Facebook Social Ads Experiments - Campaign Quality Score?

One of my favorite blogs to read is Michael Martinez’s Seo Theory. Every once in a while, Michael will post a list of different experiments to try to help the beginner/immediate SEO hone their skills and understand different aspects of optimization. Recently, I’ve been experimenting with different campaigns and ideas on some of my auxiliary facebook accounts to try and squeeze some more water out of the facebook well. I’m going to be posting some of these experiments that I have tried over the next couple of days as well as what they did/did not conclude.

Experiment 1. Keeping a Clean House - Facebook Campaign Quality Score?
Adwords is notorious for looking at your account and campaign history when deciding your quality scores and the impressions/cpcs that you will get. If you have a ton of bad performing campaigns and ads in your account, other campaigns that you create in the future may suffer in the form of higher bids and lower quality scores. So I decided to see if Facebook looks at this info with social ads. The goal of this experiment was to see whether or not campaign structure/quality affected our bid prices/impressions. A good assumption would be that a campaign with a lot of ‘dead’ ads (ones with yellow exclamation marks that haven’t been served in a few days) would have a worse quality score than one with no ‘dead’ ads, assuming of course that all other variables were held constant and that facebook did actually implement some type of quality score (the unknown in this experiment).

For this experiment I set up 3 campaigns based off another campaign I was already running so I would know the type of numbers to look for. This original campaign was pretty gross:

I would use this campaign as the basis for the 3 test campaigns. The first campaign would have all of the same ads from the original campaign, including the ones that have proved to suck. The second campaign would have all the ads from the original campaign, but after the ones that suck died out, Id delete them. The third campaign would have only the ads from the original campaign that had a ctr over 0.15%. I needed to launch these campaigns all at a time when they’d be accepted together to avoid dirtying up the statistics due to time of day (different ads perform differently throughout the day). After letting the campaigns run for a few days and stabilize, the totals for impressions, cpc and cpm should let us determine whether some kind of a campaign quality score exists. Assuming our good ads performed roughly the same across all three trials, the ad with only good ads should receive preferential treatment from Facebook if a campaign quality score exists.

Results: Campaign 3 had absolutely no advantage to it over the other two, other than allowing the page to load quicker (when you 100 ads in an adgroup it will take a few seconds for the page to load). After 3 days, the cpcs were roughly the same, averaging out at around ~30cents, with the impressions all remaining consistent as well. On my end, the conversions were all consistent, although this has little to do with the 3 campaigns used since all of them ended up serving the same top-performing ads over 90% of the time. This doesn’t conclude the absence of campaign quality score, only that in my experiment, campaigns with bad ads didn’t perform any differently than campaigns without bad ads. There are other factors that need to be tested including average ctrs across a campaign (does a campaign with 2 0.50ctr ads perform better than one with 0.75 and 0.25 ctr ads or vice versa), banned ads in a campaign (do banned ads in a campaign affect the good ads?), etc. Additionally, it would be interesting to run this same type of test across multiple accounts to test for account quality score (do accounts with lots of bad campaigns perform worse than accounts with no bad campaigns).

So, from this quick experiment we can conclude that having a campaign with hundreds of shitty ads in it does in no way affect the overall performance of the rest of your good ads. So spam away

(PS: sorry I couldn’t provide screenshots for the final results … the first shot was taken about a week ago and I just recently reformated so I don’t have a copy of photoshop installed).