Perfection is now a red flag.

When every brand uses the same AI prompts to generate “flawless” content, quality becomes invisible. It is the new background noise. In a world of synthetic perfection, the “glitch” is the only thing people trust.

But before we talk about viral hits, we have to acknowledge the shift in how the work is actually getting done.

1. The “Anti-Labor” Revolution: Data analyst and Performance Marketers are now “Vibe Coding” their work

• The Concept: Traditional data analytics and web development are too slow. Marketers are bypassing the “old way” of manual labor.

• The Evidence: Look at talknerdytome-labs. Instead of hiring a team or using rigid tools, marketers are connecting Google Ads and Facebook Ads accounts directly to Claude.

• The Shift: They are using Claude as a personal data analyst and developer to build apps and analyze sheets on the fly. It is about shipping at the speed of “vibes” rather than waiting for a dev cycle.

  ---

2. The Great “Anti” Movement: Why Normal is the New Invisible

• Anti-Normal: Look at Lauren Elizabeth. She went viral for an oddly specific, “proper” way to wear a fur coat hood. It is niche, weird, and completely un-optimized.

• Anti-AI Aesthetic: Business of Fashion reports that blurry, low-fi photography is taking over. If a photo is too clear, people assume a bot made it.

• Anti-Influencer: The Columbia Tribune highlights “Underconsumption-core.” People are showing off used-up products and 10-year-old shoes. They are rejecting the “perfect” influencer lifestyle.

• 2016 Nostalgia: We are longing for the “old” internet from before everything was governed by hyper-efficient algorithms. We miss the mess.

3. Your “Proof of Humanity” Action Plan

• Ship Human Mistakes: If your copy is too clean, add a typo. If your video is too stable, shake the camera. Prove a person was there.

• Ship Human Craziness: Do things that don’t scale. Create content about things that are “too specific” to be a mass-market ad. Specificity creates a “human signal.”

• Ship Human Weirdness: Follow the Lauren Elizabeth model. Be obsessed with something small and strange. AI is good at “average,” but it is terrible at “weird.”

• Remember When We Were Human: Use visuals that trigger 2016 nostalgia. Low-res, raw, and unfiltered.

The Bottom Line:

AI is too perfect. It can copy your style, but it can’t copy your grit

In an era of synthetic noise, “weird” is the only thing that breaks the scroll. Stop trying to be the best. Start trying to be the most human.

See you next week,
Manu the Vandal

Every week there’s a new model. Kling has a sale. Then LTX-2 drops and suddenly that’s the one you need. Normally you’re juggling subscriptions, switching platforms, renting Runpod hours. Praying your GPU doesn’t catch fire.

Weavy just plugs them all in. One platform. All the models. No GPU required. No Runpod bills. Everything works from your browser.

Why Weavy Works For Me

I don’t hate ComfyUI. I love the control. But I’m tired of the setup dance.

Weavy is different. Design-first. Brutal UX. Built by people who understand art direction, not just machine learning engineers sharing their latest discovery.

No model hunting. Everything’s already loaded. Kling, LTX, whatever came out this morning. It’s there.

Production tools built in. 3D rendering, compositing, color grading. Out of the box. No Node Manager hell. No updating, deleting conflicts, memory errors, conflicted dependencies. The tools VFX artists spent hours layering in After Effects are just... there. Ready. That’s the bridge. Old-world production techniques meeting AI speed.

Copy nodes between projects. Built an AI workflow you like? Two clicks, it’s in your next project.

Workflows become shareable apps. Build something complex, simplify the interface, send it to someone. They can use it without learning the whole system.

So I built 4 workflows. Not for clients. For fun. Here’s what they actually do: 👇

🔥 The 5 Workflows I Built (Steal These)

Each one solves something specific I deal with regularly. Links to the projects and assets included.

1️⃣ Visual Hijack 🕵️

🔗 Visual Hijack

What it does: Feed it any image or video. It breaks down the visual concept and generates variations you can remix.

Why it’s fun: Pure creative theft. See something that works aesthetically, understand the mechanics, then vandalize it into something new.

The artistic angle: This is about style transfer and concept extraction. You’re not copying. You’re learning visual language by deconstructing it. Feed it a Rothko, get the color field logic. Feed it a Tarkovsky frame, extract the compositional tension.

2️⃣ Full Brand Kit 🎨

🔗 One Click Chaos Brand Kit

What it does: Input visual guidelines. One run generates profile pics, banners, headers. All sized right, all cohesive.

Why it’s fun: Endless variation machine. Lock in a core aesthetic, then explore how far you can push it before it breaks. See which colors collapse first. Watch which compositions stay recognizable.

The artistic angle: It’s a constraint game. How much can you vary while maintaining identity? You’re essentially making generative art with brand consistency as the rule set. Break the rules intentionally and see what still reads.

3️⃣ Monkeys in Art: A Brief, Bananas History 🐒

🔗 Monkeys in Art

What it does: Animates paintings across completely different art styles and techniques. Then stitches them all together into one film tracking monkeys from classical art to modern AI-generated imagery.

Why it’s fun: For centuries, artists painted monkeys to laugh at humans. It’s satire with fur.

The Weavy workflow is basically a menu of animation techniques. Pick the style, pick the animation method, animate the painting. Then combine everything into a short film that travels through art history.

The artistic angle: Each historical style needs different technical approaches to animate properly. Oil painting movement is different from line art is different from how you’d handle a Frida-style piece. The workflow lets you experiment with all of it, then see how they flow together as one narrative.

4️⃣ The E-com Asset Bombing-Run 💣

🔗 E-com Asset Bombing

What it does: I built this for you. One flat product photo becomes a real-looking scene: better angles, better light, lifestyle frames, and assets you can actually run as ads.

The vibe: I’ve seen the viral “autopilot” setups.
n8n agents scraping the internet, calling Sora-style APIs, pumping out an endless stream. It looks like a machine. But the output looks like a parody.

When I use it: E-commerce clients need volume. 50+ pieces of content for a launch. Flat product shot → 3D renders → lifestyle contexts → UGC-style review videos. What would be multiple photoshoots compressed into an afternoon.

And it’s profitable: Controlled A/B tests show AI provided 36% revenue lift compared to e-commerce websites

5️⃣ Anime Style Lab 🎌

🔗 Anime Style Lab

What it does: Style picker loaded with famous manga and anime artists. Pick your style, set up multi-shot sequences, or jump into the Sakuga setup for pure movement experimentation.

Built-in model comparison. Drop your image, generate across Kling, Sora, Veo. Start with 4-5 second clips. Try all three. One nails camera movement. Another handles character consistency. One does speed lines right. Each model has specific strengths.

Why it’s fun: Immediate access to iconic anime styles without hunting for the right prompts or settings. Multi-shot for storytelling. Sakuga mode for fight choreography and dynamic movement testing. See which AI model actually understands fluidity versus which one just makes things wiggle.

The artistic angle: Reverse-engineering decades of anime evolution through artist-specific styles. Test movement physics. Study how different models interpret action. This is the playground for anime lovers who want to experiment without the technical barriers.

The kind of animation that takes decades to master is starting to become accessible. We're watching it happen.

🎯 Who This Is For (And Who Should Skip It)

Weavy makes sense if you:

✅ Want to experiment without setup hell
✅ Like trying new models the day they drop
✅ Need video compositing without dependency nightmares
✅ Value speed over maximum control
✅ Want to share workflows with non-technical friends

Stick with ComfyUI if you:

❌ Need bleeding-edge custom pipelines
❌ Want total parameter control

Different tools. Different moods. Use both.

🎯 Bottom Line

Weavy handles subscription chaos. New model drops? It’s already there. No switching platforms. No rebuilding workflows.

The four workflows are yours to steal. Links include project files.

Not replacing your stack. Just removing the annoying parts.

🤘 Now Let’s Vandalize Together

What’s the most absurd artistic experiment you’ve been putting off because setup seems painful?

Reply and tell me. I’ll pick one and build a workflow for it next week.

Manu

We spent $18,026 on paid ads for a family art studio.

We got 87 purchases.

That’s $207 to acquire a customer.

The cheapest product in the store? $79.

The math stopped mathing a long time ago.

This is the story of Jean-Pierre Weill Studios, a family-run business creating something genuinely rare: 3D paintings on glass. Hand-painted. Layers of depth. Pieces that start at $79 for the Mini Collection and go up to $15,900 for museum-quality Grand editions.

I’ve been working with them on creative, PPC, and SEO. You can see the full case study on my portfolio. We crafted stunning creatives. Launched high-impact ads. Optimized campaigns across Meta and Google.

And we learned something painful together.

🔨 What We Broke

The e-commerce playbook that says “spend money, get customers.”

Here’s what the data actually showed us:

Meta Ads (Nov 2024 to Apr 2025):

• Total Spent: $15,387

• Total Purchases: 82

• Cost Per Purchase: $187.65

• ROAS: 0.85x

Google Ads:

• Total Spent: $2,639

• Actual Purchases: 5

• Cost Per Purchase: $527.94

Read that Google number again. Five hundred and twenty-seven dollars to acquire a single customer. On a platform where we were competing against mass-produced wall art from companies with infinite budgets.

The ROAS of 0.85x means for every dollar we put in, we got 85 cents back. We were paying to lose money.

🧪 What We Tested

We tried everything.

Performance Max campaigns. Advantage+ on Meta. Retargeting sequences. Lookalike audiences. Black Friday pushes. Different creatives, different hooks, different audiences.

Some ads performed better than others. Our best Meta ad hit a $41 cost per purchase. But even that winner couldn’t overcome the fundamental problem:

When someone searches “unique wall art” or “3D painting,” they’re shown our handcrafted $3,100 Classique Collection next to a $29 Amazon print. The algorithm doesn’t know the difference. The shopper has to figure it out in 0.3 seconds while scrolling.

We weren’t losing because our ads were bad. We were losing because the arena was wrong.

Art isn’t a commodity. But PPC treats everything like one.

✅ What Actually Worked

Two plays. One organic. One B2B.

🌿 The Organic Play: Give the Art Away (Digitally)

We stopped trying to sell in the feed and started giving people a reason to live with the art.

Free wallpapers.

We took the studio’s most beautiful pieces and turned them into phone wallpapers. No purchase required. No email gate. Just beautiful art that people could carry in their pocket every day.

Pinterest became the distribution engine. The platform is built for this. People actively search for “aesthetic wallpapers,” “unique phone backgrounds,” “art for iPhone.” They’re not scrolling mindlessly. They’re hunting.

And when someone sets your art as their wallpaper, they see it 100+ times a day. That’s brand exposure you can’t buy. That’s the kind of slow-burn familiarity that turns into “I finally bought one of those glass paintings I’ve had on my phone for six months.”

The organic SEO work compounds on top of this. Blog content. Pinterest optimization. Building the kind of presence that doesn’t disappear when you stop paying.

🏛️ The B2B Play: Sell to the Sellers

The pieces were already in galleries. Real ones. Places where customers walk in expecting to pay for craftsmanship. Where a $3,100 price tag is normal, not insane.

So we asked: What if we stopped selling art to consumers and started selling to the places that sell art?

We used Clay to scrape gallery databases. Hundreds of them. We looked for galleries with physical stores, gift shops, places that actually move inventory.

Then we identified the decision makers: art store managers, gallery curators, wholesale buyers.

One wholesale relationship = dozens of pieces moving. No CPA nightmare. No bidding war against IKEA.

🔥 Bottom Line

Some products don’t belong in the PPC arena.

When your cost per acquisition is higher than your entry-level product, that’s not a bid optimization problem. It’s a channel problem. When your ROAS is below 1x, the algorithm isn’t broken. Your strategy is.

We learned this the hard way. $18,000 worth of hard.

The studio makes something genuinely beautiful. Hand-painted glass art with real depth, created by a family that cares about every piece. That story doesn’t compress into a 15-second scroll.

But it does resonate on Pinterest, where people actively search for beauty. And it resonates with gallery owners who understand craftsmanship.

TL;DR: We burned $18k on paid ads with a 0.85x ROAS. Google was especially brutal at $528/purchase. The product wasn’t the problem. The channel was. Now we’re running organic through Pinterest and free wallpapers, while building wholesale gallery relationships using Clay for prospecting. Zero ad spend. More brand love.

I joined an AI social media startup right after a layoff and a pivot. The company was rebuilding from scratch: no brand, no marketing assets, and a “product” that was still wireframes in Figma.

So we ran a brand awareness experiment using a channel most marketers ignore: GIPHY.

What we did

We created a small pack of useful stickers people would actually want in daily chats (not product demos, not logo spam). Then we uploaded them to GIPHY with search-driven keywords, the same words people type inside Slack, iMessage, WhatsApp, and Instagram DMs.

Because GIPHY is embedded everywhere, a sticker that ranks for common searches becomes free distribution inside real conversations.

🔥 The result

268,000+ organic impressions. $0 spend.
One sticker alone hit ~41.7k views.

No, you can’t draw a straight line from “someone used our sticker in Slack” to “someone signed up.” That’s not how top-of-funnel works. But getting your brand seen hundreds of thousands of times in non-ad contexts, in friends’ chats and team channels, is real distribution.

🎨 The Backstory: Building a Brand From Zero

When I walked in, there was nothing. No visual identity. No explainer videos. No social assets. No app to show. Just a vision for an AI-powered automation tool and a team trying to figure out what comes next.

So I called Tom Engler (hands down the best creative illustrator and animator I know). Together, we built the entire brand from scratch: logo, color palette, character mascots, the works.

Once we had the visual identity locked, we created a welcome video, which was then repurposed into Lottie animations and GIPHY stickers, literally just invested one hour on Giphy. That's when I started thinking about distribution.

The obvious plays (LinkedIn posts, Twitter threads, paid ads) felt boring. Everyone does those. And we had zero budget for paid acquisition anyway.

That’s when I remembered something I’d read in Tom Orbach’s Marketing Ideas newsletter. He wrote about using GIPHY stickers as a distribution channel (GEO). The concept stuck with me: people search for stickers and GIFs constantly, and if your brand shows up in those results, you’re getting free impressions in spaces ads can’t reach.

I decided to test it.

The play: Create branded stickers and GIFs people would actually WANT to use in their daily conversations. Not product demos. Not logos slapped on animations. Genuinely useful visual content that happens to carry the brand.

📊 The Results

Here’s what 268,000+ impressions looked like broken down:

• Calendar sticker: 41,698 views

• Coffee mug sticker: 17,862 views

• “Relaxing” character sticker: 8,505 views

• Plus dozens of other stickers adding up to 268,000+ total impressions

All organic. All free. All from people voluntarily sharing branded content in their group chats, Slack channels, and social DMs.

Now, let’s be honest about what this means.

Brand awareness is the long game. You can’t attribute a signup directly to someone seeing your sticker in a group chat three weeks ago. That’s not how top-of-funnel works.

But here’s what IS measurable: 268,000 times, someone saw this brand in a context that wasn’t an ad. In their friend’s message. In their team’s Slack. In a place where brands usually don’t reach.

For a startup with zero marketing budget, that’s not nothing. That’s distribution most companies pay thousands for.

The real question is: was there something underneath to convert all that awareness?

💀 The plot twist

The marketing didn’t fail. It worked exactly as brand awareness should.

What failed was what came next.

Some startups hire marketers hoping marketing will create product-market fit. I thought I was joining a growth role. What I walked into was a prayer: almost no active users, “traction” that was more aspiration than reality, and a short timeline that basically matched how long it takes to realize the foundation isn’t there.

Three months later, I was out, not because awareness didn’t happen, but because there was nothing underneath it to convert.

✅ The framework (steal this)

1. Make stickers people would use even if your brand didn’t exist.

2. Keyword like your audience searches (emotions + situations: “tired,” “meeting,” “coffee,” “deadline”).

3. Let compounding do the work: every use shows your brand to the whole chat.

The red flags I now ask before any role

• “Can I see activation (not signups)?”

• “What’s your retention curve?”

• “Why is this role open now, scaling what works, or hoping I’ll make it work?”

🔍 The Red Flags I Now Check Before Any Role

After this experience, I never accept a role without asking these questions:

1. “Can I see your activation metrics?” Not signups. Activation. How many people actually USE the product after signing up?

2. “What’s your current retention curve?” If they hesitate or pivot to “we’re early stage,” that’s your answer.

3. “Why is this role open now?” Listen for whether they’re scaling something that works or hoping you’ll make something work.

4. “Who held this role before, and what happened?” If the answer is “we’re building this function for the first time” at a company that’s been around for years... ask why.

Product-market fit isn’t a vibe. It’s data. If they can’t show you the data, assume it doesn’t exist.

🔥 Bottom line

GIPHY is still wide open for free brand impressions.
Just don’t build awareness for a ghost town.

If you’re a founder: don’t hire for hope—hire for scale.
If you’re job hunting: check the data before you sign.

See you next week ✌️

A while ago we had two videos live in the same campaign.

Same offer.
Same platform.
Same creator, same vibe.

One was quietly printing money.
The other was quietly burning it.

We did the usual ritual:

• Checked comments

• Blamed “learning phase”

• Invented platform conspiracy theories

But if you opened our media library, both files looked identical apart from a random suffix.

From inside the ad account, they were just two more tiles in the grid.

The real problem wasn’t the ads.

The problem was this:

Our creative library was a museum, not a database.

We were “rich” in assets and completely broke in insight.

That’s the day Marketing Vandals stopped treating creatives as files…
and started treating them as rows in a table.

Folders Are for Storage. Tables Are for Learning.

Most teams still live in folders:

• Campaign subfolders

• Date subfolders

• Filenames written by an editor on 2 hours of sleep

That system is fine for one question:

“Where did we put that video?”

But it fails on the questions that actually move performance:

• “Which kind of opening shot keeps winning?”

• “Which hooks die in week 1 every single time?”

• “What works for students but flops with founders?”

You don’t answer those in Drive.

You answer them in a table.

At Marketing Vandals, the mental shift was simple:

• Every creative = one row

• Every trait that might matter = a column

• Performance data = another table we can join in

The asset is still a file.
But the value lives in the row.

The Marketing Vandals Tables (a.k.a. Our Creative Brain)

We run on two main tables:

1. The Creative Table – everything about what the asset is

   

2. The Performance Table – everything about what the asset did

   

In the creative table, we track things like:

• What it’s about (offer, product, vertical)

• How it shows up (format, style, creator type)

• How it starts (the first few seconds, the hook)

• Where it lives (channel, market, language, campaign)

Not in paragraphs.
Not in notes.
In cells.

In the performance table, we don’t overthink it:

• Spend, impressions, CTR, CVR, CPA, ROAS, view rates, etc.

• One row per ad / asset export from each platform

Then we connect the two tables with a shared ID or filename.

Suddenly:

• That “nice video from the Black Friday push”

• Becomes

• “Row 438 that opened with a problem dramatization, ran to creators in the US, and hit a 3.4x ROAS.”

Once it’s a row, it’s no longer anecdotal.
It’s queryable.

Why We’re Obsessed With the First Few Seconds

One of the most important columns in our creative table is simply:

“What happens in the first few seconds?”

Because in paid social, those seconds decide if your ad even gets a chance to perform.

So we tag that moment consistently:

• Is it a before/after?

• A bold claim?

• A weird visual?

• A problem dramatization?

• A question straight to camera?

Not with fancy language.
With simple, repeatable labels we can filter later.

Now we can ask:

• “For this vertical, which type of intro gives us the best CTR?”

• “When we open with a weird visual vs. a straightforward product shot, what happens to CPA?”

You can’t get that answer from a folder name.

You get it from a pivot table.

Hooks, Verticals, Campaigns: The Tags That Pay Rent

There are three categories we tag religiously:

1. Verticals – Who is this really for?

   • Students, creators, beauty lovers, founders, parents…

   • We pick one primary vertical instead of pretending “this ad is for everyone.”

2. Hooks – What’s the first promise / tension?

   • Every hook gets a short description and an internal code.

   • The same hook can show up in different edits and formats, but it’s still the same idea.

3. Campaigns – What initiative did this belong to?

   • One campaign → many creatives → one column to tie them all together.

Once those tags live in the table, we can do things like:

• Compare hooks inside the same campaign

• Compare verticals across the same hook

• See which campaigns relied on “one lucky creative” vs. a broader bench of winners

That’s how you graduate from “cool, that ad worked” to “we understand why it worked.”

From Guessing to Analysis: How the Tables Actually Get Used

The magic is not in having tables.

The magic is in running the same analyses over and over.

Here are a few flows we keep coming back to at Marketing Vandals:

1. “Top Hooks This Quarter” Table

   • Filter by date range

   • Group by hook

   • Look at ROAS / CPA / spend per hook

   • Result: a ranked list of ideas, not just videos

2. “Vertical vs. Intro Type” Table

   • Rows = verticals

   • Columns = intro types (bold claim, problem, weird visual, etc.)

   • Cells = CTR / ROAS averages

   • Result: a cheat sheet for which openings to use for which audience.

3. “Format & Channel Sanity Check” Table

   • Compare performance of the same idea across formats and placements

   • Result: clarity on when to invest in extra edits vs. when one format is enough.

4. “Workhorse Assets” View

   • Filter for creatives that ran in multiple campaigns and still hit target KPIs

   • Result: the short list of assets to repurpose, reshoot, or localize.

Every new brief starts here.

Not in Figma.
Not in a brainstorm.
In the tables.

Stop Working in the Dark

Most teams think they have a creative problem.

What they really have is a data structure problem.

If your media library is just:

• A pile of exports

• Organized in pretty folders

• Remembered by “vibes” and Slack links

…then every new campaign is guesswork with extra steps.

Marketing Vandals runs differently:

• We shoot.

• We tag.

• We export performance.

• We join the tables.

• We learn.

• Then we brief the next wave based on what the tables told us.

No one remembers which video “felt good.”

We remember which patterns won.

Because once your creatives live as rows in a table, you’re not just making ads anymore.

You’re running a creative analytics engine—and turning chaos into a compounding advantage.

We didn’t “crack” the account with a new targeting trick.

We cracked it with one winning hook.

Simple story. Clear promise. Stupidly high CTR.

Then we ran that hook through our AI stack.

Suddenly, it wasn’t “one ad” anymore. It was:

• 15+ creative variations

• 4 aspect ratios (1:1, 9:16, 4:5, 16:9)

• Multiple tones, creators, and openings

• All built on the same core idea

That’s when it clicked:

The Old Playbook Is Broken

Here’s how most teams still work:

Creative team makes an ad. It runs for two weeks. Performance dips. Everyone scrambles to make “the next ad.”

Rinse. Repeat. Burnout.

The problem? On saturated channels like Meta, ads are a volume game.

The algorithm doesn’t reward your one beautiful ad. It rewards whoever can feed it enough fresh creative to find cheap pockets of attention.

One great hook isn’t the finish line. It’s the starting point.

One hook = a whole creative tree. 🌳

Your job isn’t to find a miracle ad. Your job is to find a repeatable hook and teach AI to spin it into a small army of creatives.

Here my (agile) workflow

1 Start With the Hook That Deserves 15 Lives

In crowded feeds, you don’t have room for “clever.”

You need a hook that is brutally obvious:

Problem → “This is annoying, right?” Shift → “Here’s the new way.” Payoff → “Here’s what you get.”

Think things like:

• “Turn any bad photo into a perfect profile pic.”

• “Fix your lighting in one tap—no studio, no gear.”

• “Erase anything from your photos in 3 seconds.”

Once one of these hooks proves it can pull attention and clicks, that’s the seed.

You don’t just duplicate the ad and tweak colors. You multiply the hook.

100 Hooks for Ecom

I’m going to show you how to take one video asset and multiply it into a library of 15+ high-converting variations using the modern AI stack. We aren’t just changing captions here; we are changing the video reality itself.

2 🧠 The Framework: Models vs. Tools vs. Platforms

Before we get to the tactic, you need to stop confusing the terminology. Most marketers use these words interchangeably. That’s a mistake.

To pull this off, you need to understand the AI Video Stack:

• The Engine (Models): These are the brains. Think Kling or VEO. They are the raw intelligence generating the pixels.

• The Workshop (Tools): This is where you tinker. ComfyUI is the heavyweight here. It gives you infinite control to chain models together, but it requires a “builder” mindset.

• The Factory (Platforms): This is for scale. MakeUGC and ArcadS. These platforms wrap the technology into a user-friendly interface, specifically for things like talking avatars and consistent character generation.

The “Moonshot” Idea: Stop testing text variations. Start testing reality variations.

🧪 The “Hook Library” Strategy

Here is the exact workflow to turn one video into a scalable PPC asset for E-com.

Step 1: Isolate the “Source Code”

Take your winning video. Identify the first 3 seconds (the Hook).

Maybe it’s a UGC creator saying, “Stop using regular soap.”

Step 2: Choose Your Weapon (ComfyUI vs. Platforms)

Option A: The Control Freak (ComfyUI)

If you want granular control over lighting, texture, and scene composition, use ComfyUI with the Kling model.

• The Play: Use “Video-to-Video” transformations. Keep the motion of your winning ad, but swap the environment.

• Result: The same winning hook, but now the creator is on a beach, in a luxury apartment, or in a neon-lit studio.

Option B: The Scale Master (MakeUGC / ArcadS)

If you need talking heads that look human (but aren’t), this is where the money is.

• The Play: Upload your script to MakeUGC or ArcadS. Their tech uses advanced lip-sync and facial modeling (often superior for direct-to-camera speech compared to raw models).

• Result: You generate the same winning script delivered by 5 different “Avatars” (different ages, ethnicities, and genders) in minutes.

Step 3: Build the Matrix

Now, you don’t just have one ad. You have a Hook Library.

💡 Why It Works (Psychology)

This taps into the “Mere Exposure Effect” but bypasses the “Habituation” filter.

1. Familiarity: The core message (the script/hook) remains the same, building brand memory.

2. Novelty: The visual “wrapper” changes completely.

To the human brain (and the Meta algorithm), these look like completely new ads. But they are built on the skeleton of a proven winner.

The Result: You extend the life of a winning concept from 2 weeks to 6 months.

3 Build a Media Library That Remembers What Worked

When you’re pumping out variants, chaos is a feature… until it kills you.

So you need a simple structure that answers:

• Which hooks are proven?

• Which styles won for each hook?

• What died from fatigue and shouldn’t be revived?

Even a basic spreadsheet with these columns works:

This turns your ad account into a library of proven patterns, not random uploads.

When a hook starts fatiguing, you don’t panic. You check your library, see which styles haven’t been tested yet, and spin up the next batch.

The New Question

On Meta and other saturated platforms, the algorithm rewards volume and resilience, not perfection.

So don’t ask, “What’s my next great ad?”

Ask instead:

“Which of your winning hooks is sitting there with only one life? And how fast can you give it 14 more?”

Build a Media Library That Actually Scales

I used to work on movies and ads. I come from production. Today I am a marketer who sits between creative and analytics and runs campaigns, which is a weird spot. At Lightricks, a campaign could go from brief to launch in one or two weeks, so documentation and asset sharing became critical. Since then I have carried this discipline into startups and freelance work. I have also seen teams hire actors and full crews on a hunch, the project collapses, the marketing group is let go, and there is no media library to reuse assets. No documentation, no tags, no tracking of which hook in the first three seconds worked, no trend notes, and no historical data on organic wins. A media library is not only storage. It is how you analyze performance and find shared traits of winning assets.

🎯 What This System Gives You

• Keep Web, Design, Organic, Paid, and PPC on the same backbone

• Never ask “Where is the final file” again

• Make new campaigns a duplicate, not a rebuild

• Backup for (.fig) Figma files, lotties, Rive, etc.
  

⚙️ Step 1: Duplicate Campaigns, Do Not Invent Them

Most teams start from scratch. New folders. New names. New chaos.
Fix it with one master campaign template. Duplicate it for every project.

Why it works

• Year folder keeps the archive browsable at a glance

• Date first naming makes matching creatives to reports trivial

• Chronological sort works in any file system

• Rule: time first, name after. Goodbye “final_final_v3”

Template starter

/MediaLibrary
  /2025
    /2025-12-04-[CampaignName]
      /01_Strategy
      /02_Scripts
      /03_Production
        /Footage
        /Audio
        /Stills
      /04_Edit
        /Projects
          /Premiere
          /AfterEffects
        /Exports_Working
      /05_Design
        /Source
        /Exports_Working
      /06_Analytics
        /Reports
        /Learnings
      /07_Delivery  <- single source of truth
        /Organic
        /Paid
        /PPC
      /08_Backups

Here you can download mine

Quick checklist

• Create the master template once

• Enforce date-first naming everywhere

• Duplicate, do not rebuild

Tip: Never Use Spaces in Folder Names on GitHub

• ❌ Spaces break command-line operations unless escaped (\ or %20)

• ❌ They can cause errors in scripts, automation, and builds

• ❌ Many programming tools assume space-free paths

  • kebab-case → my-project-folder

  • snake_case → my_project_folder

  • camelCase → myProjectFolder

⚙️ Step 2: Let Scripts Do The Boring Work

Even duplicating folders by hand gets old fast. Automate it.

Option 1: Python in VS Code

• Asks for date

• Asks for campaign name

• Generates the full folder tree in the right year

  
  Here you can download my Python file

  I advise further customization with Claude

Option 2: One click Mac app with Automator

• Click New Campaign on the dock

• Prompt for campaign name and date

• Build the tree and create clean Premiere and After Effects files

  Here you can download my Automatot file

Outcome
New campaigns become a one-click ritual, not a mini IT project.

Step 3: Make 07_Delivery The Single Source Of Truth

Team rule: If a file is not in 07_Delivery, it is not live, not approved, and not shareable. No exceptions.

What lives in 07_Delivery

• Final exports for Organic

• Final exports for Paid

• Final exports for PPC

Filename spec

YYYY_MM_DD_APP_GEO_PLATFORM_ASPECT_HookNN_vVV.mp4

Examples

2025_12_04_APP_US_TikTok_9x16_Hook01_v01.mp4
2025_12_04_APP_FR_Meta_4x5_Hook02_v02.mp4

Token guide

TokenMeaningExamplesYYYY_MM_DDProduction date2025_12_04APPProduct or app codeAPP, SHOP, EDUGEOCountry or region codeUS, FR, DE, LATAMPLATFORMPlatform nameTikTok, Meta, YouTubeASPECTAspect ratio9x16, 1x1, 4x5, 16x9HookNNCreative hook identifierHook01, Hook02vVVVersionv01, v02, v10

🧰 Tooling Stack That Actually Gets Used

Pick one cloud as the source of truth

• Google Drive

• Dropbox

• OneDrive

• S3 or a compatible bucket

🚦 Operating Rules That Make This Scale

• One template to duplicate, never rebuild

• Date first naming across folders and files

• 07_Delivery is the only place for shareable finals

• Keep hooks and versions two digits for sort stability

• Record learnings inside 06_Analytics after every campaign

🔎 Why Marketers Should Care

This is not an editor problem. It unlocks speed across the org.

• Analytics pulls last quarter winners in 10 seconds

• Campaign managers see which version actually shipped

• Localizations stay sane with dated, geo specific work

• Reporting ties any asset to a date, campaign, and channel

• AI workflows improve with clean, tagged inputs

🧭 The Point

Not storage. Navigation.

Marketing used to be chaotic. Now it’s… AI‑induced schizophrenia.

Every day a new video model drops. Every week someone posts a ComfyUI mega‑workflow with 900 nodes and claims it will “replace your entire creative team.” And somehow, in the middle of all the hype, you’re still trying to deliver ads that actually convert.

Here’s the truth nobody likes to say out loud: Most marketers and creators don’t suffer from lack of AI: they suffer from the classic marketing‑department disease: silos.

The (AI) Tool Trap: Engineering Before Understanding

Everyone wants to build the ultimate creative pipeline.
Today, that usually means a complex generative AI setup in ComfyUI: 55 nodes, 8 IP-Adapters, a custom LoRA stack, API triggers, webhooks—the works.

But here’s the uncomfortable question:
👉 How many of those teams have even used the basic ComfyUI playground or downloaded the free drag-and-drop workflows?

This is the same pattern that kills performance teams everywhere:
we build sophisticated systems before we understand the actual workflow.

You can’t hire a specialist and a manager for that specialist if there’s no alignment.
You usually only see this kind of aggressive scaling in VC-backed companies racing for headlines instead of results.

Meanwhile:

• Creators build in isolation, focused on mastering tools or skills: AE, Figma, Comfy, DaVinci, Acting, etc.

• Marketing and UA run campaigns in isolation.

• Leadership sits on top with almost zero visibility.

Narrow windows, Silos.
Broken communication.
Endless disappointment.

In performance, communication is not a “nice to have.”
It’s the operating system.
It’s the only way to win long term.

Before You Build an AI System Use the Damn Tool

Everyone is obsessed with API integrations, automation layers, Pipelines‑as‑a‑Service, and 20‑step prompt engineering.

But the smartest teams do the opposite:

Use the application first. Learn the tool. Ship something. Then automate.

Whether it’s ComfyUI, LTX2, Pika, Runway, Luma, Viggle, HeyGen, or any other platform, the rule is the same:

Try before you build. Use before you automate. Test before you scale.

This is the Startup Nation way: try fast, break fast, iterate, restart.

Why You Shouldn’t Pay for 40 Tools (But… You Also Should)

Yes, paying $40 across 10 different tools feels insane. But let’s do the math:

A real production day: actors, gear, location, editor, colorist, easily reaches $3,000–$25,000.

AI turns that into a few hundred dollars. And if each ad costs you $5–$40, the economics are unbeatable. With AI one hook can be 20 hooks in minutes.

This is why I built a comparison table to evaluate which tools are worth it:

Platform | Link | Free Plan | Paid Plans | API | AI Avatars | Text‑to‑Video | Text‑to‑Image | Audio | VFX | Best For | Results

It’s the simplest way to understand what to keep, what to test, and what to cancel.

The Problem Isn’t AI. The Problem Is Doing AI Alone.

UA needs to know:

• which assets perform

• which concepts scale

• which hooks die fast

• which variations convert

Creators need to know:

• what formats work on Meta

• what CTR looks like

• which angles are worth iterating

• which platform generated the winning visual

When these teams don’t speak, the entire system collapses.

Silos kill performance. Communication creates revenue.

The Hybrid Workflow: Where Creators + UA Actually Win

Here’s the new reality:

• Creators understand visuals.

• UA understands performance.

• AI multiplies both.

Your job isn’t to specialize in one software. Your job is to specialize in ship‑fast creative systems.

Yes, master ComfyUI and its nodes. But don’t do it in a cave. Do it with UA. Do it with dashboards. Do it with results.

The future belongs to generalists with taste + data. Not specialists trapped in one tool.

Final Message

You don’t need more AI. You don’t need more nodes. You don’t need more subscriptions.

You need a workflow built on communication, experimentation, speed, and iteration: where ComfyUI, Runway, Pika, Luma, and every other tool become creative boosters, not isolated toys. Everything is replaceable; nothing lasts forever. If one app starts outperforming the others, keep it and move the rest into the small experimentation budget.

Performance is a team sport. AI just finally exposed how broken silos really are.

Fix the communication, and everything else becomes easy.

Change Hairstyle

Add a product

Change Outfit

Change Location

Change Expression

Change Age

Zoom Out

Results came out more on-brand; some are just for fun, but in a few hours, I had the art, brand recommendations, and copy ideas ready.