Landing Page Builder Prompt Pack.

# Landing Page Builder Prompt Pack

Five prompts, run in order for a new page or individually for fixes. Each prompt tells you
exactly what to paste in (Inputs), gives you the full prompt text to copy, and pins the
output to a table format you can act on. Fill every placeholder before running.

Works in Claude, Cursor, or any capable AI chat. No account access required — this pack is
copy-in, copy-out. Nothing in it publishes anything. If you run these prompts through an
agent rather than a chat session, the agent operates under your Campaign AGENTS.md scope
rules: pasted copy is a read-only source and every draft is RECOMMEND-tier output.

---

## Session guardrails — paste this once, before Prompt 1

Paste this block at the top of your AI session before running any prompt in the pack.
It sets the operating rules every prompt below relies on. Fill {{APPROVER_ROLE}} (the
human who signs off on sensitive copy and launch — e.g. marketing lead, founder, legal).
If you keep a Campaign CLAUDE.md, paste its voice rules and never-say list after this
block so drafts start on-brand.

```
OPERATING RULES FOR THIS SESSION

1. DRAFT MODE ONLY. Everything you produce is a draft. Never describe copy as
   published, applied, or live. Never suggest auto-publishing anything.
2. NO INVENTED PROOF. Do not fabricate statistics, testimonials, customer
   names, review counts, awards, or performance claims. If a line needs
   evidence I have not pasted into this session, write the line and tag it
   [NEEDS SOURCE].
3. NO PRICING IMPROVISATION. Only state prices, discounts, or guarantees I
   paste verbatim. If pricing copy is needed and missing, tag [NEEDS PRICING].
4. APPROVAL FLAGS. Tag any draft that touches money, guarantees, or legal and
   compliance-sensitive claims (health, finance, "best"/"#1" superlatives,
   competitor comparisons) as [APPROVAL REQUIRED: {{APPROVER_ROLE}}].
5. CHANGE LOGGING. When I tell you a draft was accepted, record it as one
   Trigger / Action / Impact row:
   Trigger (which prompt or finding) | Action (what changed, where, by whom) |
   Impact (page section affected, draft or live).

Confirm you will follow these rules, then wait for my first prompt.
```

---

## Prompt 1 — Message-Match Audit

Diagnoses the gap between what your ad promises and what your page delivers.
Read-only: it produces findings and draft fixes, it changes nothing.

**Inputs:**
- {{AD_COPY}} (every live headline, description, display path, and ad asset, pasted verbatim)
- {{PAGE_COPY}} (the current landing page text, top to bottom in page order, hero first — include button and form labels)
- {{PRIMARY_KEYWORD}} (the search keyword or audience hook the ad targets)
- {{OFFER}} (the specific thing the page asks visitors to take — e.g. free audit, demo, trial)
- {{CONVERSION_ACTION}} (the single action that counts as a conversion — e.g. form submit, booked call)

**Prompt:**

```
You are auditing message match between a paid ad and its landing page. Work
read-only: diagnose and propose, do not rewrite the page wholesale.

CONTEXT
- Primary keyword: {{PRIMARY_KEYWORD}}
- Offer: {{OFFER}}
- Conversion action: {{CONVERSION_ACTION}}

AD COPY (verbatim):
{{AD_COPY}}

LANDING PAGE COPY (verbatim, in page order):
{{PAGE_COPY}}

TASK
1. Extract every promise the ad makes: offer, audience callout, keyword
   phrasing, urgency or price framing (if any), and call to action.
2. For each ad promise, quote the exact page line that delivers it, or mark
   it MISSING. Quote — do not paraphrase.
3. Reverse check: flag anything prominent on the page (hero headline, primary
   CTA, offer framing) that the ad never set up.
4. Rate each gap: BLOCKER (visitor will feel misled or lost), MAJOR (weakens
   the promise), MINOR (polish).

OUTPUT
Table 1 — Ad-to-page match:
| Ad promise (quoted) | Page delivery (quoted line or MISSING) | Match: Exact / Partial / Missing | Severity | Draft fix |

Table 2 — Page elements the ad never set up:
| Page element (quoted) | Why it breaks scent | Draft fix |

Then a one-line verdict — STRONG MATCH / PARTIAL MATCH / MISMATCH — and the
three fixes to make first, ranked by severity.

RULES
- Label every proposed fix DRAFT.
- For any fix touching price, guarantees, or legal claims: propose the
  direction only and tag [APPROVAL REQUIRED: {{APPROVER_ROLE}}] instead of
  writing final copy.
```

**Expected output:** two tables plus a ranked top-3 fix list, as specified in the prompt.

**Guardrail:** this prompt never edits the live page. Fixes ship as drafts; anything
flagged APPROVAL REQUIRED needs sign-off from {{APPROVER_ROLE}} before a human applies it.
Log each applied fix using the change-log rule at the end of this pack.

---

## Prompt 2 — Page Outline and Wireframe Copy

Produces a full section-by-section page draft for a stated offer and audience — real copy
in every section, plus an explicit list of what only you can supply.

**Inputs:**
- {{OFFER}}, {{PRIMARY_KEYWORD}}, {{CONVERSION_ACTION}} (as defined in Prompt 1)
- {{AUDIENCE}} (who lands here: role, situation, what they were searching or scrolling when they clicked)
- {{TRAFFIC_SOURCE}} (where the click comes from — e.g. Google Search, Meta, LinkedIn — it changes how much context the visitor arrives with)
- {{PROOF_ASSETS}} (testimonials, client names or logos you may use, case study summaries — paste the actual text; write NONE if you have none)
- {{TOP_OBJECTIONS}} (the 3-5 most common reasons prospects hesitate, in their words)
- {{BRAND_VOICE}} (two or three adjectives plus one sentence you consider on-voice)

**Prompt:**

```
Draft wireframe copy for a landing page. Every section gets real draft copy,
not lorem ipsum or notes-to-self.

CONTEXT
- Offer: {{OFFER}}
- Audience: {{AUDIENCE}}
- Primary keyword: {{PRIMARY_KEYWORD}}
- Traffic source: {{TRAFFIC_SOURCE}}
- Conversion action: {{CONVERSION_ACTION}}
- Brand voice: {{BRAND_VOICE}}
- Proof assets available (use ONLY these): {{PROOF_ASSETS}}
- Known objections: {{TOP_OBJECTIONS}}

TASK
Draft the page as a numbered section table, top to bottom:
1. Hero: headline, subhead, primary CTA button label.
2. Context: name the visitor's problem or situation in their own words.
3. How it works: 3-4 steps or components of {{OFFER}}.
4. Proof: built ONLY from {{PROOF_ASSETS}}. If NONE, design the section
   structure anyway and tag every proof line [NEEDS SOURCE].
5. Objection handling: address the top items from {{TOP_OBJECTIONS}}.
6. Final CTA: restate offer and CTA, and say what happens after conversion.
7. Form/CTA microcopy: field labels, button text, one reassurance line.

OUTPUT
| # | Section | Job it does | Draft copy | What I must supply |

After the table, list every [NEEDS SOURCE] and [NEEDS PRICING] tag in one
checklist so nothing unverified slips into the build.

RULES
- The hero must reflect {{PRIMARY_KEYWORD}} and the framing of the ad or post
  driving traffic from {{TRAFFIC_SOURCE}}.
- One conversion action; do not add secondary CTAs.
- No invented statistics, names, or endorsements anywhere.
```

**Expected output:** one wireframe table (7 rows) plus a consolidated NEEDS SOURCE / NEEDS
PRICING checklist.

**Guardrail:** the draft is a build spec, not a live page. Nothing goes into your CMS
or page builder until the NEEDS SOURCE checklist is cleared and {{APPROVER_ROLE}} has
signed off on any pricing, guarantee, or comparison copy.

---

## Prompt 3 — Headline and Subhead Variants

Generates ten headline + subhead pairs tied to the keyword the visitor just searched or
the hook they just clicked — with a risk flag on every variant.

**Inputs:**
- {{PRIMARY_KEYWORD}}, {{OFFER}}, {{AUDIENCE}}, {{BRAND_VOICE}} (as defined above)
- {{AD_HEADLINE}} (the exact headline of the ad this page receives traffic from)
- {{BANNED_PHRASES}} (words or claims you never use — competitor names, superlatives legal has rejected, category cliches; write NONE if none)
- {{REAL_DEADLINE}} (an actual expiry date or capacity limit if one exists; write NONE and urgency variants will be skipped)

**Prompt:**

```
Write 10 headline + subhead pairs for a landing page.

CONTEXT
- Primary keyword: {{PRIMARY_KEYWORD}}
- Ad headline the visitor just clicked: {{AD_HEADLINE}}
- Offer: {{OFFER}}
- Audience: {{AUDIENCE}}
- Brand voice: {{BRAND_VOICE}}
- Banned phrases: {{BANNED_PHRASES}}
- Real deadline or limit (NONE means skip urgency angles): {{REAL_DEADLINE}}

TASK
Cover these angles, at least one pair each:
- Keyword echo: mirrors {{PRIMARY_KEYWORD}} and {{AD_HEADLINE}} closely.
- Outcome: the end state the visitor wants.
- Objection flip: leads with the hesitation, then answers it.
- How-it-works: the concrete mechanism of {{OFFER}}.
- Question: the question the visitor is already asking.
- Urgency: ONLY if {{REAL_DEADLINE}} is not NONE, and only using that fact.

Headlines under 10 words. Subheads under 20 words and adding new information,
never restating the headline.

OUTPUT
| # | Angle | Headline | Subhead | Keyword present? (Y/N) | Risk flag |

Risk flag values: CLEAN (safe to test), NEEDS SOURCE (implies proof I have
not supplied), APPROVAL REQUIRED (touches pricing, guarantees, comparisons,
or regulated claims).

Close with your top 3 picks for a first test and one sentence each on why.

RULES
- No superlatives ("best", "#1", "fastest") unless I pasted evidence.
- No fabricated urgency or scarcity.
- Nothing from {{BANNED_PHRASES}}.
```

**Expected output:** one 10-row table plus a top-3 shortlist with rationale.

**Guardrail:** variants are test candidates, not replacements. Anything flagged NEEDS
SOURCE or APPROVAL REQUIRED is blocked from testing until cleared by {{APPROVER_ROLE}}.
If you A/B test headlines with an automation tool, create the test in draft or paused
state by default and require a human to start it.

---

## Prompt 4 — Objection and FAQ Section

Turns real objections from sales calls, chat logs, and support tickets into an FAQ section
that answers what visitors are actually worried about — using only policies you actually have.

**Inputs:**
- {{SALES_OBJECTIONS}} (verbatim objections from calls, CRM notes, chat, or tickets — paste 5-10, exactly as prospects said them)
- {{POLICY_TEXT}} (your actual refund, guarantee, contract, cancellation, and data-handling terms, pasted verbatim; write NONE for any you lack)
- {{OFFER}}, {{CONVERSION_ACTION}} (as defined above)

**Prompt:**

```
Build an objection-handling FAQ section for a landing page from real sales
objections.

CONTEXT
- Offer: {{OFFER}}
- Conversion action: {{CONVERSION_ACTION}}
- Verbatim objections from real prospects:
{{SALES_OBJECTIONS}}
- Official policy text (the ONLY source you may cite for commitments):
{{POLICY_TEXT}}

TASK
1. Group duplicate objections; keep the most vivid verbatim phrasing.
2. Rewrite each as an FAQ question in the customer's own words — "What if it
   doesn't work for us?", not "What is our satisfaction commitment?".
3. Draft an answer of 80 words or less: direct first sentence, then the
   reason to believe. Cite {{POLICY_TEXT}} where relevant.
4. Mark each answer's claim status.

OUTPUT
| Objection (verbatim) | FAQ question | Answer draft | Claim status | Placement |

Claim status values: SUPPORTED (backed by pasted policy), NEEDS SOURCE
(needs proof I have not supplied), APPROVAL REQUIRED (commits money, a
guarantee, or a legal/data promise).
Placement values: FAQ section / near CTA / near pricing.

Close with the recommended question order (most conversion-critical first)
and any objection the main page copy should absorb rather than the FAQ.

RULES
- Never invent a guarantee, refund term, timeline, or security commitment
  that is not in {{POLICY_TEXT}}.
- If {{POLICY_TEXT}} is NONE for an objection, state what policy would need
  to exist and tag the answer [NEEDS SOURCE] — do not improvise one.
```

**Expected output:** one table (one row per unique objection) plus an ordering
recommendation.

**Guardrail:** every answer that commits your company to anything — refunds,
guarantees, timelines, data handling — carries APPROVAL REQUIRED and must be verified
against real policy by {{APPROVER_ROLE}} before publishing. An FAQ that promises what your
contract does not is a liability, not a conversion asset.

---

## Prompt 5 — Pre-Launch CRO Checklist Review

A structured final pass before the page goes live. The model checks; a named human
launches. Run it on the staging or draft version, never as post-launch cleanup.

**Inputs:**
- {{FINAL_PAGE_COPY}} (the complete near-final page text in order, including form fields, button labels, and footer trust elements)
- {{AD_COPY}} and {{PRIMARY_KEYWORD}} (as defined in Prompt 1)
- {{FORM_FIELDS}} (every field on the form and whether each is required)
- {{TRACKING_PLAN}} (the events you intend to fire — e.g. form submit, thank-you pageview — and the tool that fires them)
- {{PAGE_URL}} (the URL the page will go live at)

**Prompt:**

```
Run a pre-launch review of this landing page. You are the reviewer, not the
approver: your output is a findings report, and a named human decides launch.

INPUTS
- Final page copy: {{FINAL_PAGE_COPY}}
- Ad copy it must match: {{AD_COPY}}
- Primary keyword: {{PRIMARY_KEYWORD}}
- Form fields and required status: {{FORM_FIELDS}}
- Tracking plan: {{TRACKING_PLAN}}

CHECKS — evaluate each and quote your evidence from the inputs:
- Message match: hero reflects {{PRIMARY_KEYWORD}} and the ad promise; offer
  and CTA match the ad.
- Clarity: a first-time visitor can say what this is, who it is for, and
  what to do next from the hero alone.
- CTA: one conversion action; the button label says what happens next; the
  CTA repeats after proof and FAQ.
- Friction: every required form field is justified by the offer's value; no
  field asks for data the page never explains needing.
- Proof integrity: every statistic, name, and testimonial has a source; no
  [NEEDS SOURCE] or [NEEDS PRICING] tags remain anywhere.
- Trust and compliance: pricing and guarantee lines match approved policy;
  a privacy link sits near the form; no unapproved superlatives or
  competitor comparisons.
- Tracking: {{TRACKING_PLAN}} covers the conversion action; test and staging
  traffic is excluded or noted as a known gap.

OUTPUT
| Check | Result: Pass / Flag / Fail | Evidence (quoted) | Fix | Launch blocker? (Y/N) |

Then the verdict: READY / READY WITH FIXES (list them) / NOT READY (list the
blockers). End with this sign-off block for the human approver to complete:

  LAUNCH SIGN-OFF
  Verdict accepted by {{APPROVER_ROLE}} — name: ______  date: ______
  Blockers resolved or explicitly waived: ______
  Log entry — Trigger: pre-launch review | Action: page published to
  {{PAGE_URL}} | Impact: live traffic begins on this page
```

**Expected output:** one checklist table, a verdict line, and a fill-in sign-off block.

**Guardrail:** the verdict is advisory. The page does not go live on a READY verdict —
it goes live when {{APPROVER_ROLE}} completes the sign-off block. Keep the completed block
with your log: it is the audit record of who launched what, when, and which flags were waived.

---

## Change-log rule (applies to the whole pack)

Whenever a draft from any prompt is applied to a real page, record one
Trigger / Action / Impact row:

| Trigger | Action | Impact |
|---|---|---|
| Prompt 1 audit: hero missing {{OFFER}} | Hero headline replaced (by: name, date) | {{PAGE_URL}} hero section — live |
| Prompt 4: refund objection unanswered | FAQ item added after policy verified | {{PAGE_URL}} FAQ section — live |

Keep the log wherever your team already keeps campaign records. The habit matters more
than the tool: every change traceable to a trigger, every action attributable to a person,
every impact stated before it happens.

---

## Placeholder reference

| Placeholder | What it is |
|---|---|
| {{APPROVER_ROLE}} | Human who signs off on sensitive copy and launch |
| {{AD_COPY}} | Live ad headlines, descriptions, and assets, verbatim |
| {{PAGE_COPY}} | Current page text, top to bottom |
| {{PRIMARY_KEYWORD}} | Keyword or hook the ad targets |
| {{OFFER}} | The specific thing visitors get |
| {{CONVERSION_ACTION}} | The one action that counts as a conversion |
| {{AUDIENCE}} | Who lands on the page and why |
| {{TRAFFIC_SOURCE}} | Channel the click comes from |
| {{PROOF_ASSETS}} | Real testimonials, logos, case studies (or NONE) |
| {{TOP_OBJECTIONS}} | Common hesitations, in prospects' words |
| {{BRAND_VOICE}} | Voice adjectives plus one on-voice sentence |
| {{AD_HEADLINE}} | Exact headline of the connected ad |
| {{BANNED_PHRASES}} | Words or claims you never use (or NONE) |
| {{REAL_DEADLINE}} | Actual expiry or capacity limit (or NONE) |
| {{SALES_OBJECTIONS}} | Verbatim objections from calls and tickets |
| {{POLICY_TEXT}} | Actual refund/guarantee/contract terms (or NONE) |
| {{FINAL_PAGE_COPY}} | Complete near-final page text |
| {{FORM_FIELDS}} | Every form field plus required status |
| {{TRACKING_PLAN}} | Events you will fire and the tool that fires them |
| {{PAGE_URL}} | Where the page goes live |