Just-in-Time Help¶
Interface — instantiates Progressive Disclosure
Contextual explanation or guidance shown at the point where the user needs it.
Just-in-Time Help delivers a small piece of explanation, an example, or a warning at the exact place and moment the user needs it — bound to the field they are in, the tool they just picked up, or the step they have reached — so guidance arrives without the user ever leaving the task to go look something up. Its defining idea is that help is pushed by the current task context and, crucially, carries the caveat or check that matters right here: it is the system offering the right sentence at the right moment, and where the moment is risky, the right validation before the action commits. It is not a manual the user must find; it is the manual coming to the user, one relevant fragment at a time.
Example¶
A developer typing in a code editor reaches into an unfamiliar library. As they type the function call, an inline popover appears showing the parameter signature and a one-line example — help bound to exactly the symbol under the cursor. A moment later they pass an argument that a recent version removed; a squiggle and an inline message surface before the commit: "parseDate(str, opts) — the locale string argument was dropped in v4; pass { locale } instead." When they try to commit anyway, an inline validation prompt holds the commit until the call is fixed. The developer never opened external documentation, and the mistake was caught at the point of action rather than in review an hour later.
How it works¶
Just-in-time help binds each fragment of guidance to a context — a field, a control, a symbol, a task step — and surfaces it when the user is in that context: on focus, on hover, on entering a step, or when the system detects a state worth flagging. Where that context is consequential, the fragment is the exception that applies here (a deprecation, a contraindication, an irreversible consequence) rather than generic prose. And where an input can be wrong in a way that matters, the help escalates from explanation to a validation prompt that checks the input and, if needed, holds the action until it is corrected. The content stays minimal and dismissible; the user can also summon it on demand when it does not fire on its own.
Tuning parameters¶
- Trigger eagerness — help that fires automatically on context versus help the user must request. Eager help catches the unaware; it also risks nagging the competent.
- Intrusiveness — a passive tooltip versus a blocking prompt. Passive respects flow; blocking is reserved for the input that must not pass wrong.
- Content depth — a single line versus a line plus a link to fuller reference. Shallow keeps the interruption cheap; deeper serves the genuinely stuck.
- Validation strictness — warn-and-allow versus block-until-fixed. Blocking prevents costly errors but frustrates users who know better than the checker.
- Dismissal memory — whether a dismissed hint stays gone. Remembering respects the expert; forgetting protects the occasional user who needed it after all.
When it helps, and when it misleads¶
Its strength is erasing the context switch: the user learns or is warned inside the task, at the moment of relevance, so they neither abandon the flow to read docs nor discover a mistake long after making it. Bound tightly to the moment of need, it is help delivered on the just-in-time principle rather than stockpiled in a manual nobody opens.[n1]
Its failure mode is over-prompting. Help that fires on every field trains users to dismiss it reflexively, so the one prompt that genuinely mattered — the contraindication, the irreversible step — is swatted away with the rest; the boy-who-cried-wolf dynamic quietly disarms the safety it was meant to provide. The mirror failure is help that only appears unbidden and cannot be summoned when the user is confused but the system did not notice. The guarding discipline is to spend interruptions sparingly — reserve automatic surfacing for moments that are stage-relevant or genuinely risky — and to always leave a way for the user to pull help on demand.
How it implements the components¶
task_stage— each fragment is bound to a specific point in the work (this field, this tool, this step), and surfaces only when the user is at that point.critical_detail_exception— where the moment is consequential, the help is the warning that applies here — the deprecation, the contraindication, the irreversible consequence — placed before the action, not buried after it.progressive_validation_prompt— for inputs that can be wrong in ways that matter, it checks the input inline and, when needed, holds the action until it is corrected.
It does NOT implement information_layer or depth_label — the click-to-open, author-labeled content block of Expandable Section; just-in-time help is not a section the user opens but guidance the system surfaces at the active step, and it can interrupt or block where an expandable section only ever waits.
Related¶
- Instantiates: Progressive Disclosure — it is the archetype's point-of-need face: the right fragment revealed by the moment, not by the user's search.
- Sibling mechanisms: Expandable Section · Advanced Settings Panel · Summary-Detail View · Drill-Down Dashboard · Tiered Decision Support · Wizard or Stepper Workflow · Staged Onboarding · Progressive Training Module
Editorial Notes¶
Form Classification¶
Form family: Interface, Display & Cue
Rationale: Just-in-Time Help operates as a user-facing prompt, display, template, or perceptual cue that shapes attention and action at the point of use because it contextual explanation or guidance shown at the point where the user needs it
Independent corroboration: The frozen evidence defines Just-in-Time Help as 'Contextual explanation or guidance shown at the point where the user needs it', so its operative form is Interface, Display & Cue.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Human-Computer Interaction
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Universal
Rationale: HCI developed context-sensitive help and progressive disclosure tied to the user's current interface state.
Related originating lineages:
- Education & Pedagogy — Instructional support traditions materially shaped guidance delivered at the moment of task need.
Review resolution: Both independent reviews place the primary lineage in human_computer_interaction. The queued differences (domain_reach_disagreement) concern secondary metadata rather than primary provenance. The final retains education_pedagogy only where a reviewer supplied a formative-lineage rationale; downstream application by itself is not treated as origin. origin_mode=cross_disciplinary_synthesis records the relationship among origin traditions, while domain_reach=universal records application breadth separately. encyclopedia_synthesis=false reflects whether either reviewer identified a corpus-specific synthesis, and confidence=high preserves the more cautious evidence assessment.
Review outcome: Reconciled after independent review; high confidence.
Notes¶
[n1] Just-in-time, borrowed from the Toyota Production System, where parts arrive exactly when the line needs them rather than sitting in inventory. Applied to help and learning, it means delivering the explanation at the point of use instead of front-loading it into a manual read (if at all) long before it is relevant. ↩