Our brand did $1.7M in sales in ONE day.
AI is changing ecom.
Agents already run key functions for us.
They help automate our best people's proven work.
I filmed a 25 min vid on how we're adapting.
& what still matters.
Like + RT + comment "AI" (must follow) & I'll DM it.
Flakes spends roughly half of this 70-second ad building the reason to switch.
The product reveal comes around 35 seconds in. Before that, the animation gives the viewer a different explanation for their frustration.
Hereās the breakdown:
1. Challenge the usual belief.
The opening introduces the familiar idea that dandruff means a dry scalp, then questions it.
2. Give the cause a character.
Animated creatures turn the adās fungus explanation into something you can see and follow.
3. Question the usual fix.
Oil being poured onto the scalp illustrates the adās claim that some attempted fixes can feed the problem.
4. Make failed attempts relatable.
A bathroom full of bottles captures āIāve tried everythingā in one frame.
5. Reveal the alternative.
Only after that setup do the Flakes bottles appear. The reveal answers the frustration established earlier.
6. Visualize the proposed mechanism.
The animation shows the formula confronting the same creatures introduced at the start. Problem and solution share one visual language.
7. Show the desired everyday outcome.
The story moves from scratching and frustration to a character enjoying his day and feeling comfortable with his hair.
8. Ask for the switch.
āSwitch to Flakes.ā A clear action, followed by a product and benefit recap.
My takeaway: identify the belief your customer brings to the problem. Build a clear, supportable explanation for why your approach is different. Make that explanation visible.
Which visual carries the strongest idea: the creatures, the bathroom full of bottles, or the animated mechanism?
The easiest way to get someone to watch a full ad?
Donāt make it feel like an ad.
Tell a simple story.
While theyāre following the cartoon, you educate them on the problem.
Then you casually introduce the product as the solution.
No hard pitch. No forcing it.
Just a story that naturally sells.
Full tutorial here: https://t.co/wN6YOnr9wO
Created with Nano Banana š on Gemini
Prompt:
Create a hyper-realistic cinematic miniature construction scene deep inside a misty tropical rainforest. Tiny workers use tweezers, drills, hammers, ropes and pulleys to build a luxurious treehouse between enormous moss-covered trees.
Show extreme macro shots of hands assembling wooden beams, hammering nails, sanding timber, fastening rope bridges and installing tiny glass windows, with rain droplets, insects and wet leaves adding realism.
End with a breathtaking wide reveal of the completed treehouse glowing warmly above the rainforest canopy, surrounded by mist, lush vegetation and a distant waterfall.
Extreme macro, photorealistic miniature diorama, realistic hands, cinematic camera movement, shallow depth of field, natural diffused light, detailed wood and moss textures, realistic shadows, 4K, seamless transitions, immersive scale, no text, no logos.
Ultra-realistic macro close-up of a person carefully cleaning a small white wireless earbud using fine precision tweezers, gently removing tiny dust, lint, and hair particles from the speaker mesh. The earbud rests on a soft blue microfiber cloth on a warm wooden desk, with the second earbud softly blurred in the background. Natural warm indoor lighting, shallow depth of field, cinematic bokeh, highly detailed hands and earbud texture, realistic metal tweezers, subtle hand movements, satisfying precision-cleaning ASMR aesthetic, smooth slow camera push-in, photorealistic 4K, vertical 9:16, no text, no logos, no
Made with Seedance 2.5
{
"title": "Tiny Workers Repairing a Broken USB Drive",
"duration": "15 seconds",
"style": "cinematic miniature stop-motion, photorealistic",
"prompt": "A macro cinematic scene of a damaged USB flash drive lying on a dark wooden desk. At the beginning, the USB drive is visibly cracked, dirty, and covered with small broken fragments. Dozens of tiny construction workers in colorful safety helmets and work uniforms gather around it, inspecting the damage and carrying miniature tools and equipment. They actively repair and rebuild the USB drive: some workers remove debris, others weld and assemble tiny components, while others carry pieces across the surface. As the repair progresses, the damaged brown exterior gradually transforms into a clean, sleek metallic silver USB drive. The workers continue polishing and inspecting the finished device. Maintain realistic miniature scale, detailed textures, believable shadows, cinematic depth of field, subtle camera movement, warm desk lighting, and highly detailed macro photography.",
"camera": {
"shot": "macro close-up",
"movement": "slow cinematic push-in with subtle lateral movement",
"focus": "shallow depth of field focused on the USB drive and miniature workers"
},
"lighting": {
"style": "soft cinematic practical lighting",
"details": "warm highlights, realistic reflections on metal, natural shadows"
},
"animation": {
"workers": "continuous coordinated construction and repair activity",
"transformation": "gradual visual transition from damaged brown USB casing to polished silver casing",
"motion": "tiny tools moving, workers carrying parts, welding sparks, debris being cleared"
},
"negative_prompt": "large humans, oversized workers, cartoon style, unrealistic proportions, floating objects, distorted USB drive, excessive camera shake, blurry details, text, logos, watermark"
}
AI is about to make brand characters a MUCH bigger part of performance marketing.
Brands can now use the same animated character across every single ad.
Same face.
Same personality.
Same voice.
Different concepts every week.
Eventually, people wonāt just recognize the product.
Theyāll recognize the character selling it.
Think Flo from Progressive or Jake from State Farmā¦
Except now even a small DTC brand can build one.
I think the smartest brands are going to create recurring animated characters that audiences see so often, they become synonymous with the brand.
This is going to be a massive part of AI advertising.
If you want my complete character production framework PDF, comment ābrandā below this video.
Our avatar research method that finds $200K/month angles:
1. Read 300+ reviews (yours + competitors)
2. Every time you see "As a ___" or "I'm a ___" ā Screenshot it
3. Organize by frequency
4. Pick the top 3 identities
5. Build entire campaigns around each one
We helped one client find "night shift nurses" this way.
Now they do $400K/month serving ONLY that persona.
DM āAVATARā and weāll do the same for you.
(your brand must be doing $100k+ months)
I made a Claude skill that animates your static ads.
One of the most overlooked ways to boost creative diversity: take your already-validated statics and turn them into video.
Built an agent powered by Claude + MakeUGC that makes it a piece of cake:
- Start with one of your Meta statics
- It analyzes it and generates two video concepts a simple animation, and a more creative one
- You approve what you like
- It animates them into eye-catching videos with Seedance
3 minutes, start to finish. The most efficient way to hit Meta Advantage+ video requirements.
Your only job is taste. Guide it, pick the best options.
Giving the whole skill away. Comment "MAKEUGC" and I'll send it to you
Seedance 2.0 with GPT image 2.0
Prompt:
CAMERA
Third person handheld smartphone camera operated by a friend walking a few steps behind the girl. Casual vlog footage, not professional cinematography. Natural handheld shake, subtle footsteps vibration, imperfect framing, occasional autofocus adjustment and realistic phone-camera exposure. The camera follows her naturally at walking speed.
LOOK
Ordinary smartphone video recorded by a friend. Natural lighting, realistic skin texture, natural hair movement, normal phone sharpening and compression, slightly imperfect exposure. No cinematic polish or beauty filter.
STYLE
Simple everyday lifestyle vlog. Relaxed, spontaneous and believable. No dramatic acting or cinematic choreography.
CHARACTER
@Image1
The same young American woman from the reference image. Long black hair, naturally pretty face, brown eyes, cream oversized sweatshirt, loose baggy pants and casual sneakers. Keep her exact appearance consistent throughout.
SETTING
A normal convenience store in Tokyo, Japan. Ordinary Japanese neighborhood outside, glass storefront, bright fluorescent interior, narrow aisles, refrigerated drink cases, Japanese snacks and drinks, baskets and a few ordinary customers.
STORYBOARD (15s, ONE CONTINUOUS TAKE)
(~0 to 4s, Following Shot)
The girl walks normally toward the convenience store. The camera follows casually from several feet behind her at walking speed. Her hair and oversized sweatshirt move naturally as she walks.she turns back towards the camera and gives a smile and a gesture to camera to follow her then She reaches the entrance and opens the glass door.
(~4 to 7s, Entering)
The camera follows her inside. She walks naturally down the aisle toward the refrigerated drinks. The camera stays behind and slightly to her side with realistic handheld movement.
(~7 to 11s, Refrigerator)
She reaches the refrigerator and stops. She opens the glass door, looks through the drinks, selects one bottled drink and takes it out. The camera moves slightly closer while remaining handheld.
(~11 to 15s, Ending)
She closes the refrigerator door while holding the drink. She turns naturally toward the camera, looks at the person filming her and gives a small genuine smile while holding the drink. The camera remains slightly shaky and ends naturally.
AUDIO
Realistic synchronized location audio throughout the entire recording. No dialogue, no talking, no narration, no voiceover and no music.
Include natural footsteps on the pavement and store floor, subtle clothing and hair movement, distant Tokyo street traffic, passing bicycles, faint outdoor city ambience, the glass entrance door opening and closing, the convenience-store door chime, quiet footsteps from other customers, refrigerator compressor hum, the refrigerator door seal opening, bottles and shelves making subtle sounds when she takes the drink, and normal low-level store ambience.
Audio should sound like it was captured directly by a smartphone microphone positioned several feet away. Slightly imperfect recording quality, natural volume changes as the camera moves closer and farther away, realistic room acoustics and subtle background noise.
REALISM RULES
One continuous take. The girl always walks forward naturally. The camera operator physically follows her from behind or slightly beside her. All actions happen in real time and in order: walk, enter, walk to refrigerator, open refrigerator, take drink, close refrigerator, turn toward camera, smile. Natural ambient audio continues continuously with no artificial sound effects.
NEGATIVE PROMPTS
Speech, talking, dialogue, narration, voiceover, music, lip-sync, silent environment, missing footsteps, missing door chime, artificial sound effects, walking backward, reverse walking, selfie camera, first person POV, floating camera, drone footage, perfect stabilization, cinematic tracking, cuts, transitions, time jumps, speed ramping, hyperlapse, exaggerated acting.
took the pixar version and pushed it into two more looks today, SIMPSONS style everyone knows, and a second FAMILY GUY style
it literally took a single prompt and now i have 2 whole new formats to test
what actually happened:
- generated one style-reference still per look, first frame of the pixar video redrawn in the new visual language, character design and board linework both restyled
- fed that still in as the style reference alongside the original video as the action/timing/dialogue reference
- locked the rule that only the rendering style changes, nothing about the action, the board content, the pointing choreography, or the voice performance
- same gaze rule, same pronunciation lock, same beat-by-beat timing as every version before it, none of that gets touched, only the skin
RT + reply 'CARTOON' and i'll send you the full workflow so you can run this on your own ad
(must follow so i can DM)
Secret of the DTC Universe #6: Avatar research is overrated.
GASP.
I know.
Let me explain:
These days, a lot of people get taught that the key to scaling eCom and DTC comes down to developing a bunch of highly specific avatars.
And, in case youāre wonderingā¦
An avatar is basically the character profile for one of your potential buyers.
So for example, letās say you sell a gut supplement and your average buyer is a 52-year-old woman.
If youāre an AvatarMaxxerā¦
You might create an in-depth brief and character sketch for each of the following avatars within that market:
- A 52-year-old woman dealing with menopause.
- A recently divorced woman trying to get back into the dating world.
- An empty nester who is suddenly focused on herself again.
- A career-focused woman who has become more concerned with her appearance.
- A woman who is struggling with confidence as she gets older.
- A busy working mom who feels tired and run down
And you might even do additional versions of these avatar briefs along different ethnic lines as well.
So likeā¦
A version of each avatar profile for white women, black women, Asian women, Hispanic women, etc.
Do that thoughā¦
And it means your next month or two is going to be filled with endless deep research into the psychographic profile for each of those groups.
Youāll go to Reddit threads, read comments on ads, and basically do all of the research stuff I taught in the āRā section of the original RMBC Method.
And thenā¦
Youāre going to need to build out some giant system for generating different ads for every single one of those avatars.
Thereās a certain class of people who hear all of this and get super revved-upā¦
āSystemsā¦researchā¦daddy like!"
But the truth isā¦
Those people are hardly ever effective marketers or particularly successful peopleā¦
And even if they areā¦
Theyāre still shooting themselves in the footā¦
Because this whole process slows everything down tremendously.
And, much worseā¦
In the vast majority of cases, doing all of this is only going to lead to very incremental gains at best.
The reason why is simple:
Itās because all of those ādifferentā avatars have much more in common than they have not in common.
For example - letās look at that 52-year-old female buyer.
No matter whether sheās black or whiteā¦
Married or divorcedā¦
Has adult children, young children, or no childrenā¦
Has a 9-5 or is a stay-at-home-momā¦
Then she almost certainly is STILL:
-Dealing with some stage of menopause.
-Experiencing less energy than she did when she was 42.
-Enjoying more confidence than when she was younger, but feeling like thatās in tension with the fact that her physical appearance doesnāt match her mental state (she doesnāt feel old, but she sees or feels signs of aging).
- In a romantic relationship, or has been in one recently.
And, if sheās in our target marketā¦
Then she STILL has a similar cluster of symptoms as virtually every other woman in that market:
Bloating, low energy, weight issues, fatigue, worse sleep, mood swings, hair shedding, etc.
Thatās the shared center of the TAM (total addressable market)...
Itās massive (see image attached with this post).
And this is why that center is where you should focus 90% of your efforts when it comes to writing ads.
It just makes the most logical senseā¦
And doing this lets you get OUT of research mode and INTO testing and scaling so much faster.
This last part is HUGEā¦
Because Iām pretty sure most of the people who love AvatarMaxxing are actually just using it as a cope for not taking action.
It gives them the ILLUSION of productivityā¦
When in reality, theyāre just pussyfooting aroundā¦
And putting off doing the ONLY thing that mattersā¦
Which is getting your ads out into the wildā¦
Putting real money behind themā¦
And then letting the market tell you whether or not youāre on the right track.
And, to be clear, this doesnāt mean your ads wonāt have any specificity.
They totally will.
Theyāll be:
Specific about the problem.
Specific about the desired outcome.
Specific about the mechanism.
Specific about the reason the person has struggled before.
Specific about the details that make the product believable.
All of that specificity is fineā¦
But none of that means you need to spend months building out some database of highly nuanced avatar profiles.
You can write one ad about menopauseā¦
Another about appearanceā¦
Another about confidenceā¦
Another about the frustration of feeling like your body has changedā¦
And all of those ads will still speak to the massive center of your target market.
-SPG
P.S. This is an important P.S. so I hope you read it.
Iām not saying you should never speak to other markets.
Of course you can test ads and landers that target different demos.
Thatās fineā¦
Although you should really only do that once youāve reached a ton of scale, and it really seems like finding more growth on at least one traffic channel will be hard.
So for exampleā¦
Maybe youāre doing nine figures on Meta selling a gut supplement to women who are, on average, 50+.
Now you want to go after menā¦
Obviously it makes sense to spend time doing research about why guys buy gut supplementsā¦
And further narrowing down who you target to a specific age-range again, like 50+ dudes.
It would be crazy not to.
And of course, itās also fine if you want to test ads speaking to specific segments of your market as well.
For example, letās stick with guysā¦
Maybe you test different ads that are targeted towards different pain pointsā¦
Or you do some ads targeted towards fit dads, while others are focused on out-of-shape retirees.
Thatās cool. Go nutsā¦
But you STILL donāt need to develop some elaborate avatar profile to do this.
Especially in the days of AI.
LLMs were trained on pretty much the entire corpus of human knowledge (including a LOT of Reddit threads)...
And your ads are mostly going to be like 1-4 minutes longā¦
So you donāt need to unlock the secrets of the universe hereā¦
Itās often as simple as being like:
āIām still in great shape even though Iām in my 50sā¦which is why I donāt understand whatās going on in my gut.ā
Orā¦
āWhen I retired after 40 years, I thought Iād finally be able to relax. Instead, Iām constantly stressed about going to the bathroom.ā
Oh wow.
Iām a witch!
I just figured out how to talk to two segments in a different wayā¦
And I didnāt spend 4 weeks of mental mastur-you-know-what to make it happen.
P.P.S. I want to be very clear.
I am not writing this email to ācall outā the experts/gurus/people who advocate for AvatarMaxxing.
If it works for them - thatās awesome.
If it works for their students - also fantastic.
There is no right way to do anything.
I just personally donāt think you need to spend much time on avatars.