Facebook Ad Algorithm Cracked: The Secrets Behind Successful Facebook Advertisers

You may be running Facebook ad campaigns, but are you getting the most out of every dollar you spend? Our research says you probably aren’t, and here’s why.

GOALS: I wanted to prove some basic relationships amongst the Facebook ad metrics, and see how else the data can guide us in creating more effective Facebook ads. It’s not only about the algorithm(s), but also how the metrics answer questions like “Does my prospect seeing an endorsement from one of their friends always improve my ad results?”

DATA: 172 ads across 10 client campaigns from August 1 – Sept 10, 2010 (roughly 40 recent days).

Question 1: Is CPC inversely proportional to CTR, as it is in AdWords?

Answer 1: Yes.

R2 correlation is 34%, almost a moderate correlation. [NOTE, we had ads with higher CTR than this, but I had to remove some of them because the outliers screwed up the charts such that we couldn't see the most basic and general relationships.]

Take-aways: Most of us are going for CPC’s lower than $0.50, even as low as $0.10 CPC. To get there, your ad CTR needs to be from 0.10% to 0.30%. Assume you start with a CPC of $0.70, and for every 0.10% CTR increase, your CPC drops $0.21. Lowering CTR is how to get more Facebook fans cheaply.

Question 2: How do Action Rate (% of those who see an ad who like your fan page) and CTR relate?

Answer 2: They are directly proportional. CTR is typically higher than AR because clicks include actions, and actions (liking) are about 50-70% of clicks. This will vary though depending on whether your ads are written more to get likes, or more to get clicks.

Strong R2 correlation at 82%.

Take aways: When you label your ads, put “like” or “click” in the name. Then you can divide out your data to see what’s really going on. Another rule of thumb is that 90% of facebook fan likes happen on the ad itself, not on the fan page. If you need to get conversions on your landing tab, you should be going for clicks, not likes, and adjust your ad copy accordingly.

Question 3: How do Social Impressions relate to Overall Impressions?

Answer 3: Social impressions (ad impressions where an endorsement from your friends can be found in the ad) are naturally less than overall impressions. Social impressions equal about 30% of overall impressions. However, if you target friends of page fans, the social impressions average 96.5%, which is odd… shouldn’t it be 100%?

Moderate correlation at 56%.

Question 4: Does an increase in Social Impression % mean a lower Cost Per Conversion? Does it mean a higher Action Rate?

Answer 4: It has no bearing on cost per conversion (negligible correlation).

Surprisingly, it also has no correlation to action rate.

Take aways: Don’t optimize for social impressions. One of our best cost-per-conversion ads had a social impression percentage of 16%. Don’t believe the hype that social endorsement on an ad translates into higher action rates. This finding is counter-intuitive, but when you argue with data, at best you’re Captain Kirk- at worst, you’re just a nutjob. I’d definitely love someone else to use their data to confirm or challenge this finding, though.

Question 5: How does your bid relate to actual CPC?

Answer: Low correlation, but you can guess what CPC will be. This is harder to figure because Facebook’s ad reports don’t put these in the same place. And I’d like to know the bid/CPC relationship for better vs worse cost per conversion ads, but can’t get that right now.

Take aways: Don’t look for low bid opportunities. This is similar to AdWords- many newbies freak out about raising their bid, or bidding high early on, even though the system’s algorithm requires that for success.

Anecdotally, I’ve found that going for the lowest CPC bids while creating ads leaves you with targeting that doesn’t get many impressions. It seems that since CTR is more the determinant of CPC, it’s better to go for targeting that makes sense and messaging that’s powerful for that group, and rely on CTR to lower your CPC. There seems to be something about the FB ad display algorithm that doesn’t favor microtargeting (and by this I mean a level of specificity 18-25 year old males who like women and are in college, interested in video games, located in Salt Lake City).

Brian Carter is CEO of the FanReach Facebook Marketing Courses, an AdWords consultant, and conduct social media training.

Related Stories
Mediabistro Course

Blogging: Analytics, SEO, and Content

Blogging: Analytics, SEO, and ContentWork with the former marketing director at Conde Nast Digital to improve your search rankings, integrate social, and increase traffic to your blog! Starting November 5, Jim Hopkinson will teach you how to analyze KPIs, monetize your blog through ads, sponsorships, and affiliates, and leverage your blog toward a larger platform such as publishing, speaking, or consulting. Register now!