Atlas Creative Taxonomy

The system. Every creative Atlas produces follows this hierarchy. The AI agent selects the methodology per client, generates diverse concepts across the grid, and explains why each choice was made.

The hierarchy

Client Strategy (set once)
  └─ Demographics / media targeting
  └─ Personas (3-5 mindsets)
  └─ Applicable human desires
  └─ Market awareness level (Schwartz)
  └─ Market sophistication level (Schwartz)

Creative Concept (one per cell in the grid)
  = Persona × Angle × Offer × Visual Format

Execution (multiple per concept)
  = Hook variant × Copy structure × Placement size

Step 1: Choose the methodology

The agent picks one based on the client's situation:

System name	When to use	Core idea
Awareness Mapping	Competitive market, buyers at mixed stages (most reno clients)	Diagnose awareness + sophistication → message writes itself
Progress Mapping	Product hired for different reasons by different buyers	Find 3-5 "jobs" the client is hired for → each job = an angle
Offer Engineering	Weak offer capping performance, or new client with no data	Fix the value equation first, then wrap creative around it
Concept Grid	Established client, need volume and diversity	Persona × Angle matrix, produce one per cell
Gap Analysis	Client with existing ads, need to find what's missing	Tag all running ads by angle/persona/format → brief the gaps

The agent states: *"For [client], I'm using [system] because [reason]."*

Step 2: Define the personas (3-5)

A persona is a mindset at the moment of seeing the ad, not a demographic.

For TodzTerior example:

Just got keys — excited but overwhelmed, hasn't started looking
Comparing quotes — has 2-3 quotes, can't tell which is real
Burned before — had a bad experience or heard horror stories, trust is zero
Budget-anxious — knows what they want but terrified of the cost
Procrastinator — keys coming soon, hasn't done anything yet

Step 3: Select angles (one per entry point)

#	Entry point	What it argues
1	Problem	Names a pain they already feel
2	Aspiration	Shows the outcome they want
3	Fear/Loss	What happens if they don't act
4	Social proof	Others like them chose this
5	Authority	An expert or credential endorses this
6	Identity	This is for people like you
7	Comparison	Why this, not that
8	Curiosity	An unexpected claim or reframe
9	Objection	Dismantles a reason not to act
10	Value/Offer	Direct financial argument

A batch of 8 must use at least 4 entry points. Two creatives from the same entry point must target different personas.

Step 4: Build the concept grid

              Problem  Aspiration  Fear  Proof  Identity  Curiosity  Objection  Offer
Just got keys    ●                  ●                        ●
Comparing        ●                        ●                             ●
Burned                                    ●       ●
Budget                   ●                                   ●                    ●
Procrastinator   ●                  ●                                   ●

Each ● = one concept brief. Pick 6-8 for a batch.

Step 5: Pick visual formats

For statics (Barry Hott native format thinking):

Format	When to use
Bold text on photo	Problem, fear, curiosity hooks
Testimonial card	Social proof angles
Before/after split	Transformation, comparison
Stat/number card	Authority, proof, curiosity
Quote graphic	Identity, founder authority
Carousel	Multi-point comparison, process explanation
Offer card	Value/offer, direct response
Native/meme style	Curiosity, pattern interrupt

Step 6: Write the copy

Each concept gets one copy brief containing:

Persona mindset — what they're thinking right now
Entry point — which of the 10 angles
Single evidence — one proof point (not a list)
Human desire — which desire drives this (tranquility, order, saving, etc.)
Must not — what the copy cannot say

Copy structure follows the angle:

Problem → PAS (Problem → Agitate → Solve)
Aspiration → BAB (Before → After → Bridge)
Fear → Loss framing → urgency
Proof → Claim → Evidence → CTA
Offer → Value stack → Price anchor → CTA

Step 7: Test using 3:2:2

Package 3 concepts × 2 primary texts × 2 headlines into DCT. Judge after 72 hours or 10-12 conversions. Winner = winning angle, not winning ad.

Step 8: Production cadence (60/30/10)

Weekly:

60% — iterations of proven winners (new hooks, new visuals)
30% — winning message in new formats (static → carousel, etc.)
10% — net-new concepts from untested grid cells

The Three-Slot Test (pre-production QA)

Every concept must declare:

Mindset — what the viewer is thinking
Evidence — one proof point
Problem — in the viewer's own words

Two concepts sharing all three → kill one.