The evidence file  /  Complexity leadership

Complexity leadership theory has been institutionalized, not validated.

It is widely cited, published in the leading journals, and taught as a mainstream framework. There is still no validated instrument behind it and no study showing that a complexity-informed intervention produces a measurable result. Both things are true, and the difference is worth money to you.


VerdictUseful vocabulary · thin evidence
  • What it describesA real and recognizable problem: organizations built for operational efficiency struggle to let anything new survive contact with the operational system.
  • Evidence for the partsStrong. Ambidexterity, network brokerage and closure, dynamic capabilities and tie strength each have substantial independent empirical support.
  • Evidence for the synthesisThin. The flagship publications are theoretical syntheses and trade articles. None of them tests anything.
  • Effectiveness evidenceNone located. No study establishes that a complexity-informed leadership intervention produces a measurable organizational outcome.
  • Validated instrumentNone in general use. There is no complexity-leadership questionnaire with published reliability data comparable to the PCQ.
  • Meta-analysisNot possible. There is no meta-analyzable quantitative body to pool.
  • How it is used hereAs a structured way of looking, labelled as such on the diagnostics page. Not as a measure, and never as a prediction.

What the theory claims

Uhl-Bien & Arena, 2018
The Leadership Quarterly, 29(1)

Complexity leadership theory holds that organizations run two systems with incompatible logics, and that the interesting question is what connects them.

Operational leadership works through the formal systems, structures and processes — selection, refinement, execution, efficiency. Entrepreneurial leadership creates the new knowledge, skills, products and processes that keep the firm viable. Enabling leadership creates, engages and protects the adaptive space between them.

Adaptive space is generated through two activities its authors call conflicting — deliberately holding the tension between the pressure to innovate and the pressure to produce — and connecting, linking people so that novelty can scale into the operational system rather than dying where it was invented.

The failure it explains is real: pilots that succeed and then have nowhere to go.

That is a genuinely useful description, and it is why the framework appears in this practice at all. Most executives recognize the pattern immediately — a portfolio of successful pilots and no change in how the company operates.

One terminology note, because older material circulates. The 2007 formulation used adaptive / administrative / enabling. The 2018 formulation uses entrepreneurial / operational / enabling. And “adaptive leadership” today more often refers to Heifetz's quite different framework. Sources that blur these are not describing the same thing.


What the research actually supports

The honest answer is that the strength sits one level down. The components of the framework are well evidenced, independently and over decades — organizational ambidexterity, dynamic capabilities, structural holes and brokerage, tie strength, semi-structures. That is where the rigor lives, and it is not a small thing.

The empirical complexity-leadership literature has also improved since 2011, when the theory's own authors noted that there were many theory papers and relatively few research papers, and could name only a handful of empirical exceptions — two of them computer simulations.

That specific complaint no longer holds. Post-2015 studies work in real organizations: agile coaches at Spotify, a Swedish municipality, consulting firms, a research institute. A scoping review counted fourteen empirical studies in healthcare alone.

But read what those studies do

They are small, qualitative, single-site and descriptive. The largest complexity-leadership-specific sample located is sixteen interviews. They document what enabling behaviours look like in practice. They do not test what those behaviours produce — and the healthcare review says so directly, noting that the studies do not assess effectiveness.

What it does not support

Two critiques sit unrebutted in flagship journals.

Neither is an outsider taking a swipe. The first was published in the same journal that launched the theory.

Rosenhead, Franco, Grint & Friedland (2019)

A structured review — around 3,000 records screened, 135 papers, detailed analysis of the ten most cited. Their finding on the evidence base is that much of the work is, in their words, abstract and untested.

On the prescriptions, they are blunter still: these are in effect motherhood statements — general, non-specific, and offered as though unchallengeable.

The part that matters most to anyone selling diagnostics: they observe that complexity leadership gives an extraordinarily limited role to tools for rational analysis, with much of the established quantitative toolkit ignored or dismissed. Their recommendation is not to abandon the theory but to pair it with analytic method — which is, more or less exactly, how it is used here.

Tourish (2019)

A different charge, and the one that bites a consultant hardest. Complexity leadership, he argues, offers a theory of complex organizations led by non-complex leaders. It declares that leadership outcomes emerge unpredictably from interaction, then grants enabling leaders a capacity for purposeful influence that its own ontology does not permit.

His warning is that, handled carelessly, complexity vocabulary becomes a new mask for old theories that legitimize imbalanced power at work.

If outcomes cannot be predicted, what exactly is a client paying for?

That question deserves a straight answer rather than a deflection. You are paying for diagnosis, for structured attention to a problem the organization has been managing by feel, and for someone willing to hold a tension that everyone else in the room is incentivised to resolve prematurely. You are not paying for a predicted outcome, and anyone offering you one from this literature is selling something the literature does not contain.

Claims to distrust

If a vendor says any of this, ask for the citation

  • “Research-backed” or “validated” complexity leadership interventionsThat literature does not exist. The framework is well established; the intervention evidence is not. These are different claims.
  • Any performance percentageFigures such as “37% more efficient” and a “25% improvement in average handle time” circulate widely. They trace to a practitioner magazine with no reported design, sample or comparison group. They are not findings.
  • “Uhl-Bien's research shows that…” followed by a causal claimHer central post-2015 output on this topic is synthesis and commentary. The accurate verb is argues or proposes, not shows.
  • A predicted or guaranteed outcomeThis contradicts the theory's own epistemology. Its authors are explicit that complexity approaches cannot predict outcomes.
  • That the framework is settled consensusTwo substantial critiques sit in flagship journals without published rebuttal.
  • A complexity leadership “assessment” sold as a validated instrumentNo such instrument is in general use. A structured assessment is a reasonable thing to offer. Calling it a validated measure is not.

How it is used in this practice

As a structured assessment, described that way on the diagnostics page and never as a questionnaire. Three tests, applied by interview and document review: whether the new way of working has resources that do not compete with a quarterly commitment, whether a named person holds authority across a functional boundary, and whether what is learned reaches the operational system through anything other than a status deck.

Those three come straight from the framework. They are useful. They are also a disciplined way of looking rather than a measurement, and pretending otherwise would fail the standard this page exists to apply.

The two measures used alongside it — psychological capital and organizational network analysis — are validated instruments with published reliability data. Pairing a structured assessment with real measures is exactly the pluralism Rosenhead and colleagues recommended.

References

Everything above, checkable

Rosenhead, J., Franco, L. A., Grint, K., & Friedland, B. (2019). Complexity theory and leadership practice: A review, a critique, and some recommendations. The Leadership Quarterly, 30(5).

Tourish, D. (2019). Is complexity leadership theory complex enough? A critical appraisal, some modifications and suggestions for further research. Organization Studies, 40(2), 219–238.

Uhl-Bien, M., & Arena, M. (2018). Leadership for organizational adaptability: A theoretical synthesis and integrative framework. The Leadership Quarterly, 29(1), 89–104.

Uhl-Bien, M., & Arena, M. (2017). Complexity leadership: Enabling people and organizations for adaptability. Organizational Dynamics, 46(1), 9–20.

Uhl-Bien, M., Marion, R., & McKelvey, B. (2007). Complexity leadership theory: Shifting leadership from the industrial age to the knowledge era. The Leadership Quarterly, 18(4).

Bäcklander, G. (2019). Doing complexity leadership theory: How agile coaches at Spotify practise enabling leadership. Creativity and Innovation Management, 28(1), 42–60.

Schulze, J. H., & Pinkow, F. (2020). Leadership for organisational adaptability: How enabling leaders create adaptive space. Administrative Sciences, 10(3), 37.

Belrhiti, Z., Nebot Giralt, A., & Marchal, B. (2018). Complex leadership in healthcare: A scoping review. International Journal of Health Policy and Management.

Bonini, A., Panari, C., Caricati, L., & Mariani, M. G. (2024). Leadership and adaptive performance: A meta-analysis. PLOS ONE.

Reviewed September 2026. Where a source was read via abstract or repository record rather than full text, no page locator is given here. If something on this page is wrong, write and say so — corrections are published.


Read next

Related entries in the evidence file

Next

This is one page of the evidence file

Every construct this practice uses gets the same treatment, including the ones that make it look good and the ones that do not.