Why does strategy execution break between departments rather than between levels?
The short answer
Strategy execution breaks at the handovers between units, because commitments are binding within a unit and treated as voluntary between units. In one development program studied, 39 percent of technical dependencies had no corresponding coordination between the teams. Across unit boundaries, work is not misprioritized — it is not even noticed.
Two division heads, one status meeting, and both are right. One says he has been waiting six weeks for the input. The other says no request ever reached him. Neither is lying.
For a long time I tried to resolve such meetings with more clarity from above — sharper priorities, tighter reporting, unambiguous responsibilities. It did not change the situation, because I was working on the wrong axis. Both units knew exactly what they were supposed to do. What was missing lay between them.
Commitments hold within a unit and are voluntary between units
Donald Sull and Charles Spinosa put this difference into one sentence in the Harvard Business Review in 2007: requests and commitments are honored within a unit and treated as optional across unit boundaries.
This is not a question of character. Along the reporting line, a procedure exists: whoever breaks a commitment has to deal with their manager, and before the next conversation about their own pay. Sideways, this procedure does not exist. In the survey by Sull, Homkes and Sull of 7,600 managers, around 80 percent say that poor cooperation between units is not reliably addressed. A commitment without consequence is a declaration of intent.
The commitment breaks on ignorance, not unwillingness
Only about a quarter of managers say their own goals are known to those responsible in other units (MIT Sloan Management Review 2018, more than 400 organizations).
That changes the diagnosis fundamentally. Whoever does not know what the neighboring unit is measured on this quarter cannot plan for its needs. That is not negligence, it is a lack of information. The two division heads in the status meeting do not have a conflict problem. They never agreed on their respective obligations, because nobody created the occasion for it.
Interestingly, those affected answer this differently depending on how they are asked. Asked about the cause, managers more often name alignment with the strategy. Asked what they concretely lack, they ask for cross-unit coordination twice as often as for a better goal cascade.
Across the unit boundary, work is not misprioritized but not seen
The hardest finding on this topic does not come from a management survey but from an aircraft engine. Manuel Sosa, Steven Eppinger and Craig Rowles recorded 569 technical dependencies between components in a development program and compared them with the actual coordination between the responsible teams (Management Science 2004).
For 39 percent of the dependencies there was no corresponding coordination. The real point lies in the detail: within an organizational unit, teams coordinated more closely the stronger their technical dependency. Across unit boundaries, the strength of the dependency was statistically irrelevant. Coordination there is not weaker. It happens regardless of how much depends on it.
That organizational structure reaches this deep into the result also shows elsewhere. In a study of 3,404 program components of a Windows version, characteristics of the organization — number of units involved, distribution across reporting lines — predicted defects better than any property of the code itself.
What is not reported gets worked around
The obvious assumption is: if it does not work sideways, someone escalates upwards. There is hardly any robust research on this, and the strongest field finding points the other way.
Anita Tucker and Amy Edmondson observed in hospital operations how nurses deal with disruptions that originate in other departments. 93 percent of the problems were worked around locally, 7 percent were reported back to the source. The workaround cost a good half hour per shift on average — time that appears in no plan. Transferring this to a management team is an analogy, not evidence — the mechanism is familiar nonetheless: working around is faster than clarifying, and nobody notices.
That is exactly why a portfolio looks healthy while the handovers between units stand still. The extra work of working around appears in no status report. It appears as utilization.
Whoever intervenes from above at this point only shifts the problem. Sull and colleagues describe exactly this relationship: interventions by top management in cross-unit questions lead to more frequent escalation in the future, because those involved learn that clarification comes from above. The sideways procedure never develops that way.
A commitment between units needs an amount and a date
The integration question is old. As early as 1967, Lawrence and Lorsch showed in ten companies that highly differentiated organizations need their own mechanisms to come back together. A replication of their measurement later failed; the idea has held.
What works in practice is narrower than an integration concept. For every measure that needs a second unit, three items belong in the decision paper: who delivers what to whom, by when, and what contribution the measure is meant to make to the goal. The third item is the uncomfortable one, and it is the effective one — it turns a request into a commitment with a counter-value.
This is what BizzPlAI is built for as a Strategy Execution System. The expected value contribution of a measure remains retrievable from the decision onward, together with the goal it contributes to. That makes one unit’s commitment to another traceable without anyone having to escalate it. It does not make the decision on priorities for you.
One more thing belongs to honesty: a survey that cleanly measures horizontal against vertical reliability does not exist to this day. For eleven years, nobody has independently replicated the finding. For a problem every management team knows from status meetings, that is a remarkable state of affairs.
A test for your next meeting: take the three measures that need a second unit. Ask the delivering unit what contribution the measure is meant to make. If only the deadline is known there, you do not have a commitment. You have a request.
Common follow-up questions
Does execution fail more at the hierarchy or at unit boundaries?
Reliably comparable figures for both axes do not exist. What can be shown is the responsibility gap sideways: around 80 percent of 7,600 managers surveyed say that poor cooperation between units is not reliably addressed (Sull, Homkes and Sull 2015). For commitments along the line, by contrast, there is a procedure with consequences.
Why are commitments between units broken?
Usually not out of unwillingness. Only about a quarter of managers say their own goals are known to those responsible in other units. Whoever does not know what the neighbor is planning cannot plan for it either.
Does it help to escalate conflicts between units upwards?
There is hardly any robust research on this, and the strongest field finding points the other way: problems are mostly worked around locally rather than reported. Escalation is therefore rarely the practice, but rather the exception that makes do for a missing procedure.
Sources
- Promise-Based Management: The Essence of Execution, Harvard Business ReviewDonald Sull (London Business School) and Charles Spinosa, 2007
- Why Strategy Execution Unravels — and What to Do About It, Harvard Business ReviewDonald Sull (MIT Sloan), Rebecca Homkes (London Business School), Charles Sull, 2015
- The Strategic Agility Project — survey in more than 400 organizationsMIT Sloan Management Review, Donald Sull and Charles Sull, 2018
- The Misalignment of Product Architecture and Organizational Structure in Complex Product Development, Management Science 50(12)Manuel E. Sosa, Steven D. Eppinger and Craig M. Rowles, 2004
- Why Hospitals Don't Learn from Failures, California Management Review 45(2)Anita L. Tucker and Amy C. Edmondson, 2003
- The Influence of Organizational Structure on Software Quality, Proceedings ICSE 2008 — 3,404 binariesNachiappan Nagappan, Brendan Murphy and Victor Basili, 2008
- Organization and Environment: Managing Differentiation and Integration — field study in ten companiesPaul R. Lawrence and Jay W. Lorsch, Harvard Business School, 1967
See how strategy arrives.
In 30 minutes, we show you how to steer your initiatives instead of merely managing them.