How Organisations Capture Decisions

How experts think, how that thinking is recorded, and how systems put it to work — a look at the published research behind decision capture.

Research Insights · Published by Elanza AI

1. Introduction

Every organisation depends on countless decisions, many of which rely on the experience and judgment of experts rather than written procedures alone. While this expertise drives effective outcomes, it is often difficult to capture, transfer, and preserve over time. As organisations evolve and experienced professionals move on, researchers have increasingly explored ways to understand expert decision making, capture the reasoning behind it, and build systems that can support the decisions that follow.

This paper brings together insights from published research to explain how these ideas have evolved and why they continue to shape modern decision science.

2. What Never Makes It Into the Manual

2.1 Tacit and Explicit Knowledge

Knowledge was long treated as a by-product of work rather than an asset in its own right. Karl Wiig pushed against that view in 1993, arguing that good knowledge underpins how successfully an organisation carries out its activities even though it is usually managed haphazardly — and that the people doing the knowing deserved as much attention as the technology around them [1].

The reason it resists management became clear once Ikujiro Nonaka and Hirotaka Takeuchi, in 1995, took up a distinction first drawn by Michael Polanyi and put it to work in organisations. Explicit knowledge can be set down in formal language — manuals, specifications, written procedures — and travels between people easily. Tacit knowledge is personal, grown from experience, and hard to put into words at all; intuitions and hunches belong here [2].

"They recognize that the knowledge expressed in words and numbers represents only the tip of the iceberg."

The Knowledge-Creating Company [2, p. 8]

Figure 1. Two kinds of knowing. An original illustration of the distinction between explicit and tacit knowledge described in the research, sized to reflect the estimate that roughly 90% of organisational knowledge is tacit and about 10% is explicit [2]. Source for the 90%/10% estimate: [3]

2.2 Beyond the Binary

Even that split is a starting point rather than a map. Maryam Alavi and Dorothy Leidner, in 2001, collected the alternatives, describing knowledge as know-about, know-how, know-why, know-when and know-with — in medicine, the difference between knowing which drug suits an illness, how to give it, why it works, when to prescribe it and how it interacts with others [4]. Working from education research, Lorin Anderson and David Krathwohl, in 2001, organised knowledge into four broad types, among them procedural knowledge — knowledge of how to do something — and metacognitive knowledge, awareness of one's own thinking [5].

The design lesson is direct: these classifications point to the need to support different types of knowledge and the flows between them [4]. A system built to hold facts will not, on its own, carry timing or judgment.

Figure 2. Knowledge is not one substance. An original diagram of the classification and medical example set out in the literature [4].

3. Asking in a Way That Works

If the most valuable knowledge resists being written down, the question becomes how to draw it out. That is the work of cognitive task analysis — structured methods for studying how skilled people actually think while doing their job.

3.1 The Critical Decision Method

One of the most widely used is the Critical Decision Method, developed in 1989 by Gary Klein, Roberta Calderwood and Donald MacGregor as a variant of John Flanagan's critical incident technique, extended for work carried out under time pressure, with heavy information and changing conditions [6]. Rather than asking someone to describe their role in general, it revisits one real, difficult incident and prompts for the cues they noticed and the judgments they made. It has been used with fireground commanders, paramedics, structural and design engineers, military leaders and programmers [6]. Robert Hoffman, Beth Crandall and Nigel Shadbolt, in 1998, later examined the method in its own right, showing the records it produces — situation assessment records and timelines — and its practical value for capturing expert knowledge and designing complex systems [7].

3.2 Classifying Elicitation Methods

There is no single technique, and that is a strength. Janet Burge, in 1998, classified elicitation methods into direct approaches — asking an expert how they do their job — and indirect approaches, for knowledge that cannot easily be expressed directly, with the choice guided by what you are trying to learn. Her survey — an online classification from Worcester Polytechnic Institute's Artificial Intelligence Research Group, accompanying her 1998 master's thesis, rather than a peer-reviewed paper — also groups methods by the kind of knowledge each one recovers [8].

Figure 3. Choosing the right way to ask. An original diagram of the classification of elicitation methods, sorted both by how they engage the expert and by what they recover. Four of the categories described in the source are shown here; it also lists methods that elicit classifications of domain entities and evaluations [8].

4. Building It Into the Work

Once decisions have been recorded, technology can carry them. Alavi and Leidner's 2001 review describes four things organisations do with knowledge — create it, store and retrieve it, transfer it and apply it — and defines knowledge management systems as software built to support and enhance exactly those four processes [4].

The role technology plays shifts at each step: providing access to sources of knowledge, gathering and storing it, linking sources so knowledge flows more widely and deeply, providing effective search and retrieval, and supporting the development of individual and organisational competencies [4]. At the point of application, knowledge is built into daily work — workflow automation routes documents, information, rules and activities automatically, and rule-based expert systems capture and enforce well-specified procedures [4]. Because these systems widen an organisation's memory, a decision made once can be applied across both time and distance [4].

Figure 4. Four processes, four kinds of help. An original diagram of the knowledge management processes and the supporting roles for technology described in the research [4].

5. Where It Has Landed

These ideas are not confined to research papers. The same body of work records what happened when organisations put them to use.

Table 1. Applications across industries.

Sector	Application
Insurance	A company hit by commoditisation and falling profits applied its best decision-making expertise through a new underwriting process supported by a knowledge management system, and moved into profitable niche markets, as reported in 1998 [4].
Automotive	Chrysler set up Tech Clubs so specialists split across teams would not lose shared expertise, as reported in 1998; Ford reported development time falling from 36 to 24 months through knowledge sharing, also as reported in 1998 [4].
Chemicals	Buckman Laboratories ran an indexed online forum that helped it move from selling products to solving customer problems, as reported in 1998 [4].
Training and system design	Knowledge captured through cognitive task analysis feeds instructional design and system development, as reported in 1998 [7].

The year given for each row is when the account was published, not necessarily when the practice began; these are secondary accounts drawn into the 2001 review from earlier sources (Davenport & Prusak, 1998; Ruggles, 1998; Zack, 1998; Gazeau, 1998), and the outcomes are self-reported and were not independently verified [4].

Understanding how people decide, and recording it deliberately, turns out to be useful wherever decisions are made.

6. Still in Motion

These works come from different fields — naturalistic decision making, knowledge management, educational psychology — and largely developed independently rather than in sequence. Read together, though, they answer a connected set of questions.

Figure 5. Converging questions across separate literatures. An original timeline of selected works discussed here; not every source in the reference list is plotted [1], [2], [4], [6], [7].

The journey from understanding human expertise to building systems that can assist decision making is still unfolding. Each study discussed in this article contributes a different piece of that journey — revealing how experts think, how their reasoning can be elicited, and how organisations can store, share and apply what they capture. Rather than offering a single solution, this body of research reflects the ongoing pursuit of making complex decisions more transparent and better understood.

The direction of travel is visible in the research itself. The 2001 review that mapped technology's role across the four knowledge processes did not present a finished picture; it set out research issues to guide further work, treating knowledge management systems as a field still being built [4]. The capture methods followed the same path — first developed in 1989, then examined in their own right in 1998 for data quality, reliability and efficiency [7].

Across cognitive science and decision science, researchers continue to investigate how expert thinking can be understood, represented, and supported without losing the context that makes human decisions effective. Taken together, these studies set out the groundwork that any decision support system inherits: what expertise consists of, how to get at it, and what happens to it once it is written down.

7. The Short Version

The expertise that makes an organisation effective is real, valuable and largely unwritten — and this research shows it can be recorded on purpose rather than left to chance. The methods already exist, and they were built for the conditions businesses actually work in: time pressure, incomplete information, and situations that keep changing [6].

The return appears in ordinary places. Teams learn the reasoning and not only the rules. Judgment stays in the business when people move on. Captured expertise flows into training and system design [7] — and, as one insurer discovered, into new and profitable markets [4]. Understand the thinking, capture it well, then let systems store, share and apply it. [4] Technology carries good judgment further than any individual can; the judgment itself still starts with people.