Ask AI for a Table, a Checklist or a Narrative: Choose the Output for the Job

AI

The assistant gives you a well-written page. You need to compare five options, so you turn the page into a table yourself. The draft was readable, but it was not prepared for the job.

Choose the output format according to what the reader must do next. Use a table for consistent comparison, a checklist for observable checks and a narrative for explanation. Combine them only where each contributes something different.

An open notebook with a handwritten weekly plan and pen.

Photo by Isaac Smith on Unsplash.

Identify the reader's next action

Ask whether the result will be scanned, compared, followed, discussed or transferred into another system. A format that looks professional on screen may be awkward in the place where it will actually be used.

For an illustrative supplier review, a director may need a table showing scope differences before reading a short explanation of the trade-off. A long narrative can make the differences unnecessarily difficult to find.

Specify the next use before asking for length or tone. Those choices should serve the task rather than become the task.

Use a table when rows really are comparable

Define the columns and explain what each field means. Require missing information to remain missing rather than letting the assistant fill every cell with a guess.

A comparison table should preserve conditions. If one delivery date depends on client input, put that qualification beside the date, not in a general note easily missed below the table.

Check that the rows represent the same kind of thing. Comparing a monthly service fee with a one-off setup charge under a single Cost column hides an important difference.

Use a checklist for things somebody can establish

Each item should describe an observable condition or action. Be clear whether the list is for preparation, checking or acceptance.

  • Preparation: information or materials needed before starting.

  • Action: steps a person should perform.

  • Verification: evidence that the result meets a condition.

  • Decision: a choice requiring an authorised person.

Avoid mixing these into one series of unexplained ticks. A checkbox beside a judgement does not establish who may make that judgement.

Use narrative for reasons and consequences

A short explanation can connect facts that a table keeps separate. Ask for the main answer early, followed by the conditions that could change it.

For the supplier example, the narrative might explain why the lower quoted fee is not directly comparable until an excluded task is priced. It should not repeat every cell or invent a preferred option.

Keep unsupported opinions and source-backed facts distinguishable. The format should make uncertainty easier to see, not disguise it.

Check the handoff format

If another person must paste the output into a worksheet or task system, agree field names and ordering. If the result is for a meeting, choose a readable level of detail.

Try one example with the receiving person before turning the structure into a template. The useful measure is whether they can use the output without rebuilding it. More elaborate formatting is not automatically more helpful.

For related practical guidance, see defining what usable output requires and specifying business requirements clearly.

Hili Consulting can help choose an output structure that supports the next person using the work. Arrange a free 30-minute consultation and bring one example of the work you want to improve.

Previous
Previous

What Context Should You Give AI for a One-Off Business Task?

Next
Next

Small-Project Stage Gates: What Must Be Decided Before Spending More?