Loading lesson
Loading lesson…
Loading lesson…
The meta-skill underneath all prompt engineering: structured thinking before AI contact.
Lesson · 20 min
Sign up to save your progress
Sign Up FreeYou have the framework. Now you need the reps. In this lesson, you will work through three real marketing scenarios -- without touching AI at all. The first two come with completed sample worksheets to study. The third is yours: you will fill out the worksheet right on this page and get it graded. The whole point is to build the muscle of thinking before prompting, so keep your AI tools closed.
Scenario 1: LinkedIn Thought Leadership Series
You are the content lead at a fintech startup that just raised a Series A. Your CEO is sharp but not a natural writer. She has strong opinions about how traditional banks are failing small businesses, and she wants to build an audience on LinkedIn. Your job: plan a 4-post LinkedIn thought leadership series for her.
The company's product is a cash flow management platform for businesses with 10-50 employees. Your CEO's key thesis is that small business owners should not need a finance degree to understand their own cash position.
Before you even think about what AI could write, fill out the worksheet. What is the specific outcome? What does success look like for a CEO thought leadership series on LinkedIn? What are the failure modes that would make these posts embarrassing or ineffective? What constraints do you know that AI does not?
See a sample completed worksheet for Scenario 1
Scenario 2: Re-engagement Email Campaign
You are the lifecycle marketing manager at a B2B SaaS company that sells project management software. You have a segment of 2,400 users who were active 6 months ago but have not logged in for the past 90 days. Your VP of Marketing wants you to "win them back."
You know from your data that the top three reasons users churned (from exit surveys) were: "too complicated for our small team," "we switched to [competitor]," and "the person who managed it left the company." Your product has shipped a simplified onboarding flow and a new "quick start" mode since most of these users were last active.
Fill out the worksheet. Think carefully about failure modes -- re-engagement emails are one of the most abused formats in marketing, and getting this wrong means these 2,400 users unsubscribe permanently.
See a sample completed worksheet for Scenario 2
Look at both completed worksheets. Before writing a single word of prompt, you now have:
This is the work. The prompt is just packaging.
No sample worksheet to peek at this time. Read the fresh scenario below, then fill out the Pre-Prompt Worksheet one section at a time, right on this page. When all four sections are filled, submit it and AI will grade your thinking against the same standard the sample worksheets were built on.
Scenario 3: Feature Launch Announcement
You are the only marketer at a 12-person HR-tech startup. The product team just shipped the most-requested feature in company history: automated interview scheduling that syncs with candidates' calendars. Your job: write the launch announcement email to your 800 existing customers.
Sales has warned you that at least 200 of those customers asked for this feature and were told "it's on the roadmap" for over a year, so the tone is delicate. The feature is included free on the two higher pricing tiers but costs an extra $29/month on the Starter tier. Your CEO wants the announcement to "make some noise." Your support lead is dreading a flood of how-do-I-turn-this-on tickets.
Fill out all four sections for Scenario 3. Be specific: real numbers, real constraints, real failure modes. When every section is filled, submit it for an AI grade.
What exactly gets produced, for whom, and what should it achieve? Name a number -- a measurable goal, not a vague deliverable like 'write an announcement email.'
What does a great version of this email look like? What would make you say 'ship it' without edits? Write criteria you could use to evaluate AI's output against a standard you set.
What are the specific ways this email could go wrong? List at least 3, specific and realistic. The scenario is full of ways this could blow up -- mine it.
What do you know that AI does not? Tone, pricing, audience history, timing. Constraints only you would know -- they close the gap between AI's assumptions and your reality.
Fill all 4 fields to enable grading.
3 minutes now, 30 minutes saved later
The worksheet takes 3 minutes. It saves 30 minutes of iteration. Every minute spent clarifying your thinking before prompting is worth ten minutes of revising AI output after. The marketers who get the best results from AI are not better at prompting -- they are better at thinking before they prompt.