AI prompts for marketing are structured instructions that turn your brand, data, and objectives into conversion‑ready assets. The best prompts define role, goal, inputs, constraints, steps, output format, and evaluation metrics—so you get on‑brand work you can ship and measure, not generic text you must rewrite.
Most “prompt lists” deliver clever words, not business outcomes. As adoption accelerates, leaders want prompts that reliably improve CTR, CVR, CAC, LTV, and pipeline velocity—not just produce more copy. According to McKinsey, generative AI could add trillions of dollars in value globally, but only when tied to real processes and measurable outputs (McKinsey). This guide gives growth leaders production‑grade prompt templates organized by funnel stage, channel, and KPI. You’ll also get a simple 7‑part structure to design high‑performing prompts, plus patterns to ground outputs in your data and guardrails. When you’re ready to scale beyond prompts to execution, we’ll show you how AI Workers can run these workflows end‑to‑end across your stack.
The core problem is that clever prompts create activity, while structured prompts create outcomes you can measure against growth KPIs.
If you’ve tried AI for marketing, you’ve seen the pattern: fun ideation, fast drafts, and then hours lost in rewrites, QA, formatting, and alignment with brand, ICP, and compliance. Generic prompts skip the essentials—your positioning, hard constraints, performance goals, and the format your team needs to launch. That gap turns AI from a growth multiplier into rework. Meanwhile, your quarterly targets don’t care how “creative” a prompt was; they care about MQL quality, paid media efficiency, funnel conversion, and time-to-test.
Growth leaders fix this by standardizing prompts around outcomes. They define role and goal, inject approved voice and data, require explicit steps, request a publish‑ready format, and include a self‑check. They also instruct the model to refuse speculation, flag missing inputs, and ask clarifying questions. The result is consistent, on‑brand assets that slot directly into your workflows—with evidence you can A/B test and attribute. Below are the patterns and 50+ templates you can paste into your tools today—and scale tomorrow with AI Workers that execute across systems (see AI Workers: The Next Leap in Enterprise Productivity).
The best way to design high‑performing prompts is to use a simple 7‑part pattern: Role, Goal, Inputs, Constraints, Steps, Output Format, and Evaluation.
A good AI prompt for brand voice defines tone, banned phrases, and approved examples, then asks the model to imitate patterns without copying.
Role: Senior Brand Copywriter for [Company] in [Category]. Goal: Produce copy that reflects our brand voice—confident, concise, evidence-led—without imitating exact phrasing. Inputs: Brand voice guide (below), 3 approved copy examples (below), ICP [Persona], Offer [Value Prop]. Constraints: - Do not invent product claims. Use only inputs provided. - Avoid passive voice, filler, and clichés. Ban: “cutting-edge,” “next-gen,” “unlock potential.” Steps: 1) Extract voice rules from the guide and examples. 2) Summarize the voice in 5 bullets. 3) Write 3 variants for [Asset Type] with [Character Limit]. 4) Add a 1-sentence rationale per variant tied to ICP pain and desired action. Output: Markdown with H2 “Voice Summary,” H2 “Variants,” each variant labeled V1–V3 with rationale. Evaluation: Check tone match (1–5), clarity (1–5), uniqueness (1–5). If any score <4, revise once.
You add data safely by sharing only necessary, non‑sensitive fields and instructing the model to treat them as the single source of truth and refuse to fabricate.
Data Policy: - Use ONLY the table below as source of truth. - If a required field is missing, ask up to 3 clarifying questions. - DO NOT infer, guess, or fabricate numbers. Inputs (sample): | Metric | Value | | CAC (last 90d) | $527 | | Avg LTV (3yr) | $5,900 | | Win Rate (SQO→Closed) | 24% | | Top Persona | VP Operations, 500–2,000 FTE | Task: - Propose 5 new paid social angles likely to improve MER, each backed by a 1–2 sentence rationale that cites the table.
The 7‑part prompt structure you should use is Role, Goal, Inputs, Constraints, Steps, Output Format, and Evaluation.
Role: [Who the AI is] Goal: [Business outcome, not activity] Inputs: [Brand voice, ICP, offer, data, examples] Constraints: [Compliance, banned terms, length, tone] Steps: [Numbered method the AI must follow] Output Format: [JSON/table/Markdown with exact fields] Evaluation: [Self-check rubric + one-shot revision rule]
The best top‑of‑funnel prompts generate high‑CTR creative and competitive SEO assets that are on‑brand, persona‑specific, and A/B‑test‑ready.
An effective high‑CTR ad prompt specifies persona pain, benefit, proof, character limits, and multiple variants for testing.
Role: Performance Marketer (Paid Social). Goal: Create high-CTR ad copy (primary text, headline, description) for [Persona], promoting [Offer]. Inputs: ICP pain: [Pain], Core benefit: [Benefit], Social proof: [Proof], Brand voice: [Guide]. Constraints: Primary Text ≤125 chars, Headline ≤30, Description ≤30. No jargon. Include 1 proof element. Steps: 1) Write 5 angles: Pain-breakthrough, Outcome, Risk-aversion, Category-contrast, Urgency. 2) For each, generate 2 variants (A/B). 3) Add a 1-sentence rationale per variant. Output Format: Table with columns: Angle, Variant, PrimaryText, Headline, Description, Rationale.
You generate SEO briefs by instructing AI to analyze competitors, cluster subtopics, specify H2/H3s, intent, FAQs, and sources to cite.
Role: SEO Strategist. Goal: Create a comprehensive content brief for the keyword “[Target Keyword]” with informational intent. Inputs: Persona [Persona], Product POV [POV], Brand voice [Guide]. Constraints: Avoid thin content. Target 1,800–2,400 words. Include 6–8 H2/H3 sections, PAA-style FAQs, and 3 credible sources (no speculation). Steps: 1) Define search intent and “must-include” subtopics. 2) Propose outline with H2/H3s and snippet-optimized first sentences. 3) Provide internal/external linking plan (anchor text + why). 4) Include a 10-point on-page checklist keyed to the brief. Output: Markdown brief with sections: Intent, Outline, Key Messages, Links, On-Page Checklist.
The strongest LinkedIn prompt asks for a contrarian POV, a hook, a 3‑point argument with a mini‑story, and a call to discussion.
Role: Growth Leader on LinkedIn. Goal: Draft 3 contrarian posts that challenge [Common Belief] with our POV. Inputs: ICP [Persona], Proof points [3 bullets], Voice [Confident, practical]. Constraints: 180–220 words. Hook first line. 1 brief story. End with a question. No emojis, no hashtags. Output: 3 posts, each with: Hook, Argument (3 bullets), Micro-story (3–4 lines), Close (question). Evaluation: Hook clarity (1–5) and uniqueness (1–5). Revise if any score <4.
You get PR angles opened by asking AI to map your data or POV to timely narratives and to produce email pitches with subject lines and quotes.
Role: PR Strategist. Goal: 5 timely story angles + reporter pitches for [Topic] tied to [News/Event]. Inputs: Unique data [Summary], Exec quotes [2 lines], ICP outlets [3 examples]. Constraints: Subject lines ≤60 chars; no hype; include one crisp stat from our data. Output: For each angle: 2 subject lines, 120-word pitch email, 1 exec quote, 1 visual idea.
The most effective mid‑funnel prompts personalize offers, clarify value, and orchestrate multi‑touch nurture aligned to persona pains and buying stage.
A great landing page prompt demands a clear hierarchy (headline, subhead, proof, CTA), risk reversals, objections, and scannable sections with wireframe labels.
Role: Conversion Copywriter. Goal: Create a high-converting landing page for [Offer] targeting [Persona] at [Stage]. Inputs: ICP pains [3], Outcomes [3], Social proof [logos/testimonial], Risk reversal [policy], Brand voice [guide]. Constraints: Headline ≤10 words; CTA verbs only; include 3 FAQs; ADA-compliant language. Output: Wireframed sections labeled H1, Subhead, Bullets, CTA, Social Proof, Feature/Benefit blocks, FAQs. Provide 2 headline/CTA variants.
You personalize 1:1 ABM emails by feeding firmographic signals and recent triggers, asking for a three‑part structure (relevance, insight, low‑friction ask), and banning fluff.
Role: ABM Strategist. Goal: Draft 3 personalized cold emails for [Account], [Title], referencing [Trigger/Event]. Inputs: Firmographics [size, industry, tools], Trigger [press/news/job post], Our POV [problem→impact→next step]. Constraints: 90–120 words. No “hope you’re well.” One ask: 15-min diagnostic. Include 1 quantified hypothesis tied to the trigger. Avoid attachments. Output: 3 emails + 1 subject line each + 1-line reason this matters, with bracketed merge fields clearly marked.
The best webinar prompt outlines a complete sequence—invites, reminders, calendar text, no‑show follow‑ups—each with value‑led hooks and next steps.
Role: Lifecycle Marketer. Goal: Create a 5-touch webinar sequence for [Topic] targeting [Persona]. Inputs: Value props [3], Speakers [2], Date/Time, Primary CTA [Register], Secondary CTA [Guide download]. Constraints: Each email ≤120 words; subject lines ≤48 chars; SMS reminder ≤140 chars. Output: Invite #1, Invite #2 (urgency), 24h reminder, 1h reminder (SMS+email), No-show follow-up (recording + CTA). Include preheader text and UTM plan.
The strongest bottom‑funnel prompts prioritize tests by impact, generate objection‑handling assets, and craft retention plays tied to usage signals.
You get prioritized CRO ideas by asking for hypotheses mapped to ICE (Impact, Confidence, Effort) with instrumentation and success thresholds.
Role: CRO Lead. Goal: Propose 12 A/B tests for [Page/Flow] to increase [KPI], prioritized by ICE. Inputs: Current metrics [CVR, bounce, scroll], Top objections [3], Device split [data]. Constraints: Each idea must include: Hypothesis, Variant spec, Primary metric, Guardrail metric, Est. uplift range, Required instrumentation, ICE score. Output: Table sorted by ICE with 12 rows; end with 3 quick wins <1 dev day.
You create ROI tools by requiring defensible assumptions, a transparent formula, and editable fields with a narrative one‑pager summary.
Role: Sales Enablement. Goal: Build an ROI model + 1-pager for [Solution] selling to [Persona]. Inputs: Benchmarks [3], Typical baseline metrics [CAC, throughput], Pricing [range]. Constraints: No hidden math. If an assumption is missing, ask questions first. Output: 1) JSON schema for calculator inputs/outputs with formulas. 2) 200-word executive summary one-pager (problem → impact → ROI) with 3 assumptions called out explicitly.
You generate retention plays by instructing AI to segment churn risks by behavior and produce stage‑specific outreach sequences with offers.
Role: Customer Marketing Manager. Goal: Draft a 3-step save sequence for customers with [Risk Signal] in [Segment]. Inputs: Plan type [ ], Tenure [ ], Usage delta [ ], Persona [ ]. Constraints: Keep tone supportive; no discounts in step 1; escalate value each step; include success metric per step. Output: Step 1 (insight + quick-win how-to), Step 2 (case study + office hours invite), Step 3 (custom roadmap + limited-time upgrade). Provide subject lines and CTAs.
The right analytics prompts force the model to define measurement, produce reproducible queries, and propose experiments with clear decision rules.
Prompts that help you measure CAC/LTV/MER reliably require definitions, data joins, lookback windows, anomalies, and decision thresholds.
Role: Growth Analyst. Goal: Calculate CAC, LTV (3-year), and MER by channel for last 90 days, flag anomalies, and recommend 3 budget reallocations. Inputs: Channel table [schema], Orders table [schema], Costs table [schema], Attribution rule [e.g., W-Shaped]. Constraints: Define each metric explicitly; state assumptions; no calculations without data lineage notes. Output: 1) Metric definitions. 2) Pseudocode or SQL outline to reproduce. 3) Table of results. 4) 3 reallocation recommendations with expected MER delta and risk notes.
You prompt AI for accurate SQL by providing the table schemas, sample rows, desired output columns, and explicit filtering and windowing rules.
Role: Marketing Data Analyst (BigQuery). Goal: Query GA4 export to get sessions, conversions, and revenue by source/medium/campaign for last 30 days. Inputs: Table schema: `analytics.events_*` with fields [event_timestamp, event_name, traffic_source.source, traffic_source.medium, traffic_source.campaign, ecommerce.purchase_revenue]. Constraints: Exclude internal IPs [list], only include sessions with event_name = "session_start". Timezone [ ], currency [ ]. Output: Valid BigQuery SQL, followed by a plain-English explanation of joins and filters.
You can standardize test briefs by asking AI to create a template with hypothesis, metrics, sample size, MDE, guardrails, and stop criteria.
Role: Experimentation Program Manager. Goal: Produce a 1-page A/B test brief and decision framework template. Constraints: Must include: hypothesis, success metric, guardrails, sample size/MDE calc placeholders, pre-registration checklist, and post-test learning prompts. Output: Fill-in-the-blank template + example filled for [Test Idea].
For a deeper view on turning prompts into shipped work (not docs), see Create Powerful AI Workers in Minutes and how teams move From Idea to Employed AI Worker in 2–4 Weeks.
Generic prompting yields drafts; AI Workers deliver outcomes by executing your process end‑to‑end across systems with auditability.
There’s a limit to what a single prompt can do inside a chat box. Growth teams win when prompts become roles—always‑on AI Workers that research, write, analyze, publish, tag, and log results in your stack. Instead of stopping at “give me 5 ad angles,” your AI Worker can generate variants, launch drafts to your ad platforms, apply UTMs, push experiments to your roadmap, and return a daily performance summary to Slack. That’s the shift from assistance to execution—where your team’s strategy compounds every week.
EverWorker lets you onboard AI Workers like employees: describe the job, attach your knowledge, and connect systems. No code. No technical complexity. If you can describe how the task is done, you can create an AI Worker to do it—at scale and with governance. Explore how AI Workers transform marketing capacity and let your team “do more with more” in AI Workers: The Next Leap in Enterprise Productivity, the platform advances in Introducing EverWorker v2, and a full library at the EverWorker Blog.
If these templates sparked ideas, the fastest win is to turn your highest‑leverage prompt into a reliable, measurable workflow. In one working session, we’ll map your instructions to an AI Worker that acts inside your systems and reports against your KPIs.
Prompts become growth when they’re standardized, measured, and embedded. Start by adopting the 7‑part structure, inject your brand/data, and require outputs you can publish or query immediately. Pilot 2–3 templates per funnel stage this week, tie them to a KPI, and review results in 7 days. When a pattern outperforms, promote it to an AI Worker that executes the entire job—so your team spends time on strategy, partnerships, and product‑led growth.
Two closing tips: 1) Protect trust with authenticity and governance; many CMOs are already investing in content authenticity and monitoring (Gartner). 2) Keep your people in the loop where judgment matters. As the research shows, the value is real—when it’s operationalized (McKinsey).
The best LinkedIn prompts specify a contrarian POV, a sharp hook, a 3‑point argument, a brief story, and a question close—plus tone, banned phrases, and word limits.
You keep outputs on‑brand by embedding a reusable brand voice block (tone, examples, banned phrases) in every prompt and requiring a 5‑bullet voice summary before writing.
You should test multiple leading models and choose per task (e.g., copywriting vs. analysis), then standardize on what performs best for your stack and governance needs.
You measure impact by predefining KPIs and success thresholds in the prompt, tagging assets with UTMs, running controlled A/B tests, and reviewing results with clear decision rules.