A customer service team lead at a homeware retailer asks Claude to turn the returns procedure into guidance for new agents. The right outcome depends on a chain of questions: whether the item was bought online, then whether it is within the return window, then whether it is damaged, with each answer deciding which question comes next. Claude produced a table with one column per question, and every cell matches the policy, yet new agents keep reaching the wrong outcome. What is the most likely cause?
- AThe table has too few columns to hold every detail of the policy, so agents are missing conditions that a wider table would show.
- BClaude summarised each rule too briefly in the cells, so agents need the full policy wording pasted into every cell of the table.
- CNew agents are not used to reading tables, so the same table should be reformatted with colour coding to make each column clearer.
- DA table sets items side by side as if they were parallel, so it hides the order in which each answer leads to the next question. Correct
Why A is wrong: Adding detail is a natural first reaction when people get things wrong. It is wrong here because every cell already matches the policy; the content is complete, and more columns would make the sequence even harder to follow.
Why B is wrong: This is tempting because brief cells can lose nuance. It does not fit the symptom: the cells are accurate, and pasting long policy wording into a table keeps the same layout that hides which question follows which.
Why C is wrong: Making a format more readable sounds sensible, which is why this appeals. It treats a structural mismatch as a cosmetic one: colour coding still presents the questions as parallel columns rather than as steps that depend on earlier answers.
Why D is correct: Tables suit information where every item shares the same attributes and can be compared across. A procedure in which each answer decides the next question is sequential, so it fits numbered steps or a decision flow that shows the order and the branches.