Free Resource — Dreamerce
The Static Ad
Creative System
5 Claude agents. Every part of the creative workflow. Built from 15 years and 3,000+ statics produced for brands scaling on Meta.
Book a Free Call →
Before You Start
Read this first.
These prompts are built around a specific philosophy: AI raises the creative floor. It does not replace the judgment that sets the ceiling.
Most people using AI for ad creative are one-shotting statics and calling it a system. That produces volume without quality. The agents in this document work in sequence. Each one feeds the next. The output improves because the inputs are structured.
How to get the most from this
- Run the agents in order. Agent 01 (Research) feeds Agent 02 (Angles), which feeds Agent 03 (Briefs). Skipping ahead weakens the output.
- Replace every placeholder in pink with real information from your brand or client. Generic inputs produce generic output.
- These prompts work best in Claude (Sonnet or Opus). Paste into a fresh conversation for each new project.
- The system is built for static ads on Meta. The principles apply broadly but the language is calibrated for Facebook and Instagram placements.
- AI generates options. You apply judgment. Never skip the review step.
✦
Want this built and run for your brand?
Dreamerce delivers 40 concepts per month with 120 variations.
Full creative strategy ownership for ecommerce brands spending seriously on Meta. Research, angles, briefs, and execution — all handled.
Book a Free Call →
Contents
5 agents. One creative system.
Each agent handles a specific stage of the static ad workflow. Together they take you from raw customer data to a brief ready to execute — in under two hours.
System Logic
Why the sequence matters.
The biggest mistake in AI-assisted creative work is jumping to output before the inputs are right. Most people open a chat window and ask Claude to "write me some ad hooks." The output is generic because the input was generic.
This system works because research comes first. Angles come from the research. Briefs come from the angles. Headlines come from the briefs. Review closes the loop. Each stage makes the next stage better.
01 Research
→
02 Angles
→
03 Brief
→
04 Headlines
→
05 Review
→
Execute
01
Customer Research
Real language from Reddit, reviews, Facebook groups, and ad libraries. Not assumptions. The actual words your customers use to describe their problem before they know your product exists.
02
Angle Extraction
Distinct creative territories pulled from the research. Each angle is a different emotional entry point into the same product. Not variations — fundamentally different reasons to buy.
03
Brief Writing
A structured creative document that tells a designer exactly what to produce. The angle, hook, visual direction, copy, and CTA. Nothing left to interpretation.
04
Headline Writing
Multiple headline options per brief, written in real customer language. Not what sounds good in a meeting. What stops the scroll in a feed.
05
Ad Review
A structured performance diagnosis of existing ads. What is the hook doing? Where does attention drop? What angle is missing? What do you test next?
→
Execute
A brief ready to hand to a designer or take into production. No briefing meetings. No back and forth. A complete creative document built from real data.
The full cycle takes under two hours. That is not a replacement for strategic judgment. It is what strategic judgment looks like when the system is working.
01 / Agent One
Customer Research Agent
Most brands research their customer once, build a persona, and never update it. That persona gets stale. The angles built on top of it get stale. This is why creative performance plateaus even when spend stays constant.
This agent pulls language from where your customers actually talk. Reddit. Review sections. Facebook groups. The goal is not to understand your product. It is to understand how your customer understands their problem before they even know your product exists.
The brands I have seen test most successfully start from what customers say, not what marketers assume. Real language always wins.
You are a customer research analyst specialising in ecommerce and direct-to-consumer brands.
I need you to analyse customer language for a brand in the [PRODUCT CATEGORY] space. The product is [BRIEF PRODUCT DESCRIPTION] and it is sold to [TARGET CUSTOMER DESCRIPTION].
Your job is to identify the exact language customers use when describing:
1. THE PROBLEM — How do they describe the pain, frustration, or situation before finding a solution? What words and phrases appear most? What emotions are present?
2. THE FAILED ALTERNATIVES — What have they already tried? What did not work and why? What language do they use when describing disappointment?
3. THE DESIRED OUTCOME — Not just the product benefit. What does life look like after the problem is solved? What do they gain socially, emotionally, practically?
4. THE HIDDEN DRIVERS — What are they embarrassed to admit they want? What deeper identity or status need is underneath the surface purchase?
5. THE OBJECTIONS — What stops them from buying? What would they say to a friend to explain why they are hesitant?
Analyse this using the following sources as your reference frame:
- Reddit communities related to [RELEVANT SUBREDDITS]
- Amazon review language for competing products
- Facebook group discussions in [RELEVANT COMMUNITIES]
- Product review sections on competitor sites
Structure your output as a research brief. Use direct quotes and specific language patterns, not paraphrased summaries. I want the raw words, not your interpretation of them.
Using the customer research brief above, map the awareness levels of the target customer for [BRAND NAME].
Classify where the majority of potential buyers currently sit across these five stages:
UNAWARE — They have the problem but have not identified it yet. What are they experiencing day to day? What language would a headline need to use to stop them in a feed?
PROBLEM AWARE — They know they have the problem but do not know solutions exist. What words describe their current frustration?
SOLUTION AWARE — They know solutions exist but have not decided on one. What are they comparing? What objections are active at this stage?
PRODUCT AWARE — They know your product exists but have not bought. What is holding them back specifically?
MOST AWARE — They are familiar with the brand and just need a reason to act now. What would push them over the line?
For each stage provide:
- The emotional state of the customer
- The language pattern that fits that state
- The type of static ad hook most likely to land (problem-agitation, social proof, outcome, mechanism, comparison, or identity)
- One example headline written in real customer language
Focus on the two stages where you believe the majority of the brand's untapped opportunity sits. Explain your reasoning.
Analyse the current creative landscape for brands competing with [BRAND NAME] in the [PRODUCT CATEGORY] space.
The main competitors are: [LIST 3-5 COMPETITOR BRAND NAMES]
Help me identify:
1. THE DOMINANT ANGLES — What creative territory is everyone already occupying? What are the most common hooks, claims, and emotional frames across this category?
2. THE GAPS — What customer pain, desire, or outcome is NOT being addressed by current competitor advertising? Where is there open creative territory?
3. THE TONE GAPS — Is the category skewing too clinical? Too emotional? Too aspirational? What tone would feel different and more honest to the customer?
4. THE FORMAT PATTERNS — What visual approaches are dominant (before/after, product-only, lifestyle, social proof, testimonial, comparison)? Which appear overused?
5. THE OPPORTUNITY ANGLES — Given the gaps above, which 3 creative directions would be genuinely differentiated in this category right now?
For each opportunity angle, write one example headline that demonstrates how it would feel in an actual ad.
Note: I will use this to brief creative that intentionally avoids the most saturated angles in the category.
02 / Agent Two
Angle Extraction Agent
An angle is not a headline. It is the emotional territory a creative occupies. Two ads can have the same visual style and a completely different angle. Two ads can share the same claim and be speaking to fundamentally different emotional states.
Most brands test variations of the same angle and call it creative testing. They change the headline but keep the same emotional frame. They wonder why performance stays flat. The answer is almost always that they are not actually testing new angles. They are testing new executions of one angle.
Not an Angle Test
Three ads with different headlines but the same outcome claim. You are testing copy, not angles. Performance will plateau because the emotional territory is identical.
An Actual Angle Test
Ad A targets failed alternatives. Ad B targets identity. Ad C targets cost of waiting. Three different emotional entry points into the same product.
Using the customer research and awareness level mapping above, generate 6 distinct creative angles for [BRAND NAME].
Each angle must:
- Occupy a genuinely different emotional territory (not a variation of the same claim)
- Be grounded in a specific insight from the customer research, not a generic benefit
- Be appropriate for a static ad on Meta (scroll-stopping, single message, no long explanations)
- Be testable independently from the others
For each angle provide:
ANGLE NAME — A short internal label for reference
EMOTIONAL TERRITORY — What is the customer feeling when this ad lands?
CORE INSIGHT — The specific piece of customer research this angle is built on
THE HOOK LOGIC — Why would this stop someone scrolling? What is the tension or curiosity?
AWARENESS STAGE — Which awareness level does this primarily target?
RISK LEVEL — Safe/proven angle or differentiated/riskier bet?
ONE EXAMPLE HEADLINE — Written in real customer language
Present the 6 angles in order from safest (closest to proven category territory) to most differentiated (genuine white space). Flag which 3 you would prioritise for the first test batch and explain why.
Pressure-test the 6 angles above before investing in creative production.
For each angle, answer the following honestly:
1. IS THIS ACTUALLY DIFFERENTIATED? Could any competitor in this category run this same angle without changing a word? If yes, it is not a real angle — it is a category convention.
2. IS THE INSIGHT REAL? Is this grounded in something a customer actually said, or something a marketer decided sounded compelling?
3. IS THE HOOK STRONG ENOUGH? A strong hook creates a gap in the reader's mind — something they want resolved. Does this angle create that gap, or does it make a statement the reader can ignore?
4. WHAT IS THE OBJECTION? What would the reader's skeptical inner voice say back immediately? Does the angle have a built-in response to that objection?
5. WHAT VISUAL WOULD CARRY THIS? Can you describe a specific image (not a vague lifestyle shot) that would make this angle land without any copy at all?
After pressure-testing all 6, rank them again. Which 3 survive scrutiny best? Which should be dropped or reworked before going into a brief?
Be direct. Do not soften the assessment to preserve all 6 angles.
Take the 3 highest-priority angles from the pressure test and expand each into 3 variations.
A variation is NOT a different headline on the same angle. A variation is a different execution of the same emotional territory.
For each of the 3 angles, produce 3 variations by changing one of the following dimensions:
- THE ENTRY POINT — Where in the customer's experience does the ad begin? (before / during / after the problem)
- THE VOICE — Who is speaking? (brand, customer testimonial, third-party, internal monologue)
- THE TENSION SOURCE — What specific fear, frustration, or desire is the sharpest version of this angle?
For each variation provide:
- A one-line description of how it differs from the parent angle
- One example headline written in customer language
- The specific visual direction (what would the designer need to produce?)
- Which metric this variation is most likely to optimise (CTR / thumb stop / conversion)
At the end, give a recommended test structure: which 3 variations (one per angle) should go live first, and why.
03 / Agent Three
Brief Writing Agent
A bad brief is the most expensive mistake in creative production. Not because the brief itself costs anything. Because every downstream decision gets made from it. A designer working from a vague brief fills the gaps with assumptions. Those assumptions are usually wrong. The concept comes back wrong. The revision cycle starts.
A complete brief removes interpretation. It tells the designer the angle, the hook, the visual direction, the copy, and the outcome. There is no ambiguity about what the creative is supposed to do or who it is supposed to reach.
The brief is the system. At Dreamerce, every engagement produces 10 concepts per week with 3 variations each. That volume is only possible because the brief is fully structured before a single design decision is made. This agent is how we get there.
Using the angle and research above, write a complete creative brief for a static ad for [BRAND NAME].
The angle I am briefing against is: [ANGLE NAME AND ONE-LINE DESCRIPTION]
The brief must include every section below. Do not leave any section vague or open to interpretation.
BRIEF REFERENCE: [BRAND NAME — ANGLE NAME — DATE]
1. OBJECTIVE
What is this ad supposed to do? (awareness / click / conversion) Which metric defines success?
2. AUDIENCE
Who specifically is this speaking to? Not a demographic. A moment. What is this person doing, feeling, or thinking right before they see this ad?
3. THE ANGLE
In one sentence: what emotional territory does this ad occupy? What is the single idea it needs to land?
4. THE HOOK
The opening line or visual element. This must create a gap — something the reader needs resolved. Write 2 options.
5. THE VISUAL DIRECTION
Specific description of the image or design treatment. Not "lifestyle shot." Specific: what is shown, from what angle, in what context, with what visual weight given to product versus message.
6. THE COPY
Primary headline (maximum 8 words)
Supporting line (maximum 15 words)
CTA (maximum 5 words)
7. WHAT THIS AD MUST NOT DO
List 2-3 things to avoid — specific to this angle and brand, not generic ad rules.
8. HOW SUCCESS IS MEASURED
Which metric is this ad expected to move? What would a good result look like in the first 7 days?
Using the creative brief above as the parent, produce 3 variations.
Each variation should keep the angle and emotional territory intact while changing one primary execution dimension.
VARIATION A — Change the hook
Keep the visual direction and copy structure. Write a completely different hook that enters the same emotional territory from a different point. Specify what changes and why it would land differently.
VARIATION B — Change the visual direction
Keep the hook and copy. Describe a fundamentally different visual treatment for the same angle. Not a colour change or crop change. A different image concept entirely that serves the same message.
VARIATION C — Change the voice
Keep the angle but shift who is speaking. If the parent brief uses brand voice, this variation uses customer testimonial framing or internal monologue. Rewrite the hook and copy to match.
For each variation provide:
- What changed and what stayed the same
- The specific hypothesis: why would this variation outperform the parent?
- Which metric each variation is most likely to optimise (thumb-stop rate, CTR, or conversion rate)
All three should be ready to hand directly to a designer. No further interpretation required.
Review the creative brief above and audit it against the following criteria. Be direct. Flag every problem.
1. SPECIFICITY — Could a designer who knows nothing about this brand produce exactly the right asset from this brief? Or are there sections that could be interpreted in multiple ways?
2. SINGLE MESSAGE — Does the brief ask the ad to do one thing, or is it trying to communicate more than one idea? An ad that tries to say two things says neither.
3. HOOK STRENGTH — Does the hook create genuine tension or curiosity? Or is it a claim the reader can agree with and scroll past?
4. VISUAL CLARITY — Is the visual direction specific enough to brief a designer without a follow-up call? If not, what needs to be added?
5. COPY DISCIPLINE — Is the headline under 8 words? Does it use customer language from the research or marketing language invented by the brand?
6. INTERNAL CONSISTENCY — Does the hook lead into the copy logically? Does the visual support the message or work against it?
Provide your assessment as: PASS / FLAG / FAIL for each criterion, with a one-line explanation and specific fix instruction for any FLAG or FAIL.
Do not approve a brief that has a FAIL in any category. It will cost more to fix the creative than to rewrite the brief now.
04 / Agent Four
Headline Writing Agent
A headline is doing two jobs at once. It has to stop the scroll and it has to earn the click. Stopping the scroll is about tension or recognition. Earning the click is about making the reader believe the next step is worth their time.
Most AI-generated headlines fail the second job. They create curiosity but deliver nothing. The reader clicks and feels misled. This agent produces headlines that do both jobs without sacrificing either.
Generic AI Headline
"Transform your life with the power of [product]." Could be any product in any category. Creates no tension. Earns no click. Blends into the feed.
Research-Backed Headline
"You've tried everything. Here's what you haven't tried." Grounded in the failed alternatives insight. Creates a specific gap. Speaks to a real emotional state.
Using the creative brief and customer research above, write 15 headline options for [BRAND NAME — ANGLE NAME].
Write 3 headlines in each of the following structures:
STRUCTURE 1 — PROBLEM AGITATION
Opens with the exact language a customer uses to describe their frustration. Does not name the product. Creates recognition before solution.
STRUCTURE 2 — FAILED ALTERNATIVES
Acknowledges the customer has already tried other approaches. Creates empathy and differentiates before making any claim.
STRUCTURE 3 — SPECIFIC OUTCOME
Names a concrete, believable result. Not "transform your life." A specific change a specific person can imagine for themselves.
STRUCTURE 4 — IDENTITY
Speaks to who the customer is or who they want to become. Not aspirational vagueness. A specific identity statement the right person immediately claims as theirs.
STRUCTURE 5 — CURIOSITY GAP
Opens with something the customer does not know but immediately wants to. Creates a gap that the ad content resolves.
Rules:
- Maximum 8 words per headline
- Use customer language from the research, not brand language
- Do not use exclamation marks
- Do not make claims that cannot be substantiated
- Read each headline aloud. If it sounds like an ad, rewrite it.
From the 15 headlines above, select the top 5 and rank them.
For each, explain:
1. WHY THIS HEADLINE — What specific quality makes it stronger than the others? Is it the tension it creates? The specificity of the language? The recognition it triggers?
2. WHO IT HITS HARDEST — Which awareness level and emotional state is this headline most precisely calibrated for?
3. WHAT IT NEEDS TO DELIVER — If someone clicks from this headline, what does the landing page need to do to not disappoint them?
4. THE RISK — What could go wrong at scale? Is there a tone risk? Does it make a claim that might not land across all audience segments?
After ranking all 5, tell me:
- Which 1 headline you would test first and why
- Which 1 headline is the highest-risk but highest-reward bet
- Which 1 you would save for after you have a winning concept and need to iterate
Do not present all 5 as equally strong. Force a genuine ranking with genuine reasoning.
Take the top-ranked headline from above and expand it into complete static ad copy.
HEADLINE: [INSERT SELECTED HEADLINE]
Produce the following:
PRIMARY TEXT (copy that appears above the image in the feed)
- Maximum 3 sentences
- Opens with the same emotional logic as the headline
- Does not repeat the headline
- Adds one specific proof point or detail that earns trust
- Ends with a natural entry point for the CTA
OVERLAY TEXT (copy on the image itself, if any)
- Maximum 6 words
- Must work without any supporting context
- Should reinforce the headline, not repeat it
CTA (call to action button text)
- Maximum 5 words
- Action-oriented but not desperate
- Matches the awareness level this ad is targeting
SUPPORTING LINE (optional secondary text beneath the headline)
- Maximum 12 words
- Only include if it genuinely strengthens the ad. If it weakens it, say so and omit it.
Review the full copy as a unit at the end. Does the headline promise something the primary text delivers?
05 / Agent Five
Ad Review Agent
Most creative reviews in ad accounts are gut-feel decisions dressed up as data analysis. Someone looks at the CTR, says "this one's not working," and kills it. But they do not know which part is not working. The hook? The visual? The copy? The angle itself?
This agent runs a structured diagnosis. It tells you exactly where the ad is losing the reader, what the data suggests about the angle, and what to test in the next iteration. It closes the loop between what ran and what to run next.
When performance drops, the first assumption is audience fatigue or algorithm changes. Usually it is just that the brand has been running the same three angles for six months.
Diagnose the performance of a static ad currently running for [BRAND NAME].
AD DETAILS:
- Headline: [PASTE HEADLINE]
- Primary text: [PASTE COPY]
- Visual description: [DESCRIBE THE IMAGE OR DESIGN]
- Angle intent: [WHAT WAS THIS AD SUPPOSED TO DO?]
PERFORMANCE DATA (provide what you have):
- CTR: [X%]
- CPC: [£/$ X]
- CPA or ROAS: [X]
- Frequency: [X]
- Spend to date: [£/$X]
- Days running: [X]
Diagnose this ad across the following layers:
LAYER 1 — THE HOOK
Is the CTR consistent with a strong hook? What does the data suggest about whether the opening line or visual is stopping the scroll?
LAYER 2 — THE ANGLE
Is the angle landing with the intended audience? What does the CPA or ROAS relative to CTR tell you about angle-to-offer fit?
LAYER 3 — THE COPY
Where in the read journey is the reader likely dropping off? Is the primary text doing its job of moving from hook to click intent?
LAYER 4 — THE VISUAL
Based on the description, what visual elements might be contributing to or working against performance?
LAYER 5 — CREATIVE FATIGUE
At the current frequency and spend, is this creative approaching fatigue? What does the trend suggest?
End with: keep, scale, iterate, or kill — and the specific reason for that decision.
Identify what my top-performing static ads have in common for [BRAND NAME].
Below are my 5 best-performing ads from the last 90 days:
AD 1: [Headline / Copy / Visual description / CTR / CPA]
AD 2: [Headline / Copy / Visual description / CTR / CPA]
AD 3: [Headline / Copy / Visual description / CTR / CPA]
AD 4: [Headline / Copy / Visual description / CTR / CPA]
AD 5: [Headline / Copy / Visual description / CTR / CPA]
Analyse and identify patterns across:
HOOK PATTERNS — What types of opening lines or visual hooks appear in multiple winners? Are they problem-first, outcome-first, identity, or curiosity? What specific language constructions repeat?
VISUAL PATTERNS — What visual approaches appear in the top performers? What does the relationship between image and text look like in the strongest ads?
ANGLE PATTERNS — What emotional territories are the winners occupying? Are there 2-3 angles doing most of the work?
COPY PATTERNS — What copy length, structure, or tone is consistent across winners?
WHAT IS ABSENT — What approaches are conspicuously absent from your winners? What does that tell you about what does not work for this audience?
Based on these patterns, produce:
1. A "winning creative template" — the structural DNA of your best ads
2. Three new brief concepts that extend these patterns into untested angles
3. One deliberate departure from the pattern — a calculated bet on what the data cannot yet tell you
Based on the performance diagnosis and pattern analysis above, generate a brief for the next creative test for [BRAND NAME].
The brief should address one of the following based on what the diagnosis revealed:
IF THE HOOK IS WEAK — Brief a new creative that keeps the same angle but enters through a completely different opening. The angle stays. Everything that creates first contact changes.
IF THE ANGLE IS WRONG — Brief a new creative that abandons the angle entirely and tests the next-priority angle from the extraction work. Explain why the previous angle failed and what the new angle addresses differently.
IF CREATIVE FATIGUE IS THE ISSUE — Brief 3 fast variations of the winning concept that introduce enough novelty to reset frequency without abandoning what is working.
IF THE AD IS WORKING — Brief the next iteration of the winning concept. What is the one thing most likely to improve it further? Change only that.
For the brief you select, produce a full creative document using the same format as Agent 03 Prompt 01.
At the end, specify:
- What this test proves if it wins
- What this test proves if it loses
- What you test next in either scenario
A test without a hypothesis is noise. A test with a hypothesis is learning.
System in Practice
The weekly creative workflow.
This is how the 5 agents fit into a working week. Not a theoretical schedule. The actual sequence used to produce 10 concepts and 30 variations per week for scaling ecommerce brands on Meta.
| Day |
Activity |
Agent |
Output |
| Monday |
Review last week's performance data. Run diagnosis on all active ads. |
Agent 05 Review |
Keep / Scale / Iterate / Kill decisions for all active creative |
| Monday |
Update customer research with new reviews, Reddit threads, or ad library observations. |
Agent 01 Research |
Refreshed research notes added to the project brief |
| Tuesday |
Run angle extraction and pressure test for the week's new concepts. |
Agent 02 Angles |
3 prioritised angles for the week's production batch |
| Wednesday |
Write full briefs for 3-4 new concepts. Build 3 variations per brief. |
Agent 03 Briefs |
9-12 production-ready briefs including variation logic |
| Thursday |
Generate and shortlist headlines. Expand winning headlines to full copy. |
Agent 04 Headlines |
Complete ad copy for all briefs. Ready to hand to design. |
| Friday |
Design review. Brief audit on any concepts needing clarity before production. |
Agent 03 Audit |
Final approved briefs sent to design. Zero ambiguity remaining. |
On the brief audit step: Every brief that goes to design gets reviewed by Agent 03 Prompt 03 first. If it has a FAIL in any category, it does not go to design. Rewriting a brief takes 20 minutes. Rebuilding a creative from scratch because the brief was vague takes two days.
Important
What this system cannot replace.
AI raises the creative floor. It does not replace the judgment that sets the ceiling.
These 5 agents are faster and more structured than doing this work manually. But they are only as good as the thinking behind them. The research inputs need to be real. The angle decisions need to be made by someone who understands what actually performs. The brief review needs a human eye.
AI
Pattern Recognition at Scale
Processing large volumes of customer language, identifying structural patterns in briefs, generating headline variations quickly. Volume tasks with clear rules.
You
Judgment About What Converts
Knowing which angle is worth the risk. Sensing when a headline sounds like an ad. Recognising creative fatigue before the data catches up to it.
AI
Structured Output at Speed
Producing complete, consistent briefs. Running the same quality audit on every piece of copy. Applying a framework without shortcuts or fatigue.
You
Reading the Account
Understanding what the ad account is telling you beyond the numbers. Knowing when to push an angle further and when to abandon it entirely.
The system works because the human judgment inside it is real. One-shotting statics with AI produces volume. This produces creative that converts.
Next Step
Want this built and run
for your brand?
Dreamerce delivers 40 concepts per month with 120 variations. Full creative strategy ownership for ecommerce brands spending seriously on Meta who are done guessing.
Book a Free Call →
We work with a small number of brands at a time. Direct oversight on every project.