sample questions

10 of the 90 questions, spread across the domains. Open one to see the answer and why.

  1. A marketing manager asked Claude, in one message, to compare five competitors' product brochures and recommend which product the company should match. The recommendation reads well, but the manager cannot tell which features came from which brochure. A director will act on it, so the manager must check the comparison before any recommendation is built on it.

    What is the most effective way to restructure the request?

    1. Send the same request again in one message, and add that every claim must name the brochure it came from.
    2. First ask for a feature table across the five brochures, check it, then ask for the recommendation from it.
    3. Ask Claude to answer as a senior market analyst, so that it weighs the five products with much more care.
    4. Upload the brochures one at a time in new chats, and ask for a full recommendation after each one.
    Show the answer

    Answer: B.

    Correct: a checked feature table, then the recommendation. The manager's constraint is to check the comparison before a recommendation is built on it, so the comparison must come back as its own step, and the recommendation then starts from a table they have corrected. Why not the others: same request with sources named — it would make claims easier to trace, but the comparison still arrives inside a finished recommendation, too late to correct; a senior market analyst — a role changes the tone, not what they can check; one brochure per new chat — it splits the work, but each chat sees one brochure, so nothing compares the five.

  2. A procurement officer asks Claude to summarise a 30-page supplier contract on one page, covering the price, the term, the renewal terms and the exit clauses. The summary reads clearly and every figure looks plausible. The director will decide from it, without opening the contract, whether to renew.

    What is the most effective way for the officer to check the summary before sending it?

    1. List the four topics they asked for and find each in the summary, then check each figure against its clause in the contract.
    2. Check each price, date and notice period in the summary against its clause in the contract, then send it if all of them match.
    3. Ask Claude to reread the contract and say whether its summary is accurate, then send the summary if Claude confirms that it is.
    4. Read the summary through as the director would, and rewrite any sentence that is unclear or sounds less certain than the rest.
    Show the answer

    Answer: A.

    Correct: the topics first, then each figure against its clause. The director will not open the contract, so a missing topic and a wrong figure would both reach the decision. The four topics the officer named are the checklist for the first, and the contract's clauses are the source for the second. Why not the others: check each price, date and notice period — it would catch a wrong figure, but a summary that skipped the exit clauses would pass it; ask Claude to confirm — a self-check can help, but it does not replace reading against the contract; read as the director would — it improves clarity, but checks the summary only against itself.

  3. A sales coordinator has planned a 250-guest conference in one chat for two weeks, and the chat now holds over 150 messages. Recent replies suggest venues that are too small and forget the agreed catering budget. The coordinator must carry on tomorrow, starting from every decision already agreed.

    What is the most effective next step?

    1. Keep working in the same chat, and restate the venue decision at the start of every message they send from now on.
    2. Start a new chat tomorrow and retype the planning brief from memory, adding the decisions that they can recall.
    3. Switch to a model with a larger context window, and carry on in the same long chat exactly as before.
    4. Ask Claude to list the agreed decisions and open items, check that list, and start a new chat with it.
    Show the answer

    Answer: D.

    Correct: a checked summary, then a new chat. The long chat buries early decisions, so a new chat restores focus, and the checked list carries every agreed decision into it, which meets "starting from every decision already agreed". Why not the others: retype the brief from memory — a clean restart would work, but any decision they forget is lost, so "every decision" fails; restate the venue each message — it keeps the venue in view, but not the catering budget, and the chat keeps growing; a larger context window — more room does not stop early decisions being buried among later messages.

  4. A facilities coordinator at a university needs a new way to settle room-booking clashes. The agreed requirements: front-desk staff decide within one working day, and no new software this term. Asked "What's the best way to handle booking clashes?", Claude returned one detailed proposal built on a booking-system plug-in. The coordinator has seen no other approach.

    What is the most effective next prompt?

    1. Paste the two requirements and ask for three different approaches, each checked against both, with what each one gives up.
    2. Ask Claude to adapt the plug-in proposal so that it works without new software, keeping as many of its steps as it can.
    3. Ask Claude to compare plug-in vendors and draft a purchase case, so that the facilities budget can cover one next term.
    4. Ask Claude to turn the proposal into a step-by-step checklist that the front-desk staff can follow from next week onwards.
    Show the answer

    Answer: A.

    Correct: options against the requirements. The coordinator has seen one approach, and Claude never saw the requirements, so nothing has been compared. Pasting the requirements and asking for several approaches with their trade-offs gives the coordinator a real choice. Why not the others: adapt the plug-in proposal — it could produce a workable process, but it stays anchored to the one idea and still compares nothing; compare plug-in vendors — it breaks "no new software this term"; turn it into a checklist — it develops materials before the design has been chosen, and the proposal needs software the term rules out.

  5. A finance team of five, on their organisation's Team plan, wants every month-end report Claude drafts for them to use the same table layout and rounding rules. The marketing and sales teams also use Claude, with their own report formats. The rules change next quarter, and the team lead wants to update them in one place.

    What is the most effective place to put the rules?

    1. Ask an Owner to add the rules to the organisation instructions, so that every chat in the organisation follows them.
    2. Put the rules in the instructions of a month-end Project shared with the five, with the reason behind each rule.
    3. Have each of the five add the rules to their own Instructions for Claude, so that every chat they open follows them.
    4. Paste the rules at the start of each month-end chat, so that Claude always has the latest version in front of it.
    Show the answer

    Answer: B.

    Correct: a shared Project's instructions. Project instructions apply in every chat in that Project, for every member it is shared with, so one edit next quarter updates all five; that meets "in one place" and keeps the rules away from other teams. Why not the others: their own Instructions for Claude — it would work for each person, but five separate copies must each be changed next quarter; organisation instructions — one place, but they reach every chat in the organisation, so marketing and sales would get finance's layout; paste at the start of each chat — it works only when someone remembers, and old copies keep circulating.

  6. A charity's fundraising officer wants Claude to find giving patterns in a donor list: names, email addresses, donor numbers, gift amounts and dates, and some notes about bereavements. The charity's policy allows donors' personal data in AI tools only once it has been anonymised. The officer must contact some donors once the patterns are known.

    What is the most appropriate first step?

    1. Remove names, emails and notes, replace donor numbers with codes, and keep the code key outside Claude.
    2. Upload the list with a line asking Claude not to keep the data, and delete the chat once the work is done.
    3. Upload the list as it is, because the charity's paid business plan does not train on its chats by default.
    4. Drop the analysis, because any donor list holds personal data and so can never be shared with Claude.
    Show the answer

    Answer: A.

    Correct: anonymise first, and keep the key outside. The policy allows the data only once it is anonymised, so names, emails and notes go, and coded donor numbers still let the officer contact the right donors afterwards through the key. Why not the others: ask Claude not to keep it — deleting the chat is a good habit, but the personal data still reached Claude, which the policy forbids; business plan does not train — the training default is real, but it does not meet the policy's anonymise-first rule; drop the analysis — unnecessary, because anonymising makes the work possible. Adapted from the exam guide's sample question 3.

  7. A marketing coordinator's saved prompt turns a week of customer survey comments into a one-page digest. This week's digest misses two of the five themes in the comments and calls one theme "trending upward", though the prompt only ever supplies one week of comments. The coordinator wants to know which change fixed which fault.

    What is the most effective first step?

    1. Switch the chat to a more capable model and re-run the prompt, because two faults at once suggest the model cannot handle the task.
    2. Rewrite the whole prompt with a role, an example digest, numbered steps and a no-trends rule, then run it on next week's comments.
    3. Add one line limiting the digest to what this week's comments show, re-run it on the same comments, then turn to the missing themes.
    4. Add "be thorough and accurate" to the end of the prompt, re-run it on the same comments, and compare the two digests side by side.
    Show the answer

    Answer: C.

    Correct: one change for one named cause, re-run on the same comments. The trend claim is a request beyond the source: one week of comments cannot show a trend. Fixing that cause alone, on the same input, shows exactly what the change did, which is what the coordinator wants to know. Why not the others: rewrite the whole prompt — it may well fix both faults, but four changes at once and new comments mean nobody can tell which change fixed what; a more capable model — it treats a prompt cause as a model problem; be thorough and accurate — it names no cause, so Claude still has to guess.

  8. An HR coordinator is asked for a two-page guide for new starters, drawn from a 40-page staff handbook. The guide must cover only what applies in a new starter's first week, and the HR manager will review it. Asking for the guide in one message gave a draft that wandered into pensions and long-service leave.

    Which TWO steps best structure this request?

    1. Ask again for the whole guide in one message, and tell Claude to keep the guide as short as it can.
    2. First ask Claude to list what in the handbook applies in a new starter's first week, and check that list.
    3. Ask Claude to summarise each handbook chapter in turn, check each summary, then join them into the guide.
    4. Then ask for the guide from the checked list, and have Claude review the draft against that list and revise.
    5. Ask Claude for three versions of the guide in one message, and send the HR manager whichever reads best.
    Show the answer

    Answer: B and D.

    Correct: a checked first-week list, then the guide drafted and reviewed against it. The draft wandered because choosing what applies in week one was buried inside the writing. A list the coordinator checks fixes that choice, and a review against the list keeps the guide to it. Why not the others: summarise each chapter — it is a real chain with checks, but it splits by chapter rather than by the first-week need, so pensions still get a summary and the work takes far longer; shorter, in one message — length was not the problem, the choice of content was; three versions — all three make the same hidden choice.

  9. An HR officer asks Claude to turn a training attendance spreadsheet into a report for the leadership team in three parts: attendance by department, the three departments with the lowest attendance, and courses with waiting lists. The report will set next year's training budget, so every figure in it matters.

    Which TWO checks would most directly find a wrong figure or a missing part?

    1. Tick each of the three requested parts off against the report, and note any part that is missing or only touched on.
    2. Ask Claude to list the figures it was least sure of, and trace only those figures back to rows in the spreadsheet.
    3. Read the commentary aloud to check that its tone and length suit a leadership team reading it before a budget meeting.
    4. Ask Claude to write the report a second time, and keep whichever of the two versions reads more clearly to them.
    5. Trace each attendance rate in the report back to its rows in the spreadsheet, and recompute a sample of the rates.
    Show the answer

    Answer: A and E.

    Correct: tick off the three parts, and trace each rate to its rows. The first reads from the request to the report and finds a missing part; the second reads from the report to the spreadsheet and finds a wrong figure. Because every figure matters, only these two cover both faults. Why not the others: figures Claude was least sure of — a fair place to start, but a wrong figure is not always one Claude doubts, so the rest go unchecked; read the commentary aloud — useful for the audience, but it tests tone, not figures or parts; write it a second time — two fluent versions can share one error, and clearer is not more correct.

  10. A finance team lead on a Team plan wants every chat their six colleagues have about the monthly cost report to follow two rules: amounts shown in thousands, and only the cost-centre names from an approved 40-row list. The rules must apply without anyone pasting them in.

    Which TWO steps best achieve this?

    1. Tell Claude in their own chat to remember both rules, so that its memory applies them for the whole team.
    2. Write both rules into the instructions of a Project shared with the team, so every chat in it uses them.
    3. Add the approved 40-row list to that Project's knowledge, so chats in the Project can check names against it.
    4. Run every monthly cost report in one long chat they keep open, so the rules agreed in it are never lost.
    5. Paste both rules at the top of each new chat, and ask colleagues to copy the same text from their message.
    Show the answer

    Answer: B and C.

    Correct: Project instructions and Project knowledge. Instructions apply to every chat in the Project, and sharing the Project reaches the lead's colleagues; the list in its knowledge lets each chat check names. Neither depends on anyone pasting. Why not the others: paste the rules each time — it would work, but "without anyone pasting them in" rules it out; their own memory — memory belongs to their account, so colleagues' chats never use it; one long chat — rules agreed early get buried as the chat grows, and the colleagues work in their own chats.