AI maturity prompts for marketers
These prompts are based on the How AI Maturity Levels Transform Your Marketing Work Week webinar, published on the AI Lab by ActiveCampaign.
Get the prompt template
How to use these prompts
These prompts are pulled from the workflow Pam Didner walked through in the webinar: the same prompts she uses to climb from a generic ask to a file-grounded, validated piece of work.
Use with: ChatGPT, Claude, or Active Intelligence. This applies to every prompt below unless noted. Copy, paste, and swap in details where you see [BRACKETS]. Every AI model behaves a little differently, so review the first output and refine from there.
Prompt 1: The intermediate blog-post prompt
Best for: Writing a 1,000-word blog post that reflects your brand voice, addresses your audience, and finds an angle competitors haven’t covered.
Why this prompt: Pam used this to show marketers what changes between a beginner prompt and an intermediate one. Most marketers stop at “be more specific” by adding word count, tone, and audience. Pam’s intermediate rung adds files: brand guidelines, your own past posts, a competitor scan. That’s where the output stops sounding generic.
Primary prompt
Analyze the uploaded files: brand guidelines, keyword analysis, buyer persona, a few of my own blog posts relevant to this topic, and competitor blog posts on the same topic.
Then scan our website ([YOUR_WEBSITE_URL]) for messaging, and scan these competitor websites ([COMPETITOR_URL_1], [COMPETITOR_URL_2]) for differentiating opportunities.
Write a 1,000-word blog post about [TOPIC]. Use a friendly, educational tone that fits our product, [PRODUCT_NAME]. Target [AUDIENCE] in [VERTICAL] in [REGION]. Emphasize how [PRODUCT_NAME] helps them [SPECIFIC_OUTCOME]. End with a strong call to action to [DESIRED_ACTION].
Variables to fill in:
- [YOUR_WEBSITE_URL] — your homepage or product page URL (e.g., activecampaign.com)
- [COMPETITOR_URL_1], [COMPETITOR_URL_2] — two or three competitor sites you want to differentiate from
- [TOPIC] — the post’s subject (e.g., “the benefits of email marketing for small business”)
- [PRODUCT_NAME] — the product or service the post supports
- [AUDIENCE] — the buyer persona (e.g., “small business owners”)
- [VERTICAL], [REGION] — the slice of the audience you’re writing for
- [SPECIFIC_OUTCOME] — the benefit the product delivers
- [DESIRED_ACTION] — the CTA (e.g., “start a free trial”)
What to expect: A draft that uses your brand voice (because the model has your guidelines and past posts), addresses your specific audience (because you named the vertical and region), and finds at least one angle the competitors haven’t covered (because the model scanned them). It will still need editing, but the editing will be sharper than rewriting from scratch.
Follow-up prompt:
Compare this draft against the uploaded competitor posts. List three differentiating points my draft makes that the competitors don’t, and three points the competitors make that I’m missing. Tell me which of the missing points I should add and which I should deliberately skip.
Prompt 2: The GEO validation prompt
Best for: Scoring a blog post for AI search visibility and getting concrete fixes, not vague suggestions.
Why this prompt: After Pam built a draft she liked, her next move was to validate it for AI search. She was specific that “identify ways to improve” wouldn’t cut it; she demanded a 1–10 score and named steps. The point of this prompt is to get an audit a junior editor could action.
Primary prompt
I want this blog post optimized for generative engine optimization (GEO), meaning visibility in AI search tools like ChatGPT, Perplexity, and Google AI Overviews.
1. Score this draft for GEO on a scale of 1 to 10 and explain the score in two sentences.
2. Give me specific steps (not “identify ways”) to improve clarity, structure, and flow for AI search.
3. Using the keyword file I uploaded, suggest keyword optimization opportunities. List the keywords you’d add that aren’t already in my list, and explain why each one fits.
4. Strengthen my call to action. Show me your version and explain why it’s stronger than mine.
5. Give me three headline options for this post and tell me which one will perform best for [AUDIENCE].
Variables to fill in:
- [AUDIENCE] — the buyer persona the post is targeting
What to expect: A numbered audit you can action one item at a time. The score gives you a baseline; the steps tell you what to do; the keyword list tells you what’s missing; the CTA and headlines give you specific copy options. Pam’s rule: apply what holds up to your judgment, ignore the rest.
Follow-up prompt
Now rewrite the introduction and the call to action incorporating the changes you suggested. Keep the rest of the draft as I wrote it.
Prompt 3: The five-question validation pass
Best for: Pressure-testing a marketing plan, decision, or analysis before you commit to it.
Why this prompt: Pam called validation “the rung most marketers under-use.” This prompt is a single block that walks AI through her five validation moments (strategy, decisions, content, data, KPIs) so you can run it against any artifact you’re about to ship.
Primary prompt
I’m uploading [ARTIFACT_TYPE]: [BRIEF_DESCRIPTION].
Run five validation passes on it and answer each as a separate section:
1. Blind spots: What am I not seeing in this work?
2. Decision risks: If this work commits us to [SPECIFIC_DECISION], what are the potential risks I can’t see?
3. On-brand check: Where does this drift from the brand guidelines and messaging I uploaded?
4. Inconsistencies: Where do the numbers, claims, or recommendations contradict each other?
5. KPI second opinion: What other KPIs could reflect success for this work, beyond the ones I’m already tracking?
For each section, give me 2–4 specific items I can act on. Don’t pad. If a section has nothing to flag, say so.
Variables to fill in:
- [ARTIFACT_TYPE] — what you’re validating (e.g., “Q3 marketing plan,” “tool selection memo,” “campaign performance analysis”)
- [BRIEF_DESCRIPTION] — one or two sentences of context
- [SPECIFIC_DECISION] — the action this work would commit you to (e.g., “moving 30% of paid ad budget to email marketing”)
What to expect: Five clean sections of critique. You’ll usually agree with about half, disagree with a quarter, and need to think about the rest. That’s the point. Pam’s framing was that AI is a constructive critic, not the final answer.
Follow-up prompt:
Take the items I marked [ACT ON] from your validation pass and tell me, for each one, the smallest possible action I can take this week to address it.
Prompt 4: The analytics insight prompt
Best for: Getting a fast read on a campaign’s performance or comparing two campaigns side by side.
Why this prompt: Pam flagged analytics as the area marketers most under-use AI for. As she put it, “all of you can prompt, but the thing we’re not doing is using prompts to do analysis for us.” Her tip was that you don’t need a long, polished prompt. “Give me insight” works, as long as you’ve uploaded the data.
Primary prompt:
I’ve uploaded performance data for [CAMPAIGN_NAME].
Provide a summary of this campaign’s overall performance. Then give me three insights I should know: what’s working, what’s not, and what surprises you given the goals: [CAMPAIGN_GOALS].
Variables to fill in:
- [CAMPAIGN_NAME] — the campaign you’re analyzing
- [CAMPAIGN_GOALS] — what success looked like for this campaign (e.g., “drive 200 webinar registrations from paid social”)
What to expect: A short summary and three insights. The insights are the value, since surface-level summaries are already in your dashboard.
Follow-up prompt:
Compare this campaign against [CAMPAIGN_B_NAME], whose data I’m uploading now. Which one performed better against its stated goal, and why? Where are the lessons from the higher-performing campaign that I could apply to the lower-performing one?
Tips for better results
- Upload, don’t describe. When the prompt expects context (brand guidelines, past posts, performance data), upload the actual file. Describing what’s in it produces a flatter result than letting the model read it.
- Be specific about what you don’t want. Pam’s “don’t say identify ways, give me specific steps” is the move. Naming what you’ll reject upfront saves a refinement pass.
- Treat the model like a critic, not a co-author. Apply the suggestions that hold up to your judgment, leave the rest. The point of validation is to surface blind spots, not to outsource the decision.
- Give yourself permission to start small. Pam’s parting advice was that no one is ahead. Twenty minutes a day with one prompt open beats a heroic weekend overhaul that doesn’t stick.
See these prompts in action
Want to see Pam run each of these prompts live? The full webinar walks through how she phrased each ask, what came back from the model, and how she refined the follow-ups in real time, including the moment she pushed Camille on the GEO score and named exactly why “identify ways” wasn’t good enough.
Related
More data from the AI Lab.