“We need to start using AI” is a reasonable ambition. It still leaves a marketing team with a difficult decision: where to begin. Content, reporting, research and lead handling all look promising. Trying to change them together makes it hard to find out what actually helps.
Start by choosing one piece of recurring work that someone on the team can describe and review. The exercise below is a suggested starting point, with illustrative marketing examples.
Look at a normal working week
Ask the people doing the work to show you recent tasks. Where did they copy information between tools? Which questions sent them searching through old documents? What needed several rounds of correction before it was ready? A real brief or report tells you more than a list of possible AI features.
- Content: turning approved product information into a first draft for a campaign.
- Reporting: preparing a written summary from figures the team has already checked.
- Enquiries: summarising a free-text request before a colleague reviews it.
- Company knowledge: finding the current answer to a recurring product question.
Choose work that happens often enough to compare several attempts. A task that is annoying once a year may be a lower priority than a small handoff repeated every day.
Check the information before choosing a tool
For each candidate, identify the information it needs and who maintains it. A writing assistant needs the current product brief. A report summary needs agreed definitions and the right reporting period. If people disagree about which source is correct, resolve that first.
Also establish what the team is allowed to share with the proposed tool. For an agency, make that decision separately for each client’s material. You can explore the workflow with approved examples before connecting live accounts.
Define the result in everyday terms
“Draft a campaign brief from these approved notes, leaving missing details as questions” gives you something to assess. “Make marketing more efficient with AI” does not tell the person reviewing the pilot what to look for.
- Name the person who owns the task and the person who will review the output.
- Describe the input, the expected output and where that output should go.
- Write down what must be accurate and what would make the output unusable.
- Record the effort needed today, including review and correction.
- Agree what evidence would justify continuing, changing the approach or stopping.
Keep the first version small enough to judge
For a content workflow, the first version might produce a draft in a document the team already uses. Publishing it, updating the CRM and notifying several teams can wait until you know the draft is useful. A working example should answer a specific question about the task.
Anthropic’s Building effective agents recommends starting with the simplest effective solution and adding complexity when it improves the outcome. For a marketing pilot, that means testing whether the extra steps actually help the person doing the work.
Count the work after the draft appears
Compare the whole task: finding the input, preparing it, generating the output, checking it, correcting it and putting it where it belongs. A quick draft that needs a long rewrite may not be an improvement. Keep notes on which corrections repeat; they help distinguish a missing instruction from a task that is a poor fit.
At the end of the pilot, the team should be able to explain what changed, what still needs attention and who will own the workflow. That gives you a reason to expand it—or a useful reason to choose a different starting point.
If you are still deciding where to begin, see how we approach AI for marketing teams.
