CV GIANCARLODINARDO.6@GMAIL.COM
Creative Lifecycle Hub

The creative system in four parts.

Four layers. Each one solves a different problem. None of them do what the others do.

Research reads the market before anything gets briefed. Reviews, competitor gaps, past performance, customer language. All of it processed into angles and personas with a traceable source. The raw material.

Coverage takes that material and audits how it sits across the account. What personas are missing creative. Which angles have been over-tested. What the account is telling Meta versus what it should be.

Hooks decide how each individual ad enters its territory. The hook is where every upstream decision (persona, angle, awareness level) becomes a signal the algorithm can read and a reason for the right buyer to stop. It is a targeting decision expressed in three seconds.

Lifecycle manages what happens to a signal once it starts working. Discovery, validation, expansion, saturation, decay. The goal is to extend the signal before decay forces the account into emergency creative mode.

Use this hub for
Navigation through the system
Input layer
Research stack → feeds coverage
Deep dives live in
The individual playbooks
The journey
From territory to signal to scale.
00
Find the angles
Research Stack · Input layer
Reads reviews, past performance data, and competitor gaps to surface angles and personas before a brief gets written. What to say and who to say it to. Coverage decides where it goes.
01
Map the territory
Creative Coverage · The loop begins
Takes what research produced and audits the account against it. Which personas have no creative. Which angles are being over-served. What territory the account is missing before the next brief cycle starts.
02
Choose the entry point
Hooks Are Targeting
Once coverage identifies the territory, hooks decides how to enter it. The hook is where persona, angle, and awareness level converge into a signal the delivery system can read. A targeting decision, not a copy decision, made before production starts.
03
Manage the signal over time
Creative Lifecycle · Loops back
Tracks a winning ad through its predictable stages and manages extension before the account becomes dependent on one creative. When the cycle closes, the results go back into coverage and the loop starts again.
The system

Four parts. One continuous flow.

The research stack feeds the loop. Coverage, hooks, and lifecycle run continuously from there. Read them in order, or jump to the part of the system that matches the problem in front of you.

00 · Research · Input layer
Where angles come from.
Reads the market before a brief gets written. Five sources, run in order: reputation audit, review mining, persona mapping, past performance, competitor gaps. Each one produces a specific output. What to brief and who to brief it for. Coverage handles the rest.
Read Playbook →
01 · Coverage · The loop begins
How the angles get distributed.
Takes what research produced and audits how it is allocated across the account. By persona. By concept. By format. Identifies what the account is missing before anything new gets briefed. Hook strategy and performance management sit in the next two layers.
Read Playbook →
02 · Hooks
How each ad enters its territory.
Coverage identified a gap. Hooks decides how to enter it. The hook is where every upstream decision (persona, angle, awareness level) becomes a signal the algorithm can read and a reason for the right buyer to stop. What happens once the ad wins is lifecycle's job.
Read Playbook →
03 · Lifecycle · Loops back
What happens to a signal over time.
Once the ad wins, lifecycle takes over. Tracks five stages: discovery, validation, expansion, saturation, decay. The job is to extend the signal and build replacements before the account becomes dependent on one ad. When the cycle ends, results go back into the next round of coverage.
Read Playbook →
Supporting Layer 00 · Research

Three playbooks that go deeper on specific parts of the research process. Not part of the main loop. Deep dives on individual sources within Layer 00.

Research source · Comments
Mining Ad Comments for Briefs
How to read objections, questions, and controversy in comment sections as a brief input. One source from the research stack, covered in full.
Read Playbook →
Research tool · AI
Where AI Fits in the Workflow
How AI accelerates research inputs (review clustering, pattern sorting, language extraction) without replacing the judgment about what matters.
Read Playbook →
Research tool · AI operations
Claude in Creative Operations
AI in the messy middle of creative work: language mining, angle expansion, pattern recognition. Makes research faster without making it different.
Read Playbook →
The system in practice

What the first two layers
actually produce.

Research and Coverage don't end in a report. They end in a creative blueprint: a matrix of distinct directions built before a single brief is written. Research surfaces the raw material, customer language, angle territory, personas defined by decision context. Coverage audits how that material sits against the account and identifies what's missing.

The map is where those two layers converge into something you can hand to a writer. Personas, awareness levels, one primary angle each. Every direction is genuinely distinct, and every one earns its own entity ID in the delivery system.

From there, Hooks decides how each direction enters the feed. Lifecycle manages what happens once something wins. When the cycle closes, the data goes back into Research and Coverage, and the map gets updated for the next round.

See it worked through: the roofing creative blueprint →
Research + Coverage → Map → Hooks → Lifecycle → loop
00 · Research
Reads the market. Surfaces angles, personas, and the customer language that makes a brief specific rather than generic.
01 · Coverage
Audits the account against what research found. Identifies the gaps: which personas have no creative, which angles are over-served.
↓ The creative blueprint
The output of Research and Coverage combined. A matrix of distinct directions across personas, awareness levels, and angles, before a single brief is written.
02 · Hooks
Takes each direction the map produced and decides how it enters the feed. Persona, angle, and awareness level compress into the first three seconds.
03 · Lifecycle
Manages the signal once it wins. Extends before decay. Results feed back into Research and Coverage, and the map gets updated.
How the pieces connect

One question per layer.
No layer repeats another.

Research reads the market and surfaces the angles, personas, and hook language that go into the brief. It is the only layer that works from external signal rather than account data.

Coverage takes that material and maps it across the account. Which personas have creative. Which angles are over-served. What needs to exist before the next round gets briefed. Coverage works from the inside out.

Hooks decide how each ad enters the territory coverage identified. Persona, angle, and awareness level are all decided upstream. The hook is where those decisions compress into three seconds: a signal the delivery system reads to find the right buyer, and a reason for that buyer to stop when the ad reaches them.

Lifecycle manages a winning ad through its predictable stages. The goal is extension before decay, not reaction after it. When the cycle closes, its data feeds back into research and coverage, and the process starts again.

Why the system works with the algorithm, not against it

How Meta's delivery system
actually works.

Meta rebuilt its ad delivery infrastructure from the ground up between 2024 and 2025. The result is a system that no longer starts with the audience you defined. It starts with the creative you uploaded and works backward to find the right person. Most advertisers are still running the old playbook.

Understanding how the delivery stack works explains why coverage, hook selection, and creative diversity are not just strategic preferences. They are what the algorithm requires to function well.

Meta Delivery Stack · 2025 onwards
Andromeda · Retrieval
Reads your creative and decides which ads are eligible to enter the auction. Filters before the bid happens.
↓
GEM · Intelligence
Learns which creative signals produce which outcomes for which users. Shapes what gets featured next across the whole system.
↓
Lattice · Ranking
Unifies optimization decisions across surfaces and placements. Transfers learnings from one placement to another in real time.
↓
Auction · Delivery
The surviving ad enters the auction. Bid matters here, but creative quality and predicted conversion rate can outweigh a higher bid.
How the stack works

Three systems. One pipeline.
Creative is the input to all of them.

Andromeda

The door into the auction

Andromeda is the retrieval engine that runs before any auction takes place. When a user is about to be shown an ad, Meta needs to select from tens of millions of eligible candidates in milliseconds. Andromeda narrows that pool to roughly a thousand options by reading the creative itself: its visual content, themes, language, format, and hooks.

It assigns each ad an Entity ID based on what the creative looks and sounds like. Ads that are visually or tonally similar get grouped into the same Entity ID, which means they compete for a single retrieval slot rather than separate ones. Running fifty variations of the same concept does not produce fifty chances to reach the auction. It produces one.

This is the mechanism behind Creative Coverage. Distinct personas, distinct angles, distinct visual environments produce distinct Entity IDs. Each one is a separate ticket into the auction. Coverage is not a strategic preference anymore. It is how the account earns its way into delivery.

GEM

The system that learns what works

GEM is Meta's generative ad recommendation model, trained at a scale comparable to a large language model. It analyzes patterns across engagement, behavioral signals, and conversion data to predict which creative will perform for which user at which moment. It feeds those predictions into Andromeda and shapes what gets prioritized across the entire system.

The practical implication is that GEM rewards accounts that give it diverse, high-quality signals to learn from. An account running one creative gives GEM one data point. An account running genuinely distinct creative across personas, formats, and hook types gives GEM a richer pattern to work from, which means better matching over time.

GEM also evaluates ads within broader user journeys rather than in isolation. An ad that performs well as part of a sequence of creative signals carries more weight than one that appears once and disappears. This is part of what the Lifecycle layer is managing: a signal that is extended thoughtfully builds more history for GEM than one that gets replaced the moment performance dips.

Lattice

The architecture that connects it all

Lattice is Meta's centralized ad ranking architecture. It unified a previously fragmented system where Feed, Reels, and Stories each had separate optimization models with no shared learning. Now a single architecture transfers behavioral signals across all surfaces in real time.

What this means in practice: a conversion insight learned from a Reel immediately improves how the system optimizes a Feed placement. An ad that builds strong signal in one context is no longer starting from zero in another. The account's history travels with it across the platform rather than being siloed by placement.

Lattice also means Meta is no longer optimizing individual ads in isolation. It is optimizing the entire system. Advertisers who provide diverse, high-quality signals across placements, objectives, and funnel stages are rewarded at the system level, not just the ad level.

What this means for the four layers

How each layer connects
to the delivery stack.

Andromeda, GEM, and Lattice each have specific requirements from the creative going into them. Each of the four layers in this system addresses a different part of those requirements directly.

00
Research

Andromeda reads creative signals to match ads to users. The quality of those signals is determined entirely by the quality of the thinking that went into the brief. Research produces the raw material: the customer language, the angle territory, the persona with a specific decision context. Better research inputs produce better creative signals. Better creative signals produce better algorithmic matching.

01
Coverage

Coverage is now an Entity ID strategy. Distinct personas, distinct angles, distinct visual environments produce distinct Entity IDs. Each Entity ID is a separate retrieval ticket. An account with genuine coverage across buyer types earns more ways into the auction than one running ten variations of the same concept. The Coverage layer was designed to produce creative diversity as a strategic output. It turns out diversity is also what the algorithm requires to function at full capacity.

02
Hooks

The hook operates on two levels simultaneously. At the human level it selects who stops scrolling. At the algorithmic level it tells Andromeda who to retrieve the ad for, before the ad ever reaches the feed. The hook type is not just a copy decision or even an audience decision in the old sense of the word. It is the primary signal the delivery system uses to determine relevance. Change the hook type and you change the Entity ID and the user Andromeda goes looking for.

03
Lifecycle

GEM evaluates ads within broader contextual journeys, not in isolation. A signal that has built genuine history in the account carries more weight than one that is constantly being replaced. The Lifecycle layer manages exactly that: extending a winning signal before decay forces a reset, building replacements before the account becomes dependent, and feeding results back into research so the next cycle starts from a better position. Patience in reading performance data is a strategic input, not a passive posture.

The bottom line

You are not buying access to audiences anymore. You are buying algorithmic distribution based on creative signals.

The lever you actually control is the creative. What you put in, how distinct it is, and how well it reflects a specific buyer's situation at a specific point in their decision: that is what determines how well the system works for you.

Broad targeting is not lazy. It is what the system is designed for. Manual interest stacks now serve as soft suggestions, not hard constraints.
Minor creative tweaks do not produce new Entity IDs. A headline swap or a color change is the same ad to Andromeda. Genuine conceptual diversity is what earns additional retrieval tickets.
The account that feeds the system diverse, high-quality signals across personas, formats, and funnel stages is rewarded at the system level. Not just the ad level.