Denver is definitely one of the strangest places I’ve been.
Went earlier this year and there was just something off about the people and always felt like something weird was in the air.
Never really experienced that anywhere else
A breath of fresh air among all the slop designs lately.
We are so back
Maturing is realizing that CPM literally does not matter.
I have on account with $8 CPMs and $50+ CPAs
And another with $150 CPMs and $15 CPAs
“Just make better ads” is actually true
Genuinely what is happening with car design?
Is design just dead?
It just keeps getting worse. I blame the Teslafication of cars.
Every brand keeps pumping out the most soulless vehicles imaginable.
Car desigbn peaked in the 1990s/2000s.
The Base 2027 BMW 3 Series Is Surprisingly Good-Looking.
Read more: bit.ly/4xXNnZk
ALT The Base 2027 BMW 3 Series Is Surprisingly Good-Looking
Will retweeted
current AI models tier list:
S - Opus 5.5 & GPT-6.1 Sol : insanely smart, fast and efficient... this is my best experience ever with AI, i literally can't complain
F - everything else
jokes aside, you literally don't need anything else... at this point usage is pretty much unlimited on a $200 plan
and if you combine both Claude and ChatGPT subs, you get:
- best writing model (Opus)
- best designing model (Opus)
- best coding model (Opus)
- best model and harness at browser & computer use (Sol)
- best image generation model (Images-2.5)
This is literally exactly what I just did and it worked lol
Do not overcomplicate it
CVR is juiced. Shoutout to Claude Code
Update:
Found a winner.
Hit a winning ad during the weekend and conversions followed.
It's true when people say you are one winning creative away from cracking the funnel.
This one was all in the angle - called out a specific persona and speaks directly to that person.
That ad has a $15 CPA at the moment.
Now that this product is validated it's time to get the bulk order placed (samples are already on the way) and start working on the branding.
The good news is that I have a proven creative and landing page structure that converts.
Time to run it up.
Update:
Found a winner.
Hit a winning ad during the weekend and conversions followed.
It's true when people say you are one winning creative away from cracking the funnel.
This one was all in the angle - called out a specific persona and speaks directly to that person.
That ad has a $15 CPA at the moment.
Now that this product is validated it's time to get the bulk order placed (samples are already on the way) and start working on the branding.
The good news is that I have a proven creative and landing page structure that converts.
Time to run it up.
Looking much better here on the new product.
What I'm looking for when selecting a product:
- Subscription
- Consumable
- Retail potential
- Solves a problem
- Can sell to older markets, specifically women
- New form factor in an existing category
- Existing and growing demand (don't want to be too early)
- New variant expansion (new flavors, forms, adjacent SKUs)
- Potential for elevated brand design (not necessary but helps to stand out and charge a premium)
- 70-80% gross margins minimum
If it checks all the boxes then I spin up landers using claude. Then just duplicate the lander and swap in new product, images, and copy while keeping the same structure
Still early into this test but already beat out the last one. Found a potential winning angle too.
For creatives, simple process:
1. Swipe top formats from competitors in same category or adjacent categories
2. Use specific prompt structure to have ChatGPT breakdown the creative and write a brief
3. Use brief to have ChatGPT generate the same ad concept but applied to your brand
4. Rinse and repeat for 20 different concepts
It is that simple. Can easily make 20 solid ads in a couple hours
I've only tested about 7 of the 20 concepts to far and one has a $9 CPA at the moment. Small spend so letting it ride for the weekend
The entire purpose of this process is to find a winning product FAST.
Spin up the entire funnel, from ads to lander all with AI.
Once an idea is validated and proven to work through ads, only then should you go forward with manufacturing and placing bulk orders
The whole goal is to test extremely fast and lean until you get a hit.
Everything else can be optimized later
I could be completely wrong but I feel like being an “all-in-one” everything supplement waters down the messaging.
Trying to solve everything feels like solving nothing.
It’s the IM8 playbook but I still don’t understand what it’s really for.
Feel like it’s better to target one specific need/problem and expand into other areas with different products tailored to each.
Just my opinion
Tmrw, a longevity wellness brand from the team behind Simply Nootropics, debuted with an 88-ingredient drink mix designed to simplify supplement routines.
Dubbed Tmrw Daily, the all-in-one longevity drink mix is formulated with NMN, CoQ10, Biotin, Choline, Inulin, Manganese, and 82 other science-backed ingredients.
Tmrw Daily costs $119 for a one-month supply although customers can subscribe and pay $89 per month.
Anthony Baxter and Stuart Vaughn co-founded Tmrw and soft-launched it earlier this year. Their other supplement brand, Simply Nootropics, generates more than $50M in revenue per year.
Looking much better here on the new product.
What I'm looking for when selecting a product:
- Subscription
- Consumable
- Retail potential
- Solves a problem
- Can sell to older markets, specifically women
- New form factor in an existing category
- Existing and growing demand (don't want to be too early)
- New variant expansion (new flavors, forms, adjacent SKUs)
- Potential for elevated brand design (not necessary but helps to stand out and charge a premium)
- 70-80% gross margins minimum
If it checks all the boxes then I spin up landers using claude. Then just duplicate the lander and swap in new product, images, and copy while keeping the same structure
Still early into this test but already beat out the last one. Found a potential winning angle too.
For creatives, simple process:
1. Swipe top formats from competitors in same category or adjacent categories
2. Use specific prompt structure to have ChatGPT breakdown the creative and write a brief
3. Use brief to have ChatGPT generate the same ad concept but applied to your brand
4. Rinse and repeat for 20 different concepts
It is that simple. Can easily make 20 solid ads in a couple hours
I've only tested about 7 of the 20 concepts to far and one has a $9 CPA at the moment. Small spend so letting it ride for the weekend
The entire purpose of this process is to find a winning product FAST.
Spin up the entire funnel, from ads to lander all with AI.
Once an idea is validated and proven to work through ads, only then should you go forward with manufacturing and placing bulk orders
The whole goal is to test extremely fast and lean until you get a hit.
Everything else can be optimized later
New brand update:
Killing this product for now and moving onto the next one.
Dealing with high CPMs in this fresh ad account, got some early ATC traction but CTRs are average and CPCs too high.
Lots of competition for this one so moving down to earlier stage products for the next tests
Inspired by watching @conortrains build in public I decided to test a new brand concept.
Built everything with Claude, 100% AI and still converted on Day 2 of testing ads.
Going to see if this has legs
Will retweeted
Media buying in 1995:
• Book one magazine ad six weeks in advance
• Target "women aged 25–54"
• Run the same creative for three months
• Spend $100k and receive a sales report next quarter
• Ask how it performed and get told, "The phones seemed busier"
• Celebrate with a three-hour lunch
Media buying in 2026:
S tier affiliate ad
they SLICED open Alzheimer's brains and found a bacteria from the mouth.
it leaked from the mouth to the brain.
53 brains. 96% of them.
it's called P. gingivalis and it's the reason gums bleed.
they put it on the gums of healthy mice. it reached the brain on its own. neurons died.
that's not correlation anymore.
(PMID: 30746447)
a tree resin kills it. everything's in the replies ↓
