Scaling CBO - My case study to $2,500/day budget (5)
So as the headlines says, we could only reach $2.5K ($2.57K/day at max).
When we scaled pass $2.57 it'd go nuts...
As you can see, that's not a case-study of "I made it! Go copy it!".
That's quite the opposite, it's a "do NOT do it! As it doesn't work" 
Here's what I was doing, PLEASE TELL ME WHAT'S WRONG WITH IT
1. The structure of the account was quite simple. First I'll make a campaign (broad) non-CBO with 4 adsets when each adset "targets" a different audience. My product is a kind of water bottle for fitness.
Adset 1: 4 ads with different pictures of people riding bikes and using it
Adset 2: 4 ads with different pictures of people running and using it
Adset 3: 4 ads with different pictures of people in the office using it
Adset 4: 4 ads with different pictures of people hiking and using it
*in each adset the copy for all 4 ads was the same
2. The broad campaign was non-CBO and the budget was $1K for each adset
3. After about $20K spend, I found the biking to be the best
4. I've duplicated the adset of the bike (stopped the rest), opened the new one at $1K/day budget and paused all 3 creatives beside the top performing one.
5. I've copied the best creative 3 times and only changed the copy this time. So now I had 1 campaign with 2 adsets, both were for biking. In one adset I was testing graphics and in one adset I was testing copy.
6. At this point, after about extra $10K spend (I barely increased the budget of it, just few % daily, as I started from $1K/day when my CPA goal is $25) I decided to move to CBO
7. Now I made the budget of the CBO around $2.3 (I was increasing the adsets budget a bit on a daily basis so it's not exactly $1K/adset).
8. Was working fine for quite some time, about 3 days.
9. Next, I've decided to increase the budget of the CBO. I wasn't going for the hard-scale of dupe&double but towards more traditional and increased it in just 10% to $2500.
10. After less than 12 hours everything collapsed. Went into roas under 1, negative ROI, started losing too much.
Please notice that it sucks to lose money but I knew it could happen so I've decided to give it a day or two.
After another $5K spend with same roas I figured that something went wrong technically.
Guys, please roast me and help me understand where I went wrong as things were going amazing at start.
Few speculations I'm having:
1. CBO scaling should be same as normal >> if something works, just duplicate and double the budget, if it works, close the old one and dupe the new one again till roas goes bad
2. If CBO under learning, would you raise the budget or let it go to ACTIVE first?
3. Would you use BID (yes, I forgot to mention that we're running on no bids) in CBO campaigns? We have other campaigns (other products) where we run $10K/day campaigns with no bid, so we don't have too much experience with that.
Thanks for all the help, proud to be part of STM forum!
Nice share!
I have been testing CBO last weeks more and more. I tested these things WITH LLA and PIXEL on a low CPL offer;
all devices/all placements - CBO
all devices/all placements - not CBO
mobile newsfeed only - CBO
mobile newsfeed only - not CBO (BEST COMBINATION!?)
For some reason the CBO are doing worse than "normal" setup. Also letting FB choose the placements result in higher CPL cost. At the moment I am only running the campaigns NO CBO - Mobile newsfeed only. I have tested around 15/20 different campaigns/setups, each with avg 3 ad sets, 3 images per ad set.
This all goes against tips many people are giving; choose all placements so FB can find the best spots, do CBO campaigns since FB will spend your budget smarter.
Maybe I am doing something wrong lol, but I find it remarkable honestly.
*btw, dont touch campaigns during "learning phase" I suggest.
So what we're doing now (last chance before we ditch the CBO - as it was a failure for us already in 2019 but I thought FB algo got better - apparently got wrost) is to run CBO with 4 adsets (all same, not testing any creatives, only copy - each adset 4 ads with different copy).
But this time we're using BID CAP, so we run each adset with different bid.
adset 1 - x BID
adset 2 - 1.5x BID
adset 3 - 2x BID
adset 4 - 2.5x BID
So far barely spend, but that would be the last chance I'm giving the CBO...
P.S - My personal thoughts about CBO: FB algo resets every time the budget goes over 15% up or down and they still can't solve this issue in scale. So the CBO keeps resetting the algo. Sure, you won't think FB that dumb to launch it that way but based on the results you can't ignore the question - are we giving FB too much credit? 
You need more creatives
Home > Paid Traffic Sources > Facebook & Instagram