Joe Szabo

running ads since 2016

[Joe Szabo @joe__sz]

@ron_ecomm 100%, I just imagined for a second that CPMs are 50% lower for that segment

[Joe Szabo @joe__sz]

can a brand doing 10x ROAS in Brazil actually win in the US?

had a call with one today. $250+ AOV, strong numbers at home, already getting stocked in US retail, now going heavy on paid for D2C.

my instinct says this is a smaller gamble than it looks. here's the reasoning.

every brand entering a new market faces the same wall. does the product work here. does the message land. who's the customer. three unknowns stacked on top of each other, and when it goes badly you can't tell which one broke.

this brand has all three of those problems too. a validated product in Brazil doesn't mean it's validated in the US. a refined message in Portuguese doesn't mean the angles translate. and the retail placement helps with credibility on a $250 considered purchase, but it doesn't tell you who's buying online.

so what's actually different about their position?

the data.

10x ROAS in Brazil doesn't predict 10x here. different auction, different CPMs, different competitors. and $250 in Brazil buys a much narrower, wealthier customer than $250 does in the US. same price, different person.

but they've spent real money learning things most brands entering the US haven't. which angle pulled hardest. which objection killed the most carts. which hook format still worked in month three. none of that is plug and play, but it's a thicker hypothesis list than starting from a blank page.

that's the edge. not fewer unknowns. better starting guesses.

anyone here taken a brand into a second market? what carried over, and what did you rebuild from zero?

[Joe Szabo @joe__sz]

@johnhickey1970 Happy Birthday!

[Joe Szabo @joe__sz]

What you are describing is the retail AI that is available to everybody, the version that is not even close to peak. Internally, all the AI companies are using models that are lightyears ahead of public ones.

Give AI 2027 (or AI 2040) a read, I am sure you will have a different perspective after.

[Joe Szabo @joe__sz]

@conortrains Legend! Thanks for sharing!

[Joe Szabo @joe__sz]

@andrewjfaris @TaylorHoliday thanks for putting this out, definitely listening to this today

[Joe Szabo @joe__sz]

love it when Meta auto enables shop (beside all their random AI bs) when duplicating an ad

[Joe Szabo @joe__sz]

how most ecom brands "fix" creative fatigue:

1. pause the winning ad
2. open a brainstorm doc
3. start from zero

congrats. you just threw away the only thing in the account with proof behind it.

fatigue doesn't mean the message stopped working.

it means people got tired of watching the same clip.

two very different problems.

so before any brainstorm, squeeze the winner:

→ same video, new first 3 to 5 seconds. cut five hooks. test them all. one will take most of the spend and it's rarely the one you'd bet on.

→ same angle, different skin. turn the UGC into a static, turn the static into a founder ad. your audience sees something new, Meta sees a new creative, and frequency resets.

→ same insight, colder audience. your best bottom of funnel ad already told you which pain converts. turn it into a problem aware hook for people who've never heard of you.

new concepts are for when nothing's working.

when something works, multiply it.

[Joe Szabo @joe__sz]

@finkd How about you start using this agent to fix the Ads Manager? There's a new bug poppin' up every other day...

[Joe Szabo @joe__sz]

I don't fully agree with this one. I know in the big 2026 there is a lot of hate towards attribution tools, but in this specific case you just check ad performance on first click (or you can go deeper analyzing customer journeys) using TW or Northbeam and you'll have a clearer picture if it feeds into your other ads or nah.

Of course there are nuances to this, it depends on how much you spend on this "loser" ad, and you can't track engagement overlap so you base it off click only. But still, if it spends a lot and it has low cpm/roas and looks like shit on first click, just go ahead and kill it. You can def find a better performer in your next batch.

[Kody Nordquist @KodyNordquist]

One of the worst mistakes that a media buyer can make is turning off a low CPM / low ROAS ad.

That ad is likely the lifeline to the ad that is generating all of your profitability.

I tend to look at an adset/campaign as a whole. If it's profitable and scaling well - leave things alone. The low performers are likely fueling the growth.

You're essentially creating "mini-funnels" within your adsets. Don't kill off your ability to remain profitable because of a "low" ROAS ad.

[Joe Szabo @joe__sz]

@conortrains Curious about the early days specifically. If I remember correctly, you mentioned starting at $100/day, how much did you burn before anything worked, and over how long?

Did any of the initial creatives hit (even slightly), or did it only come together with later batches?

[Joe Szabo @joe__sz]

@ron_ecomm One of our clients uses Joor. Good experience with it so far, so it might be worth a try in your case too.

[Joe Szabo @joe__sz]

@aaronmtrx If just starting out, agency, you can build up cashflow and capital much faster. If you got like $50k - $100k or more in the bank, def ecom.

[Joe Szabo @joe__sz]

the ones who win aren't better at picking creatives. they've just been wrong more times.

[Sarah 🦕 @SarahLevinger]

What’s one thing you wish everyone knew about running ads?

[Joe Szabo @joe__sz]

@herrmanndigital Seen the same, and it's not just about the overspending on shit traffic, the pacing also seems to be somewhat off.

This happened way too often in June/July.... suuuurely we will see a fix coming soon lol

[Joe Szabo @joe__sz]

I've been running the same setup on one of our accounts and have seen really good results.

The main issue was that new creatives received literally $0 spend, Meta didn't even give them a chance.

After running 1 ad per adset, we've seen the CPA drop by 50% in just a few days.

[Andrei Lunev, PhD @andreilunev]

🌶️ Meta Ads Tip (this setup is finally back)
Bid Caps
CBO - 1 ad per adset

[Joe Szabo @joe__sz]

how it feels when you have to shut off the killer ad that's been carrying your account for 3 months, until it hit fatigue...

[Joe Szabo @joe__sz]

@jforjacob

[Joe Szabo @joe__sz]

I've run $200k+/mo in Meta spend across multiple accounts this year and last.

the accounts that scaled the fastest had one thing in common.

I killed underperforming ads within 48 hours. sometimes 24.

most brands let bad ads run for weeks. "maybe it needs more data" is the excuse every time.

an ad that hasn't shown signs of life in 48 hours at your CPA target is cooked.

every dollar sitting in a dead ad could be testing something new.

the brands that win at scale are making 10x more kill decisions than creative decisions.

that speed is the whole game.