**The original post from Dec. 2013 has been updated in April 2014 to reflect the inclusion of the new Ad Sets in Facebook Ads.
When advertising on Facebook, I typically just put one ad per campaign Ad Set.
“Why?”, you may ask? “It’s more work!”, you may shout.
Here’s exactly why.
(Click on image to make it larger.)
One of our Private Clients asked me to take a look at his Facebook account for the first time yesterday. I had lots of suggestions for him to improve his ad performance, but one thing really stood out as a major issue.
This ad performance report was such a perfect illustration of why it doesn’t make sense to put multiple ads into one campaign that I asked if I could share it with you.
Here’s the issue with placing more than one ad in an Ad Set:
Facebook does not allow you to evenly rotate ads. If you place more than one ad in an Ad Set, whichever ad gets the early clicks is the one that will get “whooshed” the most impressions. It doesn’t matter if it’s the best converting ad, or even the one that will necessarily get the best CTR.
The ad that gets the most early clicks is the one that will get the most impressions.
In this screenshot, you can see that one of the ads received more than double the impressions of the next most frequently shown ad.
Here’s the play by play for this campaign:
- The ads went live.
- The top 5 ads in the list received the first clicks (they are sorted by impressions received) .
- The top ad in the list got the most early clicks.
- Facebook “whooshed” that ad the lion’s share of the available impressions.
- The ads that didn’t get early clicks ended up not being displayed at all.
So… one of the ads got the most clicks. It’s the winner, right?!
Wrong.
It’s just the ad that got the early clicks. That’s it. You can’t make any scientifically valid decisions with that information.
This is why we don’t recommend placing more than one or maybe two ads per campaign Ad Set.
More than that is a waste of time & creatives.
See all of those ads at the bottom of that list? They were never even shown! They might be the best performers, but we’ll never know as long as they remain lumped into one campaign with the “early clicks” ad.
Hopefully at some point Facebook will allow us to do true split testing, although according to a Facebook Spokesperson we spoke with, it is not on their agenda at this point in time.
For now, separate out your ads into their own campaigns Ad Sets and do serialized testing. It’s not perfect, but it’s the way to go at this point in time.
This issue has baffled me for the past few months. Like you said, it makes for more work, and the results that are spewed forth when the campaigns finally roll out make split-testing impossible. Until the ad sets came out I was doing completely separate campaigns to split-test ads.
While I am glad to see that I am not the only one experiencing this frustration, it is ridiculous that Facebook offers the capability to upload up to 6 images for “split testing” and then not allowing the images to run their course.
It doesn’t make much sense, right, Olivier?! I’m assuming at some point they’ll figure this out so we can do true split testing. Thanks for your comment/commiseration.
what do you mean exactly when you say “serial testing”?
Split testing diverts traffic between a control and a test in real time. The notion of serial testing is to drive all traffic at one version for some time and then switch to some other version for some time. The issue there is “uncontrolled variables”, in particular, why should we expect that one month, week, day, or even hour is the same as some other? Say we run a test for the week of June 29 – July 5 and compare it to the week of July 6 – July 12. That’s obviously wrong and we would catch that, but what else about our market might we not catch? Serial is never good testing, but it is in some cases all we have. Split if you can, but test nonetheless.