---
title: Hierarchical Task Analysis (HTA)
type: concept
tags: [task-analysis, human-factors, observation, elicitation, goal-decomposition]
sources: ["[[sources/2010-salmon-hta-vs-cwa]]"]
created: 2026-04-16
updated: 2026-04-16
---

# Hierarchical Task Analysis (HTA)

Goal-based task decomposition method from the human-factors / ergonomics tradition. Originated with **Annett et al. (1971)**; mainstream procedure codified by **Stanton (2006)**. Still the dominant task-analysis method in human factors >40 years after its introduction ([[sources/2010-salmon-hta-vs-cwa]]).

## Core idea
An operator (human or machine) is "required to do *X*" to achieve a super-ordinate **goal**. Decompose that goal into **sub-goals**, **operations**, and **plans** (sequence/conditions for operations). Recurse until the redescription is "good enough" — a judgement call, not a fixed depth.

Output: a **goal hierarchy tree** with plans at each level.

## Procedure (Stanton 2006)
1. State the overall goal.
2. State subordinate operations.
3. State the plan (sequence/conditions).
4. Check adequacy of redescription (revise if needed).
5. Consider first/next suboperation → recurse if further redescription required, else terminate.
6. Repeat until all operations terminated.

## Data collection
No single method. The standard combo is:
- **Observation** (shadowing)
- **SME interviews / questionnaires**
- **Walkthroughs** (think-aloud while operator performs the task)
- **User trials**
- **Documentation review**

**Iteration with SMEs is mandatory** — a single pass never produces a satisfactory hierarchy.

## Task-decomposition template (Kirwan & Ainsworth 1992)
For each leaf-level step, capture:
- Initiating event / cue
- Information requirements
- Interface features (controls / displays)
- Actions required
- Decisions required
- Complexity / difficulty
- Outputs
- Feedback

→ **This template is directly reusable as an observation-protocol field-note structure for BPM process discovery.**

## Granularity
HTA can decompose to **button-press level** — e.g. "click on drop-down menu, click right mouse button to activate mapping menu, click on map to drop waypoint." This is appropriate for **error prediction** (SHERPA), **training design**, and **interface evaluation** — it is **far too fine** for BPM process discovery, which should remain at activity level or higher ([[sources/2018-dumas-fundamentals-of-bpm-ch5-discovery]] §6.1 warns against step-level descent).

## Relation to BPM discovery

HTA is not a BPM method, but its discipline maps cleanly:

| HTA step | BPM discovery analogue |
|---|---|
| State overall goal | Process boundaries ([[sources/2018-dumas-fundamentals-of-bpm-ch5-discovery]] §5.3 step 1) |
| State subordinate operations | Identify main activities (§5.3 step 2) / Sharp's TRAC subprocesses ([[sources/2014-sharp-using-scope-models]]) |
| State plan | Control flow (§5.3 step 4) |
| Check adequacy + recurse | Interview → Modeling → Validation cycle (§5.2.2) |
| SME iteration mandatory | Dumas's ≥2 interview iterations per expert |
| Multi-method default | Dumas's §5.2.4 "mix methods" recommendation |

**Use for BPM observation sessions:** adopt HTA's 8-field decomposition template as the observation field-note structure. Recursion stops at the activity level (not the button-press level).

## Related
- [[sources/2010-salmon-hta-vs-cwa]] — primary source.
- [[methods/process-discovery-methods]] — observation method.
- [[concepts/process-discovery]]
- [[syntheses/qualitative-discovery-method-selection-matrix]] — observation row.
- [[syntheses/interview-structuring-for-process-models]] — iteration cycle.
- Cognitive Work Analysis (CWA) — formative/constraint-based counterpart (Vicente 1999) *(referenced-not-ingested)*.
- Kirwan & Ainsworth (1992) *A Guide to Task Analysis* *(referenced-not-ingested)*.
- Annett et al. (1971) *Task Analysis* *(referenced-not-ingested — HTA origin paper)*.