The 10 Commandments of Creative Strategy
The law. Every creative that ships from Atlas follows these. The AI follows them. The team follows them. If a creative violates one, it is killed — not revised, killed.
Built from Perdriau's concept architecture, Denney's 16 desires framework, and two years of Atlas campaign data. These are not opinions — they are the rules that separate the 4% of ads that carry 64% of spend from the 96% that carry nothing.
1. A concept is Persona × Angle × Offer × Format. Nothing less.
A concept is NOT a headline. NOT a hook. NOT a visual style. It is the intersection of WHO you're talking to, WHY they should care, WHAT you're asking them to do, and HOW you're packaging it.
Changing one variable makes an iteration. Changing two makes a new concept. Changing zero makes a duplicate. If a batch of 8 creatives shares the same persona and angle, it is 1 concept with 8 skins — and that is a waste of 7 slots.
2. The angle is the argument, not the topic.
An angle is why this person should care, framed from their worldview. It is not what the product does. It is not the feature. It is the interpretive frame that makes the feature matter.
- ❌ "We do BTO renovations" — that's a topic
- ❌ "Our BTO renovations are high quality" — that's a claim
- ✅ "Your friend's BTO quote is useless to you because resale is a different job" — that's an angle
The test: can you swap the brand name and run it unchanged? If yes, it's not an angle — it's a category truth anyone can claim.
3. Angles come from 10 entry points, not from brainstorming.
Every angle maps to one of these entry points. If you can't name which one, you don't have an angle.
| # | Entry point | What it argues | Example |
|---|---|---|---|
| 1 | Problem | Names a pain the viewer already feels | "Three quotes, same flat, three different totals" |
| 2 | Aspiration | Shows the outcome they want | "Walk into your finished home knowing exactly what it cost" |
| 3 | Fear/Loss | What happens if they don't act | "Every month without a plan is a month closer to keys with no contractor" |
| 4 | Social proof | Others like them chose this | "847 resale owners chose to start 6 months before keys" |
| 5 | Authority | An expert endorses this | "Featured in The Straits Times for renovation transparency" |
| 6 | Identity | This is for people like you | "For the couple who Googled 'renovation budget' at 11pm" |
| 7 | Comparison | Why this, not that | "A resale flat is not a bigger BTO — it's a demolition job first" |
| 8 | Curiosity | An unexpected claim | "The most expensive part of your renovation isn't the carpentry" |
| 9 | Objection | Dismantles a reason not to act | "You think it's too early. That's the reason you'll feel rushed later" |
| 10 | Value/Offer | Direct financial argument | "$666 off — but only if your keys are within 6 months" |
A batch of 8 creatives must use at least 4 different entry points. Two creatives from the same entry point must target different personas.
4. Persona is a mindset, not a demographic.
"Women 25-34 in Singapore" is a media target, not a persona. A persona is a mindset at the moment of seeing the ad.
- ❌ "HDB owners" — demographic
- ✅ "Couple who just got keys and hasn't started looking yet" — mindset
- ✅ "Owner comparing three quotes and can't tell which is real" — mindset
- ✅ "Person who's been planning for months but can't commit" — mindset
Each client should have 3-5 personas defined by mindset, not age/income. The same 35-year-old woman is three different personas depending on whether she just got keys, is comparing quotes, or has been burned by a previous contractor.
5. One angle, many hooks. Not one hook, many angles.
A hook is the first 1-3 seconds that expresses the angle. Test hooks within an angle, not angles within a hook.
If your angle is "resale is a different job than BTO," you test:
- Hook A: "Your friend's BTO quote is useless to you."
- Hook B: "A new flat has nothing to demolish."
- Hook C: "Nobody told you resale costs more because it costs differently."
Same angle, three hooks. Each hook is a $50 test, not a $500 production. The angle is the bet. The hook is the execution. Test cheap, scale what works.
6. The human desire is the engine. The product is the vehicle.
From Denney's framework — every effective ad taps one of these desires:
| Desire | In renovation context |
|---|---|
| Tranquility | "Peace of mind that the job is done right" |
| Order | "A plan you can see before you commit a dollar" |
| Family | "A home your kids will grow up in" |
| Social status | "The flat your friends ask about" |
| Independence | "Your home, your way, no compromises" |
| Saving | "Same quality, less waste, lower cost" |
| Power | "You decide what stays and what goes" |
The product (renovation) is the vehicle. The desire is the fuel. An ad that leads with the product is a brochure. An ad that leads with the desire is persuasion.
7. Format follows angle, not the other way around.
Never start with "let's make a carousel" or "we need a reel." Start with the angle and persona, then pick the format that serves them:
| If the angle is... | The format should be... |
|---|---|
| Social proof | UGC testimonial, before/after carousel |
| Authority | Founder-to-camera, press mention static |
| Problem | Story-driven reel, pain-point static |
| Comparison | Side-by-side static, "us vs them" carousel |
| Curiosity | Pattern interrupt reel, surprising stat static |
| Offer | Direct response static, countdown reel |
An angle that only works in one format is not an angle — it's a gimmick.
8. Every creative must pass the Three-Slot Test before production.
Before ANY creative enters production, it must declare:
- Mindset: What is the viewer thinking right now? (Not who they are — what they believe)
- Evidence: What single proof point supports the argument? (Not a list — one thing)
- Problem: What does the viewer call this problem in their own words?
If two creatives share all three slots, they are the same concept regardless of how different they look. Kill one.
9. Measure angles, not ads.
An ad is a disposable container. An angle is a strategic asset. Track performance at the angle level:
- Which angle has the lowest CPA?
- Which angle has the highest hold rate?
- Which angle saturated fastest?
- Which angle works across multiple formats?
When an angle works, make more hooks for it. When an angle fails, retire it — don't revise the hook. A bad angle with a great hook is still a bad angle.
10. The brief is the product. The creative is the packaging.
The quality of the output is decided before a single word is written. A brilliant copywriter given a bad brief produces brilliant garbage. The brief must contain:
- The persona (mindset, not demographic)
- The angle (entry point, not topic)
- The single evidence point (not a list of features)
- The human desire being activated
- What we must NOT say (as important as what we must)
If the brief doesn't have all five, it is not ready for production. Send it back.
How to use this
For the AI engine:
Before generating, build the Persona × Angle matrix. Assign each creative to a specific cell. Reject any batch where more than 2 creatives share the same cell.
For the Creative Strategist:
Angles come from customer evidence (reviews, WhatsApp messages, meeting notes, sales calls) — not from inside the agency. If you're brainstorming angles in a room, you're guessing.
For the Account Manager:
When a client says "the ads aren't working," the first question is which angles are running. If all ads share one angle, the problem is diversity, not execution.
For the Media Buyer:
Report performance by angle, not by ad. When you say "this ad is winning," translate it to "this angle is winning" — because that's what the next brief needs to know.