Pinned Tweet
Suno Ai Song Ads are absolutely ripping for us
Launched this ad few months back and it's still the top spender
If you haven't tried it yet give this a shot
Suno Ai Ads might be your next winner
if you're in ecom or doing retention marketing for ecom brands
you don't wanna sleep on Retention University by @ayush_mach33 and @retentiongoat
They are hands down the best in the game
Saw them grow legit from the ground level and super happy for their launch
Launched Retention University today with @retentiongoat
Ngl went way better than I expected.
Only going up from here. We will make this the best resource out there for brands that want to be cracked at Retention
It cost me millions of dollars to learn this about creatives
We stopped chasing creative volume and scaled this ad account to multi-7-figure months
Here's what everyone gets wrong
They think volume means 80+ new concepts a month. So every week the team starts from zero, builds something brand new, it dies, and you learn nothing you can use
Volume isn't the problem. What you're putting volume behind is
Once you've got your winning avatar and your winning concept, milk the fuck out of it
We take one proven concept and build 8-10 iterations of it. Same angle, same claim, same promise
Native listicle static
Comparison static, their solution vs yours
Text-heavy screenshot style
UGC talking head
AI UGC
Pixar style animation
AI song ad
POV handheld
Founder ads
Voiceover over b-roll
That's one concept, 10 ads, and every one of them reaches a different person who would have scrolled straight past the other 9
And here's the part nobody talks about
When you iterate a proven concept you're running the only clean test in the account. The message is already validated, so you're moving one variable against a known baseline.
When it loses, you learn something specific
A brand new concept changes the angle, the hook, the format and the proof all at once.
When it fails you've got nothing to learn
My agency helps 7-9 figure DTC brands scale with Creative Systems and Meta Ads
If you're spending at least $100k/mo on Meta
Book a Call with me: manofy.com
Manoj Bash retweeted
I posted about AI Song Ads a month ago. It went super viral and now Iman Gadzi taking it over to the coaching industry
Bro's using hooks from Reddit confessions and driving traffic to the advertorial page is quite a smart move lol
AI Song Ads + advertorial = $$$
If you're doing ecom and still sleeping on AI Song Ads idk what the fuck you're doing
Manoj Bash retweeted
Took a few days off X and came back to everyone fighting about incremental attribution
So here's the simple version:
When you open your dashboard
Meta claims a number
Google claims a number
Klaviyo claims a number
Add them up and the total is higher than what Shopify revenue is
Nobody's lying. They're all counting the same sale.
That's the problem incrementality exists to solve.
Platform attribution asks: which ads did this buyer touch before purchasing?
Incrementality asks: would this purchase have happened anyway?
They're completely different questions
Clearest example is branded search
Someone googles your brand name
Your paid ad sits directly above your own organic listing They click the paid one
Google reports a conversion and a beautiful ROAS
You just paid for a click you were already getting for free.
Retargeting does the same thing
Someone adds to cart, sees your ad two days later, buys.
Meta takes the credit.
But that person also got your abandoned cart email, your SMS, and still had your tab open
That revenue was already yours
So use platform attribution for creative decisions
Ad A vs ad B inside the same inflated system is still a fair comparison
And use incremental attribution for budget decisions
Manoj Bash retweeted
Long form Crazy Drama Ai Ads are absolutely ripping for us
Launched this ad few months back and it's still the top spender for a supplement brand
If you haven't tried it yet give this a shot
Crazy Drama Ads might be your next winner
Took a few days off X and came back to everyone fighting about incremental attribution
So here's the simple version:
When you open your dashboard
Meta claims a number
Google claims a number
Klaviyo claims a number
Add them up and the total is higher than what Shopify revenue is
Nobody's lying. They're all counting the same sale.
That's the problem incrementality exists to solve.
Platform attribution asks: which ads did this buyer touch before purchasing?
Incrementality asks: would this purchase have happened anyway?
They're completely different questions
Clearest example is branded search
Someone googles your brand name
Your paid ad sits directly above your own organic listing They click the paid one
Google reports a conversion and a beautiful ROAS
You just paid for a click you were already getting for free.
Retargeting does the same thing
Someone adds to cart, sees your ad two days later, buys.
Meta takes the credit.
But that person also got your abandoned cart email, your SMS, and still had your tab open
That revenue was already yours
So use platform attribution for creative decisions
Ad A vs ad B inside the same inflated system is still a fair comparison
And use incremental attribution for budget decisions
Long form Crazy Drama Ai Ads are absolutely ripping for us
Launched this ad few months back and it's still the top spender for a supplement brand
If you haven't tried it yet give this a shot
Crazy Drama Ads might be your next winner
My agency helps 7-9 figure DTC brands scale with Creative Systems and Meta Ads
If you're spending at least $100k/mo on Meta
Book a Call with me: manofy.com
I posted about AI Song Ads a month ago. It went super viral and now Iman Gadzi taking it over to the coaching industry
Bro's using hooks from Reddit confessions and driving traffic to the advertorial page is quite a smart move lol
AI Song Ads + advertorial = $$$
If you're doing ecom and still sleeping on AI Song Ads idk what the fuck you're doing
My agency helps 7-9 figure DTC brands scale with Creative Systems and Meta Ads
If you're spending at least $100k/mo on Meta
Book a Call with me: manofy.com
Manoj Bash retweeted
A lot of you asked how we come up with the AI song concepts. Here's how we actually do it
First, why they even work
A song has a totally different fatigue curve than a script. A spoken claim gets worse on view three. A catchy hook gets better on view three
That's the whole reason this ad has been our top spender for months while everything else died out in a few weeks
Okay, now let me explain the method
1. The song is about their life, not your product
biggest mistake you'll make is writing a jingle. verse is the problem, chorus is the turn, product shows up once at the end
2. The lyrics come from reviews
Pull your 2 and 3-star reviews, your support inbox, Reddit
You're hunting for sentences people actually said. Our best lyric was taken almost word for word from a review.
3. Pick the emotion before the genre
frustration → country or blues
relief → acoustic, folk
confidence → hip hop
absurd → pop punk
genre comes from what they feel, not what you like.
4. Write the chorus first and make it one sentence
your chorus is your hook. if that line wouldn't work on its own as a static, the song won't convert
5. Be way more specific than feels comfortable
"3am" not "at night." melody carries specificity in a way a script can't, and specific is what makes it relatable
6. Generate long, cut short
Suno gives you 4 minutes you need 15 seconds. Make 20, keep 2, cut the best 8 bars.
7. Keep the visual cheap and native
The song is the ad. Over-produce the visual, and you kill the native feel that got you the attention
8. One winning song is ten ads
Same lyrics, new genre
Same melody, new lyrics
lyrics on screen vs off
That's why ours is still running. We didn't find an ad, we found a winner template
Steal it.
A lot of you asked how we come up with the AI song concepts. Here's how we actually do it
First, why they even work
A song has a totally different fatigue curve than a script. A spoken claim gets worse on view three. A catchy hook gets better on view three
That's the whole reason this ad has been our top spender for months while everything else died out in a few weeks
Okay, now let me explain the method
1. The song is about their life, not your product
biggest mistake you'll make is writing a jingle. verse is the problem, chorus is the turn, product shows up once at the end
2. The lyrics come from reviews
Pull your 2 and 3-star reviews, your support inbox, Reddit
You're hunting for sentences people actually said. Our best lyric was taken almost word for word from a review.
3. Pick the emotion before the genre
frustration → country or blues
relief → acoustic, folk
confidence → hip hop
absurd → pop punk
genre comes from what they feel, not what you like.
4. Write the chorus first and make it one sentence
your chorus is your hook. if that line wouldn't work on its own as a static, the song won't convert
5. Be way more specific than feels comfortable
"3am" not "at night." melody carries specificity in a way a script can't, and specific is what makes it relatable
6. Generate long, cut short
Suno gives you 4 minutes you need 15 seconds. Make 20, keep 2, cut the best 8 bars.
7. Keep the visual cheap and native
The song is the ad. Over-produce the visual, and you kill the native feel that got you the attention
8. One winning song is ten ads
Same lyrics, new genre
Same melody, new lyrics
lyrics on screen vs off
That's why ours is still running. We didn't find an ad, we found a winner template
Steal it.
My agency helps 8-9 figure DTC brands scale with Creative Systems and Meta Ads
If you're spending at least $250k/mo on Meta
Book a Call with me: manofy.com











