Kristian Gustafson: Structured intelligence analysis for the modern military

Intelligence analysis has always demanded disciplined thinking, yet for much of its history it has relied heavily on intuition, personal judgement, and the craft knowledge of experienced analysts. [1] As the modern security environment has grown (arguably) more complex—characterised by ambiguous indicators, rapid tempo, and deliberate adversarial deception—the limits of intuition have become increasingly evident. Contemporary military decision-making requires assessments that are transparent, defensible, and able to withstand both scrutiny and uncertainty. The strongest argument for structured analytical techniques (SATs) is therefore straightforward: they impose rigour, reduce avoidable error, and provide commanders with a clear understanding of how an assessment was reached. In an environment where decisions carry operational and strategic consequences, structured analysis is not simply a methodological preference; it is a professional obligation.

Much of the momentum toward more formal analytical methods emerged from repeated historical failures. Intelligence organisations throughout the twentieth century often relied on gifted individual analysts or ad hoc processes. Failures especially around the 2003 invasion of Iraq cast long shadows in intelligence structures in the US and UK, leading to direct political pressure to formalise analytical standards and processes. [2] The lack of effort at structuring assessment may lie behind the disastrous failure to predict the fall of Kabul to the Taliban in 2021. [3] Its rigorous application may be the reason that US and UK analysts successfully predicted the full-scale Russian invasion of Ukraine in 2022, whilst equally sophisticated states such as France and Germany—and even Ukraine’s own government—refused that analysis. [4]

These episodes demonstrated that even highly experienced analysts are vulnerable to cognitive shortcuts, institutional pressures, and mirror-imaging. Structure became necessary not because analysts were unskilled, but because the task itself was uniquely difficult: data are fragmentary, adversaries are deceptive, and outcomes are rarely certain. The military profession long ago recognised parallel needs in planning, adopting structured tools such as the UK Combat Estimate, the American MDMP, or the NATO 6-step Operational Planning Process to discipline tactical and operational thinking. [5] Despite SATs becoming a standard, the Intelligence Communities of NATO countries have so far failed to implement a common and shared structured approach to intelligence analysis. [6]

This is unfortunate, as the challenge to all modern states is underlined by the nature of intelligence problems. Analysts rarely work with complete or reliable information; instead they must draw inferences from partial signals, “essentially geared to penetrating those areas in which concealment and deception are endemic.” [7] Under these conditions, intuition alone is vulnerable to well-documented cognitive biases. Confirmation bias, for instance, leads analysts to overweight information that supports their existing beliefs. Anchoring can cause them to cling too closely to initial estimates, even when new evidence emerges. Availability bias encourages overreliance on vivid or recent events. [8] These are not moral failings but predictable features of human cognition. Structured techniques—such as key assumptions checks, analysis of competing hypotheses, indicator & warning matrices, decision-tree analysis and red-teaming—exist precisely to mitigate these vulnerabilities. They force analysts to articulate reasoning, challenge assumptions, and examine alternative explanations systematically. And they are well developed now: analyst handbooks proliferate in government, and Pherson & Heuer’s Structured Analytic Techniques for Intelligence Analysis provide strong handrails for new analysts. [9]

For military users who may be less familiar with these techniques, the value lies in what they make visible. A structured assessment provides clarity about what is known, what is uncertain, and how confidence was derived. This transparency supports better command decisions. When a commander receives an intelligence estimate built on explicit assumptions, clearly defined indicators, and a documented evaluation of alternative hypotheses, they can judge the robustness of the assessment and its relevance to operational planning. By contrast, an unstructured “expert judgement” product may be compelling on the surface (it might offer a good “story”) but offer no way to evaluate whether it is sound. [10] The issue is not that intuition is worthless—indeed, seasoned analysts often generate valuable insights—but that intuition without discipline cannot be audited or defended.

Time pressure, a defining feature of military operations, further strengthens the case for structure. The “time problem” in intelligence arises not only from the need to detect signals early but from the human struggle to recognise significance while events are still unfolding.[11] Commanders and analysts alike are prone to hindsight bias: once an event has occurred, it seems obvious in retrospect, leading organisations to believe they “should have known.” Structured approaches help counter this by generating explicit indicators in advance, enabling the identification of weak signals before they coalesce into unambiguous threats. They also create shared frameworks that help commanders interpret ambiguous situations without assuming that intelligence can predict events with certainty.

Critics have sometimes argued that structured analytical techniques do not reliably increase the accuracy of intelligence assessments. [12] Such a view mirrors some initial overenthusiasm at the effectives of SATs, and just as much misunderstands their primary purpose. SATs are not diagnostic tools in the medical sense; they are thinking tools. Their central value lies in improving the quality, transparency, and defensibility of reasoning. Even if accuracy improvements are modest or context-dependent, the discipline they impose reduces the risk of catastrophic misjudgement, particularly in high-consequence military environments. They also facilitate organisational learning. A structured assessment leaves a traceable record that can be reviewed, compared, and revised as events develop, as happens for example within the UK Cabinet Office with formal reviews of intelligence products. As we learn from Tetlock and Gardner [13], feedback is crucial to improving the accuracy (or, more specifically, the “Brier Score” [14]) of analysts, and auditable analysis allows professional reviews of intelligence products to help improve the individual analyst and improve processes within government.

Structured techniques also enhance communication between analysts and military decision-makers. Intelligence is only useful if it is understood, and misunderstandings between producers and consumers are common. Analysts may believe they have conveyed nuance, uncertainty, or conditional assessments, while commanders may perceive confidence or precision that was never intended. The adoption of probability- and confidence-based language, including frameworks such as the Professional Head of Intelligence Assessment (PHIA) scale, helps bridge this gap. [15] It provides a consistent lexicon for expressing uncertainty, enabling decision-makers to integrate intelligence assessments into planning processes more effectively. This author’s recent primary research into UK national intelligence products, conducted along with government, found that structured reasoning paired with structured communication, results in intelligence that is more actionable, more reliable, and more attuned to the needs of its military audience. [16]

SATs are not designed to replace judgement; they are designed to discipline it. Creativity remains essential in identifying novel patterns, generating hypotheses, and anticipating adversary behaviour. Structure simply provides a scaffold that ensures creativity does not drift into speculation. The two are complementary, not contradictory. In fact, many of the most innovative analytical leaps arise from structured activities—such as red-team exercises or alternative futures analysis—that deliberately force analysts to consider perspectives they might otherwise overlook. [17] Even properly designed and structured wargames can be treated like an analytical tool with clear benefits in situational understanding and a clear framing of options. [18]

For modern military organisations facing agile adversaries and complex operating environments, the adoption of structured analytical methods is therefore not simply best practice but operational necessity. Uncertainty cannot be eliminated, but it can be managed. Bias cannot be removed, but it can be mitigated. Adversarial deception cannot be wished away, but its effects can be constrained by disciplined reasoning. Structured methods achieve this by making thinking explicit, exposing assumptions to challenge, and enabling effective dialogue between analysts and commanders.

Perversely, many seem to be rushing past structured analysis and shoving it aside in favour of the unproven promise of Artificial Intelligence. Whilst AI can already automate routine, time-consuming tasks such as summarising reporting, processing imagery, or handling large data streams, it remains poorly suited to the core challenges of intelligence work: ambiguity, uncertainty, and adversarial deception. [19] AI systems depend on large quantities of reliable data and struggle with the fragmentary, contradictory, and deliberately manipulated information that defines real intelligence problems. They also “hallucinate,” importing or inventing false information in ways that analysts may not immediately detect, and cannot at the moment clearly lay out their reasoning. Because intelligence assessments ultimately require synthesis, and the ability to judge intent—capabilities AI cannot replicate—AI should be treated as an aid to human reasoning, not a substitute for it. Its promise is significant, but its peril lies in assuming that computational pattern-matching can replace the experienced human analyst, structuring their thinking in an auditable way in making sense of a deceptive and adversarial world.

Ultimately, structured analysis should enhance trust. Commanders do not need perfect intelligence; they need to understand the basis of the assessments on which they must act. After all, command decisions will rest on the commander’s judgement, not that of the perhaps quite junior analyst.  But when an intelligence product shows its workings (highlighting evidence, assumptions, gaps, dissenting interpretations, and the rationale for its conclusions) it empowers military leaders to make more informed decisions. In critical, time-pressured combat situations, this transparency is not optional. It is the foundation of intelligence professionalism.

Dr. Gustafson has conducted consultancy and advisory work for the UK Cabinet Office, and the MOD’s Development, Concepts and Doctrine Centre, including an integral role in developing UK Joint Intelligence Doctrine.