Strategic questioning prompts for autonomous marketing
This resource is based on Why "perfect prompts" won't save your marketing—and what will, featuring Bob Pearson, published on the AI Lab by ActiveCampaign.
Get the prompt template
How to use these prompts
These are ready-to-use prompts pulled from Bob Pearson’s approach in the linked article. Copy, paste, and swap in details where you see [BRACKETS]. Every AI tool and model behaves a little differently, so treat what comes back as a starting point — review the output and refine from there.
Prompt 1: Market share driver analysis (strategic questioning)
Best for: Getting an AI research tool to surface the top questions worth investigating, rather than a generic blog draft.
Use with: Any LLM (ChatGPT, Claude, Perplexity)
Prompt 1
What factors are driving recent changes in market share in the [INDUSTRY] category — innovation, pricing, distribution, regulations, new competitors, consumer behaviors? Please list the top three questions in each area, and what would make [TARGET_AUDIENCE] care about this topic right now.
Variables to fill in:
- [INDUSTRY] — the category you sell into (e.g., “B2B email automation”, “healthcare SaaS”)
- [TARGET_AUDIENCE] — the specific buyer or user group (e.g., “heads of demand gen at Series B startups”)
What to expect: A structured set of investigative questions, not finished copy. Use the output as a research agenda for your next campaign brief or interview round.
Follow-up prompt
Rank those questions by how likely each is to reveal a non-obvious insight for [TARGET_AUDIENCE]. Then suggest one proprietary data source we would need to answer each question with confidence.
Prompt 2: PhD-student research brief (vs. Google-style query)
Best for: Moving from shallow search to deep research. Pearson’s rule: “Do you treat AI like Google, or do you treat AI like you have a PhD student next to you?”
Use with: Perplexity (best for citation-rich output) or Claude (for longer analytical synthesis)
Prompt 2
Find investigators who have run [TYPE_OF_STUDY] in [FIELD] in the last three years, who appear to be leaders in the field, whose work reached [STAGE_OR_MILESTONE], whether they succeeded or failed. Tell me what they learned, and show me exactly how many [RELEVANT_METRIC] they recorded per [UNIT].
Variables to fill in:
- [TYPE_OF_STUDY] — trials, customer research projects, pricing tests, campaigns, etc.
- [FIELD] — your category or topic
- [STAGE_OR_MILESTONE] — the threshold that matters (e.g., phase three, Series B, 100k subscribers)
- [RELEVANT_METRIC] — what counts as volume in your world (participants, customers, pipeline, opens)
- [UNIT] — the grouping (center, company, cohort, region)
What to expect: A detailed response in 5–15 minutes (Perplexity’s typical turnaround). You will not have everything you need, but you will know what is left for a human colleague to fill in.
Follow-up prompt
Here is what you returned. Here is what my team already knows: [PASTE_NOTES]. List the three biggest gaps between those two, and draft the exact questions I should send to a subject-matter expert to close each gap.
Prompt 3: Tailored content from a single source of intelligence
Best for: Producing variations of the same core content for different segments, locations, or languages — without re-briefing from scratch every time.
Use with: Claude (strong at document-based analysis) or ChatGPT
Prompt 3
Using the intelligence brief below, generate content for each of the following [SEGMENT_TYPE]: [LIST_OF_SEGMENTS]. For each, adjust tone, examples, and proof points so the message fits the segment. Keep the core offer and call to action consistent across all versions.
Intelligence brief:
[PASTE_BRIEF]
Segment-specific guardrails:
- [GUARDRAIL_1]
- [GUARDRAIL_2]
Variables to fill in:
- [SEGMENT_TYPE] — health systems, regions, languages, customer tiers, etc.
- [LIST_OF_SEGMENTS] — the specific segments (e.g., “Denver, Cleveland, Miami” or “Spanish, Mandarin, English”)
- [PASTE_BRIEF] — the proprietary intelligence you built from your own data pond
- [GUARDRAIL_1/2] — rules that must hold (e.g., “never promise a specific recruitment number”, “regulated claims only”)
What to expect: Multiple variations you can edit down, rather than one generic draft you have to rewrite for each segment. This is the “system generates content for every investigator with slight tweaks” pattern Pearson describes.
Follow-up prompt
Review the drafts against the guardrails above. Flag any line that stretches a claim, and propose a tighter version that keeps the meaning but respects the rule.
Prompt 4: Define your AI “team” — role briefs for each tool
Best for: Deciding which AI tool handles which kind of work, instead of defaulting to one for everything. Pearson treats each tool like a team member with a job title.
Use with: Any LLM — this is a workflow-planning prompt for yourself or your team
Prompt 4
I want to assign AI tools to roles on my marketing team. For each tool below, write a one-paragraph “job description” based on its actual strengths and the kind of work I should send its way. Then list one type of task I should NOT send to that tool.
Tools I currently have access to:
- [TOOL_1]
- [TOOL_2]
- [TOOL_3]
- [TOOL_4]
My current marketing priorities:
- [PRIORITY_1]
- [PRIORITY_2]
- [PRIORITY_3]
Variables to fill in:
- [TOOL_1‑4] — your actual AI stack (e.g., ChatGPT, Perplexity, Pi, Claude)
- [PRIORITY_1‑3] — what your team is focused on this quarter
What to expect: A role map you can share with your team so nobody wastes time forcing the wrong tool onto the wrong task. Pearson uses ChatGPT as his “smart assistant”, Perplexity as his “head of market research”, Pi as his “social media director”, and Claude for longer document-based analysis.
Follow-up prompt
Now write a quick decision tree: “If the task is X, send it to tool Y. If the task is Z, send it to tool W.” Keep it to one page.
Tips for better results
- Write questions, not prompts. A good prompt often starts with a question you genuinely want answered, not a command to “write me a blog post.”
- Feed AI your own data. Public training data puts you in the same ocean as your competitors. Proprietary sales data, research notes, and customer intelligence are what produce different answers.
- Build the follow-up into your workflow. Pearson’s pattern is “Perplexity returns a detailed response, then a human colleague fills in the gaps.” The AI does not finish the job alone.
- Measure precision, not just time saved. Pearson’s rule: if you save time and waste what you gained, it is all for naught.
Related
More data from the AI Lab.