AI that pays for itself: a free course for marketing leaders

No signup. No technical experience. Use the tools you already have.

• Build. Create a repeatable weekly marketing intelligence brief.
• Review. Use a practical checklist before trusting or sharing AI-generated work.
• Measure. Calculate hours saved, estimated value and annualised return.

The curriculum

#	Lesson	Pillar	Minutes
1	Audit: pick work worth delegating	Audit	5
2	Implement: brief an AI worker like a team member	Implement	8
3	Test: review before you trust	Test	5
4	Optimise: prove the return	Optimise	6
5	Choose your AI worker	Choose	6

Lesson 1. Audit: pick work worth delegating

The best place to start with AI is not the newest tool. It is a recurring piece of work with a clear input, a reviewable output, and a measurable cost.

Chat answers a question or develops an idea. AI work takes an outcome, works through multiple steps, uses files or connected tools, and returns a finished deliverable. Automation repeats a proven workflow on a schedule or trigger.

Four criteria for a strong first workflow

1. It happens repeatedly.
2. The required inputs can be identified.
3. A person can review whether the output is good.
4. The current time or cost can be estimated.

Marketing examples. Weekly performance summary. Competitive intelligence brief. Campaign creative brief. Content repurposing plan. Search-term review. Customer-review analysis. Executive marketing update.

Do this now. Write down the workflow name, how often it happens, the current minutes per run, your estimated hourly value, and what a good output must contain. You need all five before the rest of the course means anything.

Lesson 2. Implement: brief an AI worker like a team member

AI workers perform better when they receive a real brief rather than a one-line command. Give them an outcome, context, inputs, constraints and a definition of done.

The workflow prompt

You are acting as a marketing operations analyst.

OUTCOME
Create an executive-ready weekly marketing intelligence brief that helps a
marketing leader decide what deserves attention next week.

CONTEXT
We sell [offer] to [target audience]. Our primary business objective is
[objective]. The reader is a marketing leader who needs the important signals,
not a channel-by-channel data dump.

INPUTS
Use the files, notes, metrics, and links I provide. Clearly identify any missing
information that would materially change your conclusions.

DELIVERABLE
Create:
1. A five-bullet executive summary.
2. The three most important performance signals.
3. Risks or anomalies that need investigation.
4. Three prioritized recommendations for next week.
5. A short list of sources and inputs used.

CONSTRAINTS
Do not invent metrics, customer facts, or source material. Separate confirmed
facts from your interpretations. Show calculations when comparing performance.
Do not publish, email, edit campaigns, or change an external system without my
approval.

DEFINITION OF DONE
The brief should be concise, decision-ready, supported by evidence, and readable
in under five minutes.

Practice dataset

Fictional company: Northstar CRM
Audience: Mid-market B2B marketing teams
Objective: Increase qualified demo requests

Current week vs. previous week:
- Paid search spend: $12,500 vs. $11,900
- Paid search leads: 210 vs. 195
- Paid search demos: 28 vs. 23
- LinkedIn spend: $8,100 vs. $7,900
- LinkedIn leads: 84 vs. 91
- LinkedIn demos: 14 vs. 12
- Email sends: 42,000 vs. 39,000
- Email click-through rate: 2.8% vs. 2.5%
- Email demos: 19 vs. 16
- Organic sessions: 18,400 vs. 17,600
- Organic demo requests: 37 vs. 34

Copy the workflow prompt, copy the practice dataset, open the assistant of your choice, paste the prompt, paste the data underneath it, and send.

Lesson 3. Test: review before you trust

AI can produce polished work that is still wrong. A marketing leader remains accountable for the facts, the decisions and any external action.

Review checklist

• Every important claim can be traced to an input or source.
• Calculations and percentage changes are correct.
• Facts and interpretations are clearly separated.
• Missing data is acknowledged.
• No confidential information was exposed unnecessarily.
• Recommendations connect to the stated business objective.
• The output does not take external action without approval.
• A qualified person has reviewed anything customer-facing or high impact.

The review prompt

Act as a skeptical marketing operations reviewer. Audit the draft against the
source material. Identify unsupported claims, incorrect calculations, missing
context, vague recommendations, privacy risks, and actions that require human
approval. Do not rewrite the brief yet. Return a prioritized issue list and
explain what evidence is needed to resolve each issue.

Safety. Connect only the files, accounts and applications needed for the current task. Keep approval gates around publishing, spending, deleting, sending or changing live systems.

Lesson 4. Optimise: prove the return

A workflow is valuable when it creates a better result, saves meaningful time, or improves a business outcome. Measure the baseline before scaling it.

Record four numbers: minutes before AI, minutes with AI including review, runs per month, and hourly value. Then:

• Minutes saved per run = minutes before minus minutes with.
• Hours saved per month = minutes saved per run, times runs, divided by 60.
• Estimated monthly value = hours saved per month, times hourly value.
• Estimated annual value = monthly value times twelve.

If minutes with AI are not lower than minutes before AI, the workflow has not paid for itself yet. That is a real result, and it is worth writing down.

The optimisation prompt

Review this workflow as an operations designer. The current outcome is
[outcome]. It originally took [baseline time] and now takes [assisted time],
including human review. The quality issues we observed were [issues]. Recommend
the three highest-impact changes to the brief, inputs, review process, or tool
setup. Preserve human approval before external actions.

Automate only when the workflow has run successfully more than once, inputs are consistent, the review process is defined, someone owns exceptions and failures, and the measured return justifies the complexity.

Lesson 5. Choose your AI worker

You do not need every AI platform. Choose the smallest tool that can complete the workflow safely and produce the required deliverable.

ChatGPT Work. Research and analysis that ends in a polished deliverable. Documents, spreadsheets, presentations and reports. Working across connected tools and files. Example: turn campaign data, planning notes and research into an executive-ready weekly brief.

Claude Cowork. Longer multi-step knowledge-work projects. Working with local files or browser workflows through the desktop app. Scheduled briefs and recurring operational workflows. Example: review paid-search terms, explain potential negatives, and wait for approval before any campaign change.

Perplexity Computer. Source-heavy research and competitive intelligence. Research across many web sources. Scheduled monitoring and recurring intelligence work. Example: monitor competitors, summarise positioning changes, and produce a cited monthly intelligence report.

For this course, start with ChatGPT Work. Consider Claude Cowork when the workflow depends heavily on local files or longer desktop sessions. Consider Perplexity Computer when cited external research and ongoing monitoring are the centre of the task.

Capabilities, pricing, plan access and mobile availability change frequently. Confirm current access before standardising a team workflow. Tool information last reviewed July 2026.

You now have a workflow, not just another AI tool

You selected work worth delegating, built the brief, added a review process, measured the potential return, and chose the right type of AI worker. The next step is to test the workflow with real, non-sensitive inputs and improve it based on evidence.

Take the free AI ROI Scorecard next, or book a call and we will walk through your first workflow with you.