Strategy Execution Unravels and What to Do About It, Harvard Business Review (2015)
What it cannot tell you
The five clarities describe what a plan needs from the organisation, not whether the underlying strategy is sound. A plan can score clear on all five and still fail because the choice itself was wrong. The model also treats each clarity as present or absent; it does not measure degree once a basic threshold of clarity has been met.
How Atlas reads it
The five clarities are the questions put to the organisational read, not a score returned to a leader. They map onto the layers the organisation is read in, and the weakest of the five sets the rating for the pillar it belongs to. Where signal on one of them is thin, that clarity is marked blind rather than estimated.
Questions
Direction, priority, role, ownership and dependency. Direction is knowing which strategy applies now. Priority is knowing what to choose and what to drop. Role is knowing your own contribution. Ownership is knowing who decides and who is answerable. Dependency is knowing which interfaces the work runs through.
Because research such as Donald Sull, Rebecca Homkes and Charles Sull's 2015 Harvard Business Review study shows specific, measurable gaps rather than general organisational fog. Adding clarities beyond the five that cannot be observed turns the model into a checklist. The point is a short list that predicts where a plan stalls, not coverage of everything organisational.
Dependency. Direction, priority and role each have an owner somewhere in the reporting line, but an interface sits between two owners and belongs to neither. Sull, Homkes and Charles Sull's 2015 Harvard Business Review research found only 9% of managers can rely on colleagues outside their own unit all the time.
Not reliably. The five are not additive, because the plan runs through all of them in sequence. A clear direction with unclear ownership produces confident movement in several directions at once, which is often more expensive than movement that has not started.
By reading the plan against the organisation rather than by asking people to rate themselves. Documents, decision records, roles and observed operating patterns are read together, and where the signal for one clarity is insufficient it is marked blind instead of estimated.