AI for digital marketing beginners: five tasks worth practising

Give AI a bounded task and a source to work from

“Create a winning marketing strategy” is a very large request with very little context. It invites a confident answer built on guesses. A smaller request is easier to review: organise these customer questions, draft three headlines from these facts, or explain this labelled sample table.

Write the business goal, audience, known facts, limits and expected output first. Keep confidential information out of the prompt. If an input is an assumption, label it. If a fact is unknown, ask the tool to leave it unknown rather than fill the gap.

The examples below are authored demonstrations of a workflow. They are not live model responses or client campaign results. Use them to understand the review step, then inspect whatever your chosen tool actually returns.

Task 1: turn a brief into research questions

A fictional café wants useful table enquiries. Known facts: it has a menu and seating. Prices, opening hours and availability are unknown. List eight questions a potential visitor may need answered. Group them by menu, location, visit planning and enquiry. Do not invent answers or search volumes.

A useful direction might include: where can I see the menu, where is the café, what should I know before visiting and how can I ask about a table? The output is a list to investigate. It is not evidence that customers asked those exact questions.

Review for duplicates, missing practical questions and unsupported assumptions. Then check the real business’s current information where available. Keep a source note beside every answer you plan to publish.

Task 2: draft copy without adding unsupported claims

Draft three plain-English headlines for a fictional home-cleaning quote page. The page should ask for area, property size and cleaning type. Do not invent ratings, prices, service areas, response times or guarantees. Make the next action clear.

The edited direction keeps the known service and next action. It still needs the business’s real offer, service area and process before publication. Sounding fluent is not the same as being ready to use.

Task 3: turn the message into a creative brief

Using only the checked cleaning-service message, create a short brief for a designer. Include audience, purpose, message, visual requirements and next action. Mark any missing service facts. Do not request customer photographs or testimonials we do not have permission to use.

A useful brief tells the designer what the visual needs to communicate. For this exercise, it might ask for a clear service explanation and a visible quote action. It should not ask for a fabricated before-and-after photograph or a badge claiming an award.

Check whether someone could make the creative without guessing the audience or promise. Cut vague requests such as “make it viral.” Replace them with a specific communication job and the facts the design can safely show.

Task 4: explain numbers without inventing a reason

These are fictional practice figures. Campaign A: spend 20000, enquiries 100, qualified enquiries 10. Campaign B: spend 20000, enquiries 60, qualified enquiries 24. Calculate cost per enquiry and cost per qualified enquiry. Separate arithmetic observations from possible explanations. Do not infer revenue, profit or the cause of the difference.

The checked arithmetic is straightforward: A costs ₹200 per enquiry and ₹2,000 per qualified enquiry. B costs about ₹333.33 per enquiry and ₹833.33 per qualified enquiry. B has the higher cost per enquiry and lower cost per qualified enquiry under the same qualification rule.

That does not prove which audience, message or channel caused the difference. It does not establish profitability. A good review keeps those missing facts visible rather than turning a table into a confident story.

Use the calculator to check the arithmetic independently. For a real report, verify the figures against the source records and use the same period and cost definitions.

Task 5: ask for a review, then do the actual review

Review this draft against the supplied brief. List each factual claim and whether the brief supports it. Flag generic wording, unclear next actions and missing information. Suggest edits, but do not add new facts.

This can produce a useful checklist, but it is still another AI output. Read the draft yourself. Check the source, the offer, the audience, the calculation and the action. Test any actual link or form. A second prompt does not establish that a fact is true.

Before publishing, ask: is the information accurate, useful and allowed to be shared? Does the page answer the visitor’s question? Can someone take the promised next step? Keep a short note of what you verified and what remains uncertain.

Keep the draft, the checks and the revision

Save the original brief, your prompt, the actual output from the tool, the issues you found and the edited version. That record shows your contribution. If you later share it in a portfolio, distinguish your work from the generated draft.

Google’s guidance on AI-assisted website content puts the focus on useful, accurate work and warns against producing many pages without added value. Generating more text is not a substitute for answering a real question. Apply that idea to your own practice: finish one useful output before starting ten more.

Try the AI workflow examples and then connect the exercise to the full learning plan. The aim is to become better at briefing, checking and improving work, not simply faster at requesting it.