GUIDES · REFERENCE

The AI Marketing Prompt Library

Q: Do these prompts work with any AI model?

Yes — they're written to work with any general-purpose language model (ChatGPT, Claude, Gemini). Where a prompt needs a specific detail from your business, that's marked with brackets to fill in before using it.

Q: Why do the prompts ask for real customer language or past examples?

Because generic prompts produce generic output. Feeding a model real customer quotes, past top-performing copy, or specific numbers consistently produces more usable drafts than asking it to invent tone and detail from nothing.

Q: Should I use AI output directly, or always edit it?

Treat every output here as a first draft. The pattern that performs best across this whole cluster is AI draft, human edit — not either extreme alone.

Every prompt in this cluster's individual guides, collected in one place and organized by stage — research, strategy, creative, landing pages, campaigns, and analytics.

Rohan Alexander · 10 min read · Updated July 2026

The AI Marketing Prompt Library — key topics (AI Marketing guide by Zephra)

Where this sits: Marketing → Digital Marketing → Performance Marketing → AI Marketing → Prompt Library (a reference resource spanning every stage of the Zephra Marketing Operating System).

On this page Quick answer The one rule that makes these actually work Research & strategy prompts Creative & ad copy prompts Landing page prompts Google Ads prompts Meta / Facebook Ads prompts AEO & GEO prompts Analytics & reporting prompts Governance & agent-oversight prompts Common mistakes Before-you-prompt checklist FAQ How Zephra helps

Quick Answer

This library collects every reusable prompt across the AI Marketing cluster in one place, organized by the stage of the Marketing Operating System it supports. The single biggest lever on output quality isn't the prompt wording — it's feeding the model real customer language, past examples, and specific numbers instead of asking it to invent tone and detail from nothing.

The One Rule That Makes These Actually Work

Every prompt below performs meaningfully better with three additions: 2-3 real customer quotes or reviews, one or two examples of your own past best-performing copy or pages, and an explicit statement of what NOT to claim (guarantees, comparisons, numbers you can't back up). Paste these into any prompt below before running it — this single habit matters more than any specific wording choice.

Research & Strategy Prompts

"Given this business description [paste], draft one sentence each for Positioning and Offer before I write any ad copy."
"Summarize the 5 most common objections a prospective customer for [product/service] would have, based on these reviews: [paste reviews]."
"Review this campaign brief against the 14-stage Zephra Marketing Operating System and tell me which stages are missing or thin."
"Based on this description of our current marketing setup [paste], score us on the 5-stage AI Marketing Maturity Model, capability by capability."

Creative & Ad Copy Prompts

"Write 5 Facebook ad headline/description pairs for [offer], using these 3 real customer quotes as tone reference: [paste quotes]. Do not invent statistics or guarantees."
"Rewrite this AI-generated ad copy in a more conversational, less corporate tone, matching this example of our past copy: [paste example]."
"Write 3 Google Search ad variations for [offer] targeting [audience segment], keeping tone [reassuring/urgent/playful — pick one]."

Landing Page Prompts

"Draft a landing page outline for [offer] that repeats this exact ad headline: [paste headline], with one clear CTA repeated 2-3 times."
"Write an objection-handling section addressing this specific hesitation: [paste objection], for a [business type] landing page."
"Generate 5 FAQ entries for a landing page about [offer], based on these actual customer questions: [paste questions]."

Google Ads Prompts

"List the specific approval gates I should require before trusting an AI agent to reallocate ad budget automatically."
"Draft 5 negative keywords to exclude [irrelevant searcher type] from a [category] Google Ads campaign."
"Summarize whether this account [paste conversion volume/history] has enough data for Target CPA bidding, per the 15-30 conversion/month threshold."

Meta / Facebook Ads Prompts

"Given this ad's frequency [X] and click-through rate trend [paste], tell me if this creative needs refreshing."
"Suggest 3 genuinely different creative angles (not crops of one image) for [offer], to feed into Advantage+ testing."

AEO & GEO Prompts

"List 10 realistic prompts a prospective customer might ask an AI assistant when researching [category], to test our Share of Model against."
"Rewrite this page's opening paragraph to directly and factually answer [specific question], the way an AI assistant would want to cite it."

Analytics & Reporting Prompts

"Summarize last month's results in a way that answers: what should change in our Strategy stage based on this?"
"Turn this raw performance export [paste] into a plain-language summary a non-marketer stakeholder could understand in 3 sentences."

Governance & Agent-Oversight Prompts

"Draft a rollback test plan for a new marketing automation before we trust it with live budget."
"Summarize this week's agent decision log in plain language and flag anything that looks like it needs human review."
"Draft an approval workflow for marketing spend changes above [$X], suitable for a team with [N] stakeholders."

Common Mistakes

Using a prompt without adding real customer language or past examples first.
Asking for 15-20 variations instead of 3-5 genuinely different angles.
Never stating what NOT to claim, letting the model invent guarantees or comparisons.
Treating any single output as launch-ready without a human edit pass.

Before-You-Prompt Checklist

☐ Have 2-3 real customer quotes or reviews ready to paste in
☐ Have 1-2 examples of past best-performing copy or pages
☐ Know what NOT to claim before generating anything
☐ Plan for a human edit pass before anything launches

FAQ

Do these prompts work with any AI model?

Yes — they work with any general-purpose language model. Bracketed details should be filled in before use.

Why do the prompts ask for real customer language or past examples?

Generic prompts produce generic output — real quotes and past examples consistently produce more usable drafts.

Should I use AI output directly, or always edit it?

Treat every output as a first draft — AI draft, human edit is the pattern that performs best across this cluster.

HOW ZEPHRA HELPS

You can run these prompts manually in any AI chat tool.

Zephra runs this same prompt library internally, automatically pre-filled with your actual customer reviews, past top-performing copy, and compliance guardrails — so every draft starts from your real data rather than a blank prompt.

Start Free Audit →

Sources & Further Reading

WordStream — 2026 Google Ads Benchmarks Report — Current cross-industry CPC, CTR, conversion rate, and cost-per-lead benchmarks.
Google Ads Help — About Smart Bidding — Google's own documentation on how Smart Bidding uses conversion signals.

Figures and platform mechanics referenced in this guide are cross-checked against the above as of publication; confirm current figures directly with the source before making budget decisions.