I get a lot of questions about how I test creatives. So here's the exact process I use to test creatives on accounts spending $500-$50,000 per day:
> Audience I do all of my tests with the winning audience in the ad account In most cases/ad accounts, that's Broad, but there are few ad accounts where I use single interest or stacked interests for all of my tests I usually exclude purchase X days and often web visitors 30d
> Setup for image tests I usually use DCO With a total of no more than 20 possible combinations (e.g. 1x2x8=16), so FB doesn't spread the budget too thin Usually 1 (winning) copy, 1 or 2 headlines, and 4-8 image variations After a while, I check the breakdown by asset...
extract winning combinations with the post ID and start running that winning ad in the scaling campaign and ASC campaign (where usually all the winners are)
> Setup for video tests Usually variations of creatives with a different hook, length or thumbnail I'm doing all of my tests grouped per concept or angle Meaning, each new angle is being tested in its own ad set, always in ABO
I build the ads manually Basically, 1 ad set, multiple ads (usually 2-5), and letting FB to chose a winner(s) Sometimes I use DCO, but sometimes FB is funky and can't show thumb-stop and hold rate so I like to do it this way Again, the winner is getting transferred to ASC
> Setup for new iterations of a winning creative Usually, this is a new hook, a faster video, a different thumbnail, etc.. The goal is here to try to beat the "original", that's why I either: - test only iterations on their own or - test against "original" with no social proof
Cause otherwise, new creatives would have a low chance of beating the existing winner considering both FB and users would favor the ad with accumulated social proof It's really straightforward how I'm doing my tests Everything is clean & easy to evaluate & it works
What about you - how are you setting up your tests?
Do you run a DTC brand and need help scaling your store? 🚀 We'll help you maximize the efficiency of your FB ad account and add an extra 5-6 fig/month WITHOUT any extra ad spend Go here & stop making the same mistake and burning your money on ad spend: https://bit.ly/inspiremarin
@IstvanicMarin Aligned here
@IstvanicMarin this is awesome. super detailed!
@dtc_alchemist That's the only way to make it work
@IstvanicMarin Bro how many times are you gonna tweet this 😂😂😂
@gold_tesla This is the most frequent question I get, so it makes sense to repeat it every 2 months + I always add something extra
@IstvanicMarin @readwise save thread
@IstvanicMarin Is it really so easy and simple as you make it out to be :) :) or is it your process driven approach that makes it look simple or your deep understanding?
@2105Rakesh It's really that simple as explained
@IstvanicMarin Testing Mantra is litt🔥
@IstvanicMarin Side note. What are you using to make the visual board? Miro?
@RealBasilKhan this was done in Figma
@IstvanicMarin Hey @IstvanicMarin What's the max audience size you test with broad? 20m? If it's bigger than that, how do you choose which interests, etc?
@DevrathSoni Broad is broad, no restrictions at all I don't care about max size audience
@IstvanicMarin This framework is the red pill of FB Ads, I am doing pretty much same with 2-3 changes
@IstvanicMarin @readwise save thread
@IstvanicMarin Gm! i´m appliying your creative testing structure but apparently only works if i test my videos with DCO, any recommendation? the problem of using DCO is that i can´t scale because i lost all the social proof. I´m using TikTok creative formats to give you more info, thanks marin
@pedro_ecom take winning combinations from DCO with post ID that will carry your social proof and add them to your regular/scaling campaigns
@IstvanicMarin How long will you run a DCT for image ads before turning it off if it's not converting?
@JacWilliams13 i would check the breakdown to see on what assets FB spent money if it spend equally on most, and it's bad after X time AOV (depending on the product price), I would kill it If it spent on just one, and it's bad, I would launch new DCO without that one asset