Home > Paid Traffic Sources > Facebook & Instagram

FB ads: how to deal with this? (9)

03-14-2014 08:23 AM #1 kamikaza (Member)
FB ads: how to deal with this?

I've been testing few offers and found that they convert. For both of these offers i had 6 banners (12th march, banners split test). I took the best banners on 13th of march and made 4 campaigns for each offer, containing only the best banner but with different headlines + my old headline i have used in the previous day. I'be been running campaigns during the same time of the day on both days.
Here's my stats for the best banner of the day:

Offer #1:
13march
$0.34, freq 1.9 , ctr 0.068 , cpm 0.5
12march
$0.22 freq 2.45 ctr 0.111 , cpm 0.24

~$10 spent each day
Demo is the same, it's not holidays but after optimizations it get only worse. wtf?! i run the same best banner from 12march + new combinations, ctr must go only up...
I had the same campaign running for 2 days for another age group..let's see

13march
$0.34 freq 1.91 ctr 0.068% , cpm $0.21
12march
$0.25 freq 2,51 ctr 0.093% , cpm $0.23



Offer #2, Demo for this offer is not the same as for offer #1; ~$10 spent each day

13th
$0.46 freq 1.57, ctr 0.027, cpm 0.12
12th
$0.33 freq 2.24, ctr 0.063 , cpm 0.23

The same story. 2nd day performs worse then first..

How often do you deal with this kind of situations? How do you get around it? it's kinda hard to optimize anything when stats jump like that Any advice on this? thanks

03-14-2014 08:29 AM #2 dario (Member)

How big is the audience? BTW I suggest to let the banners run for more than $10 before deciding which is better

03-14-2014 08:51 AM #3 kamikaza (Member)

audience is big enough: 400k and 700k

03-14-2014 09:20 AM #4 zeno (Administrator)

How are you bidding? CPM? CPC? Ad performance varies day to day and can depend on how the ad 'sticks' at the days dawn. One day of data does not a comparison make...

03-14-2014 09:34 AM #5 kamikaza (Member)

i'm on OCPM. should i switch to cpm ?

03-14-2014 10:22 AM #6 zeno (Administrator)

No, stick with oCPM or CPC. With oCPM the CPM bid will vary day to day and during the day so it's not surprising that you see this kind of variation - you need more data to see how the CTR levels out over several days. However, one underlying problem could be your ad CTR being low in general... you really want sidebar ads to stay >0.100 CTR. You can be profitable on below that sure, but you want FB to want to deliver your ads, and you do that by making them engaging.

03-14-2014 08:46 PM #7 magnetic (Member)

Quote Originally Posted by kamikaza View Post
i'm on OCPM. should i switch to cpm ?
oCPM is a learning algorithm and it will choose the best target groups for you. Are you somehow tracking your goals e.g. via FB's tracking pixel? In addition, Facebook told us that the minimal recommended target group for oCPM is at least 500k users + you need to run the campaign for a few days with slightly bigger budget than a standard campaign to get full advantages of oCPM.

03-15-2014 01:23 PM #8 kamikaza (Member)

Ok i see, thank you for comments! Must be i am doing something wrong..

Can you guys share some info on how do you start with fb ads if you are on budget? how many days do you run campaign for an offer you want to test (first round of testing)? like you have 5-6 images, you don't know if the offer converts at all.. how much would you set budget for your campaign to , if let's say , offers payout is $2 ?
And if you will find the best converting banner.. the next round of testing (let's say it would be best headline) .. how much days will you run the next round of testing ? will the budget remain the same as for round 1 of testing? do you bid cpc or ocpm?

I'd appreciate any advice.

03-16-2014 12:44 AM #9 zeno (Administrator)

There are lots of ways to go about this.

An example that may work for you:

Campaign 1 (Budget $7.50) (Male, 18-21)(US)(Interest targets A)
-- Advert 1a (image 1)(ad copy1)
-- Advert 1b (image 1)(ad copy1)
-- Advert 1c (image 1)(ad copy1)
Campaign 2 (Budget $7.50) (Male, 18-21)(US)(Interest targets A)
-- Advert 2a (image 2)(ad copy1)
-- Advert 2b (image 2)(ad copy1)
-- Advert 2c (image 2)(ad copy1)

Run both campaigns simultaneously. For this example the ads are direct linked. As the ads start to get clicks, see which ads of a/b/c 'stick' and pause the ones that have a worse start. Get a days worth of data. Did you get any conversions? What ad CPC did you get? Were the ad CTRs crap? Say the ads converted at 20%, would their CPCs be far too high to work with a $2 offer?

This is where you have to decide if there is ANY potential to those specific campaigns. If you got no conversions - maybe the angle/targeting/offer sucks - but did you get enough clicks to say this? Were they reasonably spread out across the day? If the ad performance is horrible it will stifle almost any campaign. Choosing ad A over ad B often takes a fair amount of data unless one bombs completely, but whether or not a general angle/campaign has potential is much easier to see.

If it worked so-so, say -50% ROI, then run the ads another day. Roll out another $7.50 ad campaign each day and see if you can get that CTR up > lower CPC. After a few days you should have a decent idea of how well the demographic converts on that offer with you specific angle/targeting. Is there any potential there?

If not, try a different angle and targeting. Hit a higher age bracket. They are usually more expensive but many convert 2x as much. Change the angle completely and target to suit.

There really is no magic bullet for low budget testing - there are far too many variables that influence campaign performance. What you do need to do is sort out a campaign structure (such as that above) and get some data, then think carefully about what that data means and how you should move forward. Only testing yourself and getting experience will help you gain confidence there, not anything anyone can tell you on a forum.

Home > Paid Traffic Sources > Facebook & Instagram