- This topic has 13 replies, 1 voice, and was last updated 1 week ago by bjjjohn.
June 6, 2021 at 6:20 pm #22132seohelperKeymaster
What’s up everyone! Just wanted to drop in and share some insights into what it takes to manage $20K-60K+ a day in spend on facebook in DTC ecom. (I’ve done $150K-250K revenue days on facebook, personal best in terms of ROAS was a bit over $200K in revenue at about $60K in spend on a single one of our brands, not including black friday which was insane)
Just a caveat here, how I run ads might not work for you, especially if you’re super low in spend. Different brands require different strategies, and most importantly, my own strategies are constantly developing. How I test and scale on facebook now is completely different than how it was 6 months ago for example. Also another caveat, some of the tactics we use are really only necessary at a super high level as you’ll see here, if you’re a mom and pop shop they won’t be necessary (for example running multiple facebook pages which I’ll get into).
When I first got started in online advertising, I was always searching for the ‘perfect’ way to run ads through shitty gurus, and honestly there is NO perfect way. I recommend learning the basics and devising your own strategy, which is what I ended up doing. Another thing, at lowish spend (less than $5K-10K+ a day I would say, you’re usually going to get decent fluctuations in performance day to day on facebook. Consistency on facebook comes from high spend and feeding the algo as many data points as possible.
I’m fortunate enough to be in a network of the most elite DTC brand owners so I’ve accumulated a ton of knowledge about what works at this level of scale, but this game still requires constant learning! This isn’t set in stone but its just what I’ve found works for me, so here it goes.
# Naming Conventions
Consistent naming conventions are super important for analyzing data in ad reporting at a glance. You can figure out your own but here are mine if you’re looking for a quick idea:
## Campaign Names:
TOF: Prospecting (Top of Funnel)
SS: Super Scaling (these campaigns are typically $2K-10K daily budget)
X.XX numbers at the end of campaign names or ad sets names: date of launch, i.e. 5.15 is May 15
Campaign name example: SS – TOF – CBO – Beast – 6.05
## Ad set names:
Targeting – Countries – Age – Placement – Attribution – Date of launch
E.g. Broad – US + CA – 18+ – Auto – 7dc1dv – 3.15
e.g. INT – Theme parks – US – 18+ – Auto – 7dc – 3.24
E.g. LLA – Lookalike (US, 10%) – 2+ Purchase 180 Days – US – 18+ – Auto – 7dc – 2.16
## Ad Names:
Brand – FB Page – video/image number – ad copy number – lander/advertorial number – post ID – date of launch
PP – vv100 – adc49 – lp3 – 123434341834813 – 8.08
PP – p3 – vv100 – adc72 – lp53 – 123434341834813 – 8.08
# Account Structure – Testing (Post ID’s)
## Testing Campaigns (always running):
# T – TOF – ABO – Interest Testing – 5.15
* Testing random interests found in facebook audience insights, similar interests to winning interests, etc using best 2-4 post ID’s to “feed” the pixel data
* Audience insights is phasing out so this might not be useful in the future
* Small budget ad sets of $30-50
* Can dupe winners out 2x in same campaign at slightly higher budget of $50-60
I do this with lookalikes too but I do not run interests or lookalikes with any real budget whatsoever nowadays. I literally run all creative testing and scaling with completely wide open targeting
# T – 1 – Creative – TOF – ABO – Broad – 2.18
* Phase 1 testing campaign
* All new videos/images get launched here
* I like to do them in batches of 3-4 new videos/images at a time in a single broad ad set with the budget set to 1.5-2x AOV
* Broad targeting (US + CA, 18+ so we determine how effective the creatives truly are without being skewed by very good lookalikes/interests etc. In the case of more niche products, can try broad interest targeting, like interest ‘fitness’ if selling fitness apparel or ‘coffee’ if selling coffee product, with detailed targeting expansion checked ON)
* Using best copy variation, best offer, best lander/advertorial
* Winners graduate to testing phase 2
# T – 2 – Ad Copy – TOF – ABO – Broad – 2.19
* Phase 2 testing campaign
* Take each winning winning creative from phase 1 and put it into its own broad ad set in this second campaign, testing 4-5 different ad copy angles (separate ad), still using best lander
* E.g. ad set naming convention:
* img192 – Broad – US + CA – 18+ – Auto – 7dc – 3.02
* Means img192 is the constant image across the 4 ads, with 4 different copy
* Winning ad copy variants graduates to step 3
# T – 3 – Lander – TOF – ABO – Broad – 2.19
* Phase 3 testing campaign
* Here’s what differentiates us from most ecom brands. We test a TON of advertorials, like 3-5 new advertorials a month focused on different angles. Seriously at scale this is what separates winners from losers. In this campaign I’ll also test running direct to our top sales lander as well as one of the ads. We NEVER run direct to a shopify store, we have a subdomain with dedicated landing pages/advertorials that we run to with custom checkout that converts MUCH higher and has a much higher AOV with it’s upsells.
* Take winning video/images + copy combo and test 3-5 different landers/advertorials as mentioned
* E.g. ad set naming convention:
* vv65 – adc220 – Broad – US + CA – 18+ – Auto – 7dc – 3.21
* Denotes that vv5 and adc220 were the winning variables from previous test, now testing 3-4 different landers/adverts with these two winning combos
* By now the creative has run through 3 different testing campaigns/phases. If still performing, it can be moved to bigger budget testing to see its scaling potential
* Can also be moved to optional step 4 for generating more winning post ID’s
* Also optional: Winner of this test can be moved back to step 2, testing more ad copy focused around the advertorial if a specific advertorial won during this test
# T – 4 – Page – TOF – ABO – Broad – 2.19
* Optional step 4
* This is another tactic that I don’t see many bigger brands using. In this campaign I’ll take the winning ads from the previous steps, and re-create them on 3-4 different facebook pages that aren’t our main brand page. These are ‘blog’ style pages. For example the name of one of the pages if you own a furniture store might be “Home Decor Insider”. What you don’t want to do is create fake influencer pages like “Katie’s Home’s” or something like that as that’s not allowed.
* Take the winning video/image + copy + lander/advert combo and test it on 3-4 different facebook pages to generate more winning post ID’s as mentioned.
* The point of this is multi-fold:
* Generate as many winning post id’s as possible because at scale you’ll need them
* Distributes negative feedback score away from your main brand page (negative feedback can become an issue at scale, especially last year with covid shipping delays)
* Different pages perform differently in the auction, some page names may resonate with people more and get cheaper cpc’s and cpm’s.
As you can see here the point in all this testing is generating as many winning post ID’s as possible.
# BPA – TOF – ABO – Broad – 2.19
* BPA meaning best performing ads
* This campaign is for testing all the winning post ID’s from steps 1-4 at higher budgets.
* Like to do them in ad sets with batches of 2-4 ads
* Also broad ad sets, but can also try with different LLA’s or broad interests
* Budget 1.5-3x AOV, and scale it but dupe. I.e. start the ad set at $300, if doing well over the course of 3 days or so, dupe out at double $600. From here you’ll get a sense of how it does at higher budgets. Sometimes it can do very well in the smaller 1-4 step testing, but falls flat here. If it was getting decent metrics in testing, but falls flat here, you can try duplicating the ad set and trying it again, or testing with a couple different audiences.
# DCT Testing (if applicable)
* DCT seems to work better with lower CPA products, or requires a very high budget for higher CPA products
* I haven’t had much success with dynamic creatives for testing, and especially now with the ios update facebook doesn’t show in breakdowns which creative variables are getting the purchases so they seem essentially worthless.
* If i were to do creative testing for DCT I would do something like:
* One broad ad set for each new video/image
* $100-300 budget
* 1x new video/image, 2 best copy + 1 new copy, 1 best headline + 1 new headline
* Pull winning post ID’s out, follow testing steps 3-4 above to test different landers/adverts/offers/fb pages
* What i DO like dynamic creative for lately is time sensitive sales, like black friday where I don’t have a ton of time to test stuff. What I usually do is toss in a ton of my existing winning videos/images/copy/headlines (I might just add a black friday sale specific line to the top of the ad copy) running to my best advertorial/lander and let it rip at about $1000 a day budget. If it does good after 1 day I’ll duplicate it out into a cost cap/bid cap at $5K-10K a day or whatever
# CBO Angle testing:
This is a CBO with 5-7 ad sets, each ad set is a separate angle containing winning ads from the above campaigns, that get added to their respective angle ad set. Budget is about $1K per day for me. All ad sets wide open broad targeting
Here’s the fun part. My methods of scaling nowadays have evolved with what works on facebook. The good thing is with this level of spend I learn quickly what is or what does not work on facebook anymore so it keeps me current. I have a few different scaling campaign structures that I’m currently running simultaneously. This is what I’m finding works right now:
# Scaling Campaign 1
Lowest cost CBO -> 1 ad set (completely Broad) -> Best 6-10 post ID’s from testing campaigns. I’ll add new post ID’s/turn off ads if performance is on a decline over a week period. I will increase the budget by 20-30% a day if performance has been consistently good over a 2-3 day period.
# Scaling Campaign 2
Same as above, except this campaign is made up entirely of non-brand page post ID’s from the page testing campaigns
^ These campaigns are both often running at $2-5K+ a day
# Scaling Campaign 3 – Bid Cap ABO
I duplicate the best ad sets 3x from the CBO angle testing campaign into a separate ABO campaign, each running at a different bid. Ad set one’s bid cap is set to target CPA + 25%. So if my target cpa for example is $50, the bid cap would be set to $62.5. Ad set two is set to +50% ($75) and ad set 3 is set to +100% ($99.99, I round down in this case as my theory is if i set the bid to $100, I’ll be put into a higher tiered auction pool and may get outbid, dont quote me on this lol)
I set budgets at about $1K-5K per ad set here. And because you can have one of these campaigns for each angle, you can see how quickly scale can build up here.
# Scaling Campaign 4 – Cost Cap ABO
* Same as above, but the cost caps for this campaign will be +15%, +25% and +50%
# Scaling Campaign 5 – Cost Cap CBO
* 4 completely broad ad sets duplicate of each other, all with the same cost cap. This campaign contains the best 6-12 post ID’s overall from all testing campaigns. You’ll have to play with the cost cap here to get it to spend properly. This campaign is generally a big one for me usually with a $10K daily budget. I’ll also have a minimum ad set spend of about 3-5x the CPA set for each ad set
The point in having so many scaling campaigns is multi-fold:
* Prevents reliance on a single scaling campaign on poor days. For example one or two of these campaigns might do mediocre one day, but the rest are crushing and make up for it
* Optimizes differently and hits different points in the auction by utilizing both CBO and ABO
If you want to go crazy you can also take these exact scaling campaigns and scale them across multiple accounts as well. For that $200K day I had $10K+ cost cap campaigns scaled across like 4 different accounts.
And that’s it! Like I said this is not end all be all of running ads, just what I’ve evolved to do after spending high budget day in and day out for single brands
The most important thing about scaling with this level of spend and what separates the brands who do great online and those who don’t is content. We’re testing about 10-15 NEW video ads per WEEK + variations of winning videos on top of that (different hooks for example)
Audience “hacking” is no longer really a thing and hasn’t been for a while. I don’t run any interests at scale for the most part and lookalikes I barely use nowadays either (they worked great last year up until Q3-Q4). literally just wide open 18+ targeting. broad targeting might not work as well if you have a super niche brand
It’s true that nowadays facebook has certainly become a lot more difficult. We aren’t spending as much on it compared to last year (though still a lot and it’s our primary DTC revenue driver still), we’re trying to crack other traffic sources to diversify for cold traffic, especially with Tik Tok, Youtube, GDN and Snapchat. Snap is spending about $3K-5K a day at so-so ROAS. if anyone is an expert at tik tok hit me up and let’s connect!June 6, 2021 at 6:39 pm #22140Happy–Human
Have you tried Google Shopping?June 6, 2021 at 6:47 pm #22133Cryb3by
Why do you direct traffic to a custom landing page instead of a product page?June 6, 2021 at 7:32 pm #22135heysupbruh
What are advertorials? That’s a new term for me!June 6, 2021 at 7:32 pm #22141wuzzar321
Hey, thanks for the write-up, do you use responsive ads at all?June 6, 2021 at 7:51 pm #22134psynomix
How long did it take you to get the ability to spend $60,000 in a day? Did it require a special Facebook rep? Did it require asking Facebook manually for the increase in spend?June 6, 2021 at 8:44 pm #22136ultronic7
Would you advise in scaling by duplicating Winning adsets 3-5 times in a cbo and also more importantly have any of ur campaigns / adsets been put into limited learning and how did you deal with that . I have been told to ignore if performance is good?June 6, 2021 at 9:10 pm #22137chewster1
Wow tanks for the insight and everything behind it.
At my work we would cancel this as a ROAS of 1.3 (200/150) is way too low it’d only really break even.
We even self manufacture so have very low costs already, am genuinely curious how there is any profit margin left in your strategy, you must have very low costs as a business?June 6, 2021 at 9:11 pm #22142rsjabberv
Hi thanks for sharing. Q; how do you deal with audience overlapping? Do you exclude and merge if needed?
(sorry if you already mentioned this in your post, I will read it in full later, when I am not this tired..)
Thanks!June 6, 2021 at 9:29 pm #22138mmnagra31
Great read, will analyze this tomorrow. Ad account management is the first mistake many seem to be doing (me too).
One question. I am selling watches, do I need to just use “watches” as interest, would that count as the broad interest you mentioned.
The landing page testing is a game changer, people that come to the landing page on regular ecom sites most of the time scroll and leave the site, the only things they see are product, price, and some other basic content that they see everywhere, free shipping, free returns blabla.
I want to test my landing pages with story telling, creating a connection with the customer, getting their attention. Then compare Google Analytics funnel data to see which landing page is the winner (don’t forget to use UTM).June 6, 2021 at 9:57 pm #22143Perdurabo306
This was great, the campaign/ad naming conventions will be very helpful, thank you. Question for you, what’s the best place to go to keep in the know re: what’s working in the d2c marketing world?June 6, 2021 at 9:58 pm #22144jamesbwbevis
Does only e-commerce work on fb?June 6, 2021 at 10:08 pm #22145bjjjohn
That naming convention 😍! Out of interest, who is making the calls on creative? I’ve seen great set ups with creatives that hold everything back. Is the data feeding back into the creatives to refine what works or is it on creative feel for each product?June 6, 2021 at 10:43 pm #22139alvingjgarcia
I don’t have any questions but I didn’t just want to say thank you for sharing this. It will definitely help so many people and it brings real tangible value to this sub. So thank you!
- You must be logged in to reply to this topic.