Briefing Mindset
The habit of prompting Claude the way you would brief a brilliant new colleague: highly capable, and entirely without context on your company, project and standards until you supply it. A good prompt contains everything that good delegation contains.
Exam context: The foundation of the whole domain. The right answers supply the context a new colleague would need; the wrong options assume that Claude will infer what is needed.
See also: 1.1 Creating Effective Prompts for Business Tasks
Five Components of an Effective Prompt
The five parts of a complete business prompt: the task (a verb plus a deliverable), the context Claude cannot know, the actual source material, the output specification, and an example when standards matter.
Exam context: When a question compares prompts, count the components. The right answer is almost always the option that supplies the missing one; the wrong options improve the wording while the missing component stays missing.
See also: 1.1 Creating Effective Prompts for Business Tasks
Output Specification
The part of a prompt that states the deliverable's format, length, tone and structure: three bullet points per workstream, neutral tone, under 400 words.
Exam context: State length as a number, not as "short". A prompt with no output specification leaves format and length to chance, and leaves the rework to you.
See also: 1.1 Creating Effective Prompts for Business Tasks
Role Framing
Opening a prompt by asking Claude to respond from a named role, such as "as an HR communications specialist". A role shapes vocabulary, tone and priorities, and even a single sentence makes a difference.
Exam context: Paired with audience naming, which changes the output more than most instruction tweaks: the same analysis briefed for the board and for the project team should read differently.
See also: 1.1 Creating Effective Prompts for Business Tasks
Source Material
The actual inputs a task needs: the report, the meeting notes, the data, pasted into the conversation or uploaded as files. Claude works from what it is given.
Exam context: Describing a document from memory instead of attaching it is a recurring wrong option; the from-memory version adds errors and omissions the real document does not have.
See also: 1.1 Creating Effective Prompts for Business Tasks
Task Decomposition
Structuring a complex, multi-part request into pieces Claude can do properly, one at a time, instead of putting all of it into a single prompt. The official remedy list for an unhelpful answer includes breaking complex requests into substeps.
Exam context: Match the technique to the task. Independent parts need separate chats. Connected parts in a known order, with no check between them, need numbered steps in one prompt. Connected parts that you must check or decide on before the next part need a staged conversation. Unclear structure needs a plan first.
See also: 1.2 Decomposing Complex Requests
Numbered Steps
The simplest decomposition technique: listing a task's connected parts as an explicit sequence inside one prompt. Claude follows numbered steps far more reliably than one long paragraph of requests. Numbered steps need an order that you already know.
Exam context: The right choice when order or completeness matters, the order is known, and you do not need to check a result between steps. If you must check or decide before the next step, stage the work instead.
See also: 1.2 Decomposing Complex Requests
Staged Prompting
Working through a deliverable whose parts build on each other stage by stage in one conversation: outline first, review and adjust, then draft section by section. Each stage's approved output becomes the starting point for the next.
Exam context: The right answer for connected work where you must check a result, or make a decision, before the next part can be right. Put the stage break at that check. A document with several sections in one voice needs a check after each section. Corrections at the outline stage take seconds. The same corrections after 20 pages do not.
See also: 1.2 Decomposing Complex Requests
Plan-First Prompting
Asking Claude to propose an approach before executing: "before writing anything, list the sections you would include and what each covers". Reviewing a plan takes seconds; reworking a wrong 2,000-word draft does not.
Exam context: Useful only when the right structure is genuinely unclear. Offered where the structure is already known and agreed, it is a small unnecessary step that the exam offers as a tempting improvement.
See also: 1.2 Decomposing Complex Requests
Mega-Prompt
The anti-pattern (a common practice that gives bad results) of putting an entire multi-part deliverable into one prompt and asking for the finished product in a single pass. Attention spreads unevenly, so some parts come back strong and others given little attention.
Exam context: One of the two decomposition failure modes offered as wrong options; the other is fragmenting connected work across separate chats. Both are plausible, but both are weaker than the technique that fits. Use numbered steps in one prompt when the order is known and no check is needed. Use a staged conversation when you must check or decide between parts.
See also: 1.2 Decomposing Complex Requests
Memory (claude.ai)
Claude's cross-chat memory, built as a set of individual topics covering your role, projects, preferences and ongoing work context. It arrived on 11 September 2025 for Team and Enterprise plans. It reached Pro and Max from 23 October 2025. As of 14 November 2026, it is on by default for Free, Pro, Max and Ultra plans. As of that date, on Team and Enterprise plans an owner decides whether memory is available. In organisations created before 31 October 2026, memory stays off for each member until that member enables it. In most newer ones, memory starts on.
Exam context: Memory is context about you, saved as individual topics, not transcripts or working documents. It does not carry an approved outline or tone guide into a new chat; explicit carry-forward remains your job.
See also: 1.2 Decomposing Complex Requests
Chat Search
The ability, on paid plans, to ask Claude to search past conversations and reference relevant details. It is retrieval you must prompt for, not automatic carry-over.
Exam context: Wrong options treat chat search or memory as a way to move working material between chats automatically; the right answer pastes the output in, re-attaches the file, or keeps it in a Project.
See also: 1.2 Decomposing Complex Requests
Iteration
The normal working cycle with Claude: output, specific feedback, better output. First outputs are drafts, and the professionals who get the most from Claude plan a round or two of feedback into every task.
Exam context: Iteration is not a failure state. The exam contrasts it with the two wrong approaches: accepting a mediocre first draft, and abandoning the tool after one attempt.
See also: 1.3 Iterating Prompts to Improve Output
Targeted Feedback
Follow-up that names the gap between the output and the need: the audience, the tone, the length, the scope, or the missing constraint. "Cut it to 200 words, keep the pricing paragraph, lose the marketing adjectives" is targeted; "make it better" is not.
Exam context: The tested skill of objective 1.3. The vague retry reproduces the problem because no new information enters the conversation; the right answer always names the fix.
See also: 1.3 Iterating Prompts to Improve Output
Re-Briefing
Rewriting the original prompt and starting a new conversation when the draft is not close and the brief itself was wrong. The brief asked for the wrong task, used the wrong framing, or left out something the whole draft depends on, such as the reader. Correcting a wrong brief across many turns works against the framing instead of fixing it. Re-brief also when a long series of corrections has made the conversation contradict itself. If the brief was right and Claude misread it, refine instead: a follow-up that corrects the misreading is enough.
Exam context: The wrong choice between refining and re-briefing happens in both directions. One is correcting a fundamentally wrong brief turn after turn. The other is restarting when the output needed only one specific fix, or when Claude misread a correct brief. Questions test whether you can tell which situation the scenario describes.
See also: 1.3 Iterating Prompts to Improve Output
Research
The claude.ai feature for multi-source questions: Claude works agentically, running multiple searches that build on each other and delivering a report with easy-to-check citations. Ordinary web search suits quick factual queries answered in one or two lookups.
Exam context: Brief Research like a research question, not a topic: scope, angle, and what the findings must inform. Research for a quick fact is over-engineering; a single search for a multi-source comparison is too little searching.
See also: 1.4 Adapting Strategy to Task Type
Diverge-Then-Converge
The brainstorming sequence: first ask for quantity and range with judgement withheld, then converge in a second step by applying criteria, shortlisting and developing the ideas that remain.
Exam context: Constraints applied at the divergence step limit the variety that is the entire purpose; criteria belong in the convergence step. The tightly constrained brainstorm is a common wrong option that uses the briefing of another task type.
See also: 1.4 Adapting Strategy to Task Type