In this guide
  1. Start with the decision
  2. Prepare a safe source pack
  3. Ask for structure before prose
  4. Turn the update into a recommendation
  5. Check the finished report yourself
  6. Try this in your next update
01

Start with the decision

Before you open an AI tool, finish this sentence: after reading this report, the reader needs to decide ____. If there is no decision, identify what they need to understand or monitor. This keeps a reporting exercise from becoming a longer account of everything you did.

For example, a weekly sales report might need a decision about where to concentrate the team’s effort. A project update might need approval to change a deadline. Give that purpose a prominent place in your brief.

02

Prepare a safe source pack

Collect the period covered, verified figures, important changes, open risks and the questions that still need answers. Separate confirmed information from assumptions. Use only tools and data your workplace permits.

For a practice run, use fictional data. Do not upload customer records, employee information, unpublished financial results or confidential project documents to an unapproved service. Removing a name alone may not make a document safe to share.

03

Ask for structure before prose

Ask AI to organise the information before asking it to write the final report. This makes omissions and unsupported conclusions easier to spot. Review the proposed outline against your source pack before moving on.

Practice prompt

Organise the notes below into a weekly report for a manager deciding next week’s priorities. Use four headings: What changed, Evidence, Risks, and Decision needed. Use only the supplied facts. Label missing information as a question. Do not invent explanations or numbers.

04

Turn the update into a recommendation

A report becomes more useful when it explains what the evidence means and states the next action. Keep an observation separate from an interpretation. An observed drop in activity is not, by itself, proof of its cause.

Fictional example

Before: We contacted 40 prospects and booked 6 meetings this week.

After: We booked 6 meetings from 40 prospect contacts. We do not yet have enough information to explain the conversion rate. Next week, I recommend recording the lead source and response reason for each contact so we can identify where to focus.

The stronger version adds a useful next step without inventing a success claim.

05

Check the finished report yourself

Check every number, date and comparison against the original source. Read each explanation and ask whether it is supported or merely plausible. Confirm that any proposed action fits your authority, budget and team commitments.

Then shorten the opening until the reader can see the main change and requested decision quickly. If the AI draft sounds more confident than the evidence allows, change it. You are accountable for the report, even when a tool helped you write it.

06

Try this in your next update

Choose one report you already write. Define its purpose, prepare a safe set of facts and test the four-heading structure. Compare the result with your usual version: is the decision clearer, is the evidence traceable and has anything important been lost?

Keep the parts that work. A good workflow earns its place by making the report more useful; it does not need to replace every step of your existing process.

Before you move on

  • The reader’s decision or information need is explicit.
  • Every figure matches a verified source.
  • Assumptions and missing information are labelled.
  • The recommendation follows from the evidence.
  • The tool and input data meet workplace requirements.

TAKE THE NEXT STEP

Make this lesson your own.

The companion PDF includes examples and a worksheet you can keep.

Explore the companion guide →

Email delivery coming soon.

This preview uses fictional examples where indicated. Read our editorial approach.