Skip to content

Automate a weekly marketing performance report

A scheduled set of agents pulls channel data, explains the changes and drafts a report for review.

Recommended: ArchitectAutomationData

Best for

  • Marketing leads reporting every Monday
  • Agencies managing many client accounts
  • Teams tired of copy-pasting from dashboards

What you need

  • Exports or API access for your ad and analytics tools
  • A shared spreadsheet or report template
  • An agreed list of metrics that matter

Starter prompt

Copy it, fill in the parts in square brackets, and paste it into the AI assistant your organisation allows.

You are the analyst agent in a weekly marketing report. You receive this week's and last week's figures for [channels] as a table. For each channel, state the change in [metrics], then give the most likely reason using only the campaign notes provided: [campaign notes]. If the notes do not explain a change, write 'Reason unclear, check with [owner]'. Flag any metric that moved more than [threshold] percent. End with three actions for next week, each tied to a specific number in the table. Data: [paste table].

How to use it

  1. 1.Fix the metrics

  2. 2.Build the data agent

  3. 3.Add the analyst agent

  4. 4.Deploy with a reviewer

  1. 1

    Fix the metrics

    Agree with your team which metrics and channels the report covers. Write down what counts as a change worth flagging.

  2. 2

    Build the data agent

    Set up an agent that pulls last week's figures from your tools into one table on a schedule. Check its output against the dashboards for two weeks.

  3. 3

    Add the analyst agent

    Pass the table to a second agent using the prompt above. Write a few test weeks with known answers and check it explains them correctly.

  4. 4

    Deploy with a reviewer

    Schedule the pipeline to run early Monday and send the draft to a named reviewer. Nothing goes to leadership until they approve it.

Watch out. Agents can invent confident reasons for a change. Keep the rule that unexplained changes are marked unclear, and have a person approve every report.

What you end up with

A reviewed performance report waiting for you each Monday, built without manual copying.

Read why AI matters in marketing & content

Related use cases