As of 13 August 2026, AI can only partly generate weekly business reports.

What to actually do

1. Hand it to a person

   The route this page recommends

   A person who owns the outcome does this end to end, worth it when the failure is dear.

2. Use a tool built for this

   Second choice

3. Do it yourself

   The distant third

   A chat interface, power-user skill, and roughly 1 hour until you can act on the result.

   How to actually do it

   1. Open the systems that contain the week's figures, such as sales, finance, marketing or support data, and export the relevant records with column headings and the reporting period.
   2. Gather the previous comparable period, approved targets and the written definitions of each metric, then remove unrelated personal data from the files.
   3. Paste the exports, definitions, targets and business context into the prompt, or upload them to a suitable AI reporting tool such as Rows.
   4. Ask the model to produce the report, calculation checks and audit table in the required format, using only the supplied data.
   5. Compare every total, percentage, variance and rate in the draft with the source exports and recalculate a sample of figures in a spreadsheet.
   6. Ask a colleague who understands the business to confirm that the metric definitions, explanations, risks and proposed actions match the current situation.
   7. Correct the report, record any unresolved data-quality issues, and send or publish the checked version with its reporting period and sources.

   Prompt

   Create a weekly business report from the data and instructions below.
   
   Business context: [describe the business, audience and decisions this report supports]
   Reporting period: [start date] to [end date]
   Source data: [paste the data or attach the export]
   Metric definitions: [define every column and calculation, including revenue, orders, customers, conversion, costs and targets where relevant]
   Previous-period data: [paste the comparable period, if available]
   Targets or budget: [paste the approved targets, if available]
   Required format: [state the sections, length, currency, rounding and chart requirements]
   
   Rules:
   - Use only the supplied data. Do not invent figures, sources, explanations or targets.
   - State the source and reporting period for every key figure.
   - Show the calculation behind every percentage, total, variance and rate.
   - Distinguish clearly between actual figures, comparisons, forecasts and interpretations.
   - Flag missing, duplicated, inconsistent or unusually changed data instead of silently correcting it.
   - If the data does not support a conclusion, say so.
   - Round only in the displayed result and retain enough precision for checking.
   - Write a concise executive summary, followed by key metrics, changes from the comparison period, target performance, risks, questions for investigation and proposed actions.
   - Put uncertain interpretations in a separate section called 'Needs human confirmation'.
   - Finish with a table listing each key figure, its source column or table, its calculation and any data-quality issue.
   
   Before writing the report, list the checks you will perform. Then produce the report and the audit table.

   Open it prefilled in ChatGPT or Claude, or copy it into Gemini, which takes no prefill link.