Home > Paid Traffic Sources > Facebook & Instagram

---

Would it work? CBO campaign to test for best pocket/pool within an interest/LLA. (19)

---

Dear Friends,

a question i have that i can't find answer for, would you recommend testing pockets(or some call it pools) within an interest?so an example would be, i want to find the best pocket/pool within an interest so instead of creating one adset with interest A, i created a campaign with 5 adsets all targeting interest A.I let CBO tell me which pool is the best and i subsequently off the bottom 4 adsets and scale the winning adset a see fit. does this work?

---

Within the same targeting, ad sets can behave way differently yes. So even duplicating exact same settings gives difference results, hence it's good to test it on multiple "pools" indeed.

---

I suggested that exact same approach here:

https://stmforum.com/forum/showthrea...aling-(Part-3)



Amy

---

Thank you.

some are saying CBO is not accurate unless it the daily budget on it is at least $200?

im thinking of doing a small test with $20 per day

do you think budget affects?

---

Nick Peroni suggests a budget of $100/day for CBOs. So you could start with that.

OR: I would suggest to take into account your offer payout when setting the budget. It just makes sense to me to set a budget of 2x payout per adset for example (so for 5 adsets that would be 10x payout for the CBO).

If you set the budget too low, you may not be giving each adset a chance to prove themselves by making conversions (assuming that's your conversion goal), and FB may be 'forced' to favor certain adsets based just on CTR alone.

That would be my speculation anyway, and the approach I take.

My concern for you is when you said "a small test". I want to ask: Are you running proven ads with proven audiences for your CBOs? Ideally you should use CBOs for scaling, not for testing ads and audiences. For testing it would be better to use ABO, so that every ad and audience receive sufficient budget and gets tested adequately. Testing audience pools using multiple adsets in this stage would be overkill IMO.

THEN, once you have a profitable ads+audience combination, break that out into its own CBO with the same 5 adsets and set a big enough budget for each adset to stand a chance at making conversions. This would be the scaling stage.

But that's just my approach - feel free to test your own!



Amy

---

got it.


could i ask you another question while you are here?

what does a profitable ad combo in look like in the beginning? and what defines a profitable ad combo?

I tested a creatives, followed by ad copy, followed by interest, and found a profitable adset ROI 2x about about 5 days, subsequently it died off with cpc hitting $8. and i off-ed it.

how do you know you found a winning adset when it happens?

---

I'm always here and please feel free to ask any questions - I don't have ALL the answers but there are other members with more FB experience than I do (e.g. @stickupkid) that can help!

I prefer to scale ad+audience combos that can do 50% ROI or above (preferably way higher but 50% would be the absolute minimum). Because with scaling the ROI would usually go down, so it wouldn't make sense to try to scale something that's just breaking even.

If an originally-profitable adset suddenly gets hit with high CPCs, I would try one or more of these things:

-See if I could ride it out by running it a couple more days - sometimes FB is either PMS'ing or testing new audience pools and just hitting a bad pool. And/or...

-Pause the camp and turn it back on for another test the next day or a few days later. Sometimes the camp is just "having a bad day". If performance doesn't pick up by noon I would pause and try another day.

-Duplicate the adset and hope it hits a better audience pool. And/or...

-Set up a new campaign from scratch, but using the same ad(s) and audience targeting (as the profitable combo). It's kind of like rebooting a computer. Doesn't always work but I had luck with it at least once.

-Create and test more ads for the same angle as the profitable ad(s).

-Check your frequency - audience may be saturated - although if the cpc increase is sudden, not very likely to be the case.

-Test higher and/or lower bids, different bid models.

-You can still take a chance and try to scale it anyway, using the duplicating the adset 5x in a new CBO approach, and see how that goes.

-If nothing you do is helping - I'm out of ideas. I'd suggest to just test more ads/angles with more audiences to find new profitable combos to scale.

Hope that helps!



Amy


Sent from my iPhone using STM Forums

---

Thank Amy.


may i check with you?

which post shud i look at regarding facebook ad placements?

i heard opposing views on instagram feed and facebook placements (other than facebook newsfeed)

some are saying stick to fb news feed only and avoid instagram because its cpm is too expensive

some are saying use all placements and instagrams is looking more promising

i am confused

---

Several choices:

1)If you want to keep things simple, test FB news feed first. That's where a lot of your traffic would come from anyway even if you were to target all placements on autobid. Prices may not be as cheap as if you were to target all placements, but it would be quicker and easier to create ads that look good for one placement, rather than having to use different image/video sizes for different placements in order to make the ad look great in all placements without distortion/warping. Please see this post:

https://stmforum.com/forum/showthrea...-Videos-Part-1

Then, once you have some profitable ads+audiences, you can always create ad versions with appropriately-sized images/videos for the other placements.

2)OR, you could target all placements to get somewhat cheaper prices, but then you'd need to create ads in different image/video sizes to make sure your ads will look good for all placements. For initial testing of ads and audiences I don't like to take this trouble, but you may feel differently.

3)OR, you can get the best of both worlds: Produce ads that look good for the news feed, and then also target other placements where the same ads would look good.

Again, once you have profitable ads+audiences, you can always test the other placements.


I would suggest not to overthink this. In the grand scheme of things, choosing what placements to test first won't make or break your campaign. Choosing what audiences to target, crafting your ads skillfully, running the right offer, knowing how to test and scale - these are the most important things IMO.


Amy



Sent from my iPhone using STM Forums

---

Thanks Amy. Really appreciate your advice.

do you do 1 on 1 consulting?

---

You're so welcome! Thanks for being an STM member!

I don't do consulting outside of the forum, but if you have further questions please feel free to ask them on the forum. I (and the other mods) will do my best to help.


Amy


Sent from my iPhone using STM Forums

---

Amy,

When it comes to CBO,

Do you use 1 day click or 7 days or 1 day click?

I did read somewhere that 1 day click does not work well for CBO budget below 400?

What are your views on this?

---

Hmm....to be honest I haven't given attribution window too much thought, or tested it as much as I should have.

I like to go with 1 day click because that would 1)enable FB to optimize the campaign faster, and also 2)optimize towards showing ads people that have the tendency to take action faster.

However, you need to ask yourself this question: For whatever it is you're promoting, how long does it usually take for a customer to convert after they click on the ad?

If most people convert the same day after clicking on the ad, go with 1 day click. But if a lot of following up / preselling over several days needs to be done before the typical customer would convert, then 7 days would be the better window to choose.

If you're not sure, you can either:

1)Start with 7 days, run for a while to collect conversions, then see what attribution window most of the conversions have taken place in. Then decide whether to switch to 1 day or not.



2)Run an A/B split-test, with AdsetA and AdsetB set to different attribution windows, to see which one will win out with statistically significant results.


If it's an affiliate offer you're running, you can also ask your AM to get an idea on how quickly people usually convert on the offer.

Hope that helps!



Amy

---

Thanks Amy

---

@stickupkid

Hey. I am curious what your views are on this too. Deeply appreciate your opinion

---

low key/impulse offers, like sweeps - > 1 day seems to be most relevant
high key/cpa offers, like ecomm - 7 days seems to be most relevant

CBO seems to work "only" with higher budgets, thats what I experience too.

---

Hey Amy,

in the post you shared here,

it says to duplicate to total 3 ad per adset for the CBO campaign,

am i right to assume 3 ads will open 3 pool instead of 3 ads targeting the same pool?

---

sorry i meant this link regarding my latest question

---

That's correct - at least based on my understanding.

Majority of the time FB will end up sending traffic only to one of those ads - the one FB deems to be the 'best' (even if all 3 ads are exactly the same).

However: I'm not so sure it's AS important to test audience pools anymore. I don't have solid proof, but I suspect that FB has been testing different audience pools automatically even when there's only one ad in the adset. Reason why I suspect this may be true is because even when I've "cut down" to having only a single ad running in an adset, performance would automatically improve for a while (I'm guessing FB may be "honing into" better and more targeted/suitable audience pools), and then once in a while the performance would drop all of a sudden in a very drastic way (I'm guessing FB may be testing new audience pools). Of course the changes in performance can be attributed to a slew of reasons other than audience pool testing. But I guess putting 3 of the same ads in an adset shouldn't make anything worse - something to test more when I get the chance.

Facebook has gotten better and better at identifying the best audiences. It used to be that we advertisers needed to do more testing and optimization in that area. In recent months/years the focus has shifted more towards keeping accounts alive (and getting new accounts), and having good creatives and a funnel that converts well.

And that is why I would recommend to test broad audiences if you're running anything that REMOTELY has broad appeal. Just target by age+gender and give FB some time to find the best audiences. Of course that doesn't always back out, especially if you have a niche offer. But good to keep in mind to test.



Amy

---

Home > Paid Traffic Sources > Facebook & Instagram