Facebook Ads Don’t Work. Here’s Proof
In October of 2007 I decided to test Facebook Ads to see if the hype was substantiated. The prospect of being able to advertise to a large group of people by leveraging specific social profile data seemed like a gold mine.
Based on my experiences in the past few months I can say that Google has nothing to worry about. Facebook Ads are not even close to being an Adwords killer.
Let me take you through an example comparing Facebook Ads and Google Adwords for something that was inspired by Sugarrae – a microsite advertising Stephen Colbert’s new book I Am America and So Can You.
I setup a brand new site to showcase Colbert’s book at colbertbook.com – a few informative pages with affiliate links to the hardcover and the audiobook. I tracked traffic and outbound clicks via Google Analytics.
I setup the following Facebook Ads…
and setup the following targeting parameters:
You are targeting people between 18 and 65 years old in the United States who like Colbert, Colbert Report, or Daily Show.
I targeted all ages to get the widest audience, but these people had to have listed Colbert Report, Colbert, or the Daily Show in their profiles – a seemingly laser-targeted demographic.
How did Facebook Ads perform? Here are the stats:
- impressions: 225,875
- clicks: 178
- CTR: 0.08 %
- average CPC (cost per click): $0.22
- average CPM: $0.17
- total cost: $38.38
Here’s a Google spreadsheet showing you the complete stats for the 2 Facebook ads from Nov 7th to Jan 5th.
I set the max CPC between $0.11 and $0.31 and varied it at different times. Observing how max CPC affected impressions was very odd – it seemed to fluctuate wildly. I’d set it to 11 cents – it would get a ton of impressions, then drop to nothing. I would raise the bid just to jump start it again to get more impressions. It seemed like there was no consistency to when Facebook decided to run your ads and at what prices.
As you can see the CTR was abysmal. That is probably the most telling part.
For comparison I also setup Google Adwords for this site with the following ad
and bid on keywords related to the book. Here are the Adwords stats for Nov 6th to Nov 27th:
- impressions: 15,386
- clicks: 688
- CTR: 4.47 %
- total cost: $109.63
As you can see the numbers are quite different, (this was for a shorter period of time).
Now let’s take a look at the stats provided by Google analytics.
Visitors coming from Facebook:
- pages per visit: 1.22
- Bounce rate: 82.84%
- average time on site: 13 sec
- 3.92 % clicked to Amazon
- 7.84 % clicked to Audible
- total conversion rate (clicked on product links): 11.76 %
Visitors coming from Google:
- pages per visit: 1.61
- Bounce rate: 67.21%
- average time on site: 42 sec
- 12.31 % clicked to Amazon
- 9.94 % clicked to Audible
- total conversion rate (clicked on product links): 22.26 %
Do keep in mind that Google Adwords ads were run on “I Am America” book related keywords and not just “Colbert Show” related keywords, but that is at the crux of this whole thing. Google Adwords allows you to target specific intent at the time of a Google search. There’s no such thing with Facebook ads.
Comparing every single metric, Google Adwords visitors were far more engaged, far more valuable, and far better targeted. The traffic was cheaper and more consistent. The volume was far higher.
It’s not even close. Facebook Ads just don’t work.
And for all the hype that it’s gotten, it’s not even in the same ballpark as Google’s Adwords. Not even the same league. While targeting users by their listed interests may sound promising, it’s just not very effective.
P.S. For another example check out John Mendez’s Facebook test from a while back – similar results. There’s also shady who talks about how restrictive the ad process is. If you want to hear a real expert’s option, check out Danny Sullivan’s take on Facebook Ads (video) – he really knows what he’s talking about.
Elizabeth said,
February 6, 2008 @ 3:06 pm
Wow. Thanks for posting this, I have long been curious if Facebook was all hype or not. A colleague of mine had very similar results to yours, but I wanted to see more than one example before drawing too many conclusions.