Facebook Advertising Test Results
Facebook has been receiving a lot of criticism lately for their lack of a cohesive mobile advertising strategy. I’ve been a voice in that chorus. Before their IPO, they lost General Motors – one of their largest advertisers and arguably their largest automotive client. But at least 50% of their traffic comes from desktop computer users, so for that and a variety of other reasons, I continued the modest Facebook advertising campaigns Moss Bros. Auto Group was running each month.
Metrics Do Not Measure Up
The results of those campaigns never seemed to add up. We had tried cost-per-click (CPC) campaigns, cost-per-thousand (CPM) campaigns, various geographies, and used broad as well as specific category/keyword targeting. Results were usually disappointing regardless of our parameters, generating click-through rates of less than 0.1%. Bust most importantly, Facebook’s metrics never matched the traffic we received on our website according to Dealer.com (our website provider) or Google Analytics. Here are figures from one particular campaign:
Time frame: 8/29/12 to 9/3/12
Facebook click claim: 85
Dealer.com inbound traffic count from Facebook: 30
Google Analytics inbound traffic count from Facebook: 24
Even the best case comparison has a massive discrepancy of 65%! Keep in mind, I was paying an average of $1.18 per click. That means I could have seen more than double my actual traffic if Facebook wasn’t misrepresenting its figures. It was that discrepancy -that misrepresentation- that lead me to run a test.
I had been wanting to explore the benefits of Facebook advertising for my personal blog. Could it help me capture a new audience? Could it garner some additional attention to this new online endeavor I had been working so hard to grow? This seemed like an ideal moment to test it while simultaneously learning for my employer.
Test Parameters
The setup was simple enough and I hoped to learn a few things.
- Determine how Facebook was obtaining their figures
- Did Facebook prioritize CPC campaigns over CPM – thereby earning more revenue for themselves with less “work” (server time, clutter for users, etc.)
- Facebook’s effectiveness at growing a blog’s audience
Here’s how I setup the test:
- Create identical campaigns: graphic, headline, copy, geography, category targeting, date range, and budget
- One variable: payment metric. One campaign was CPC, the other CPM
- Analyze traffic data according to Facebook and Google Analytics.
I must humbly admit that the campaigns were failures at generating new repeat traffic for my blog. You’ll see in the screen captures below that I generated traffic and maxed out my budgets on both campaigns, but they weren’t engaged readers. Bounce rate was high and traffic has remained flat since the campaigns. Sure, that could be my fault. After all – its my blog. But let’s look at the campaigns themselves.
Facebook Advertising Test Results
The image to the left (click to expand) shows the results of the campaigns as well as their setup.
Obviously, Facebook did a great job prioritizing the CPC campaign over the CPM. They showed it to fewer people, fewer times, and reached the budget cap. In short – more profit for less work.
And that’s my issue. The ads were identical and so was the targeting. The results should have been the same. The ONLY thing Facebook could have done to skew the results was misrepresent the actual delivery of the ads. There is no way they were shown to the same audience, despite my instruction to do exactly that.
And that is something I want to be very clear about. I had a contract with Facebook. Specific terms were agreed upon. Furthermore, the Facebook Statement of Rights and Responsibilities and Advertising Guidelines don’t disclose that they may adjust your targeting based on payment method. In today’s digital world, advertisers have become conditioned to simply accept the “secret sauce”. We take for granted that we don’t know exactly how our results are generated, but we expect them nonetheless. This is a huge mistake for our industry and will only weaken our ability to accurately measure ROI, select the best media for our products, and hold that media accountable for a mutually beneficial relationship.
Facebook’s Response
Since this test, I have stopped all advertising on Facebook. The ads they run on the right side of the site are obviously rigged for Facebook’s benefit. That said, I have a second test planned: sponsored stories. These ads appear on desktop and mobile devices so I think its worth another foray.
I really wanted an answer from Facebook and two weeks ago I got it. A fortuitous event occurred when I attended Digital Dealer 13 and one of the keynote speakers was from…you guessed it…Facebook! He was a Client Partner for the automotive industry. In addition to the keynote speaker, another Client Partner was hosting a breakout session about Facebook advertising. After the session, I introduced myself and presented my problem. He was a very nice guy and honest in his response. It came down to two things: “priority” and “assumption”.
You see, Facebook assumes your intention by how you pay for your ads. If you want clicks, they assume you will select CPC billing. If you want to brand something, they assume you will select CPM billing. They prioritize the delivery of CPC ads because the goal is to generate clicks and they aim to provide them. I believe this is a faulty assumption. Logically, because all of their ads must include a link to something (Facebook Page, website, whatever), they should assume ALL campaigns want clicks. If I just wanted to “brand something”, I wouldn’t include a link. My ad would consist of a pretty picture and impactful text. Done.
Toward the end of his explanation, his colleague (who would later deliver the keynote) joined the conversation and concurred with the explanation I just received. Both said that Facebook is constantly tweaking its ad platform and new products will soon be introduced that I may want to look at in addition to the current “True View” system they have for measuring ad metrics. I intend to look into them all.
Conclusion
Facebook is gaming the system. I knew as much after the test. I haven’t advertised since. But is that a good idea? Along with Google, Facebook is an 800lb. gorilla. They have 1 billion users worldwide. Is that an audience that -as General Motors did- I can simply create a Page for and hope they find it? The alternative is that I now know how the Facebook advertising system works and I can probably work more effectively within it.
My jury is still out on whether or not I will use Facebook advertising again for a campaign. Thanksgiving is coming up and we always do a substantial amount of advertising for the holiday season. Facebook may not be included. Either way, I promise to keep you posted.
Your Experience
What has been your experience with Facebook advertising? Will you reconsider running ads there? Do you think I’m nuts and the test proves nothing? Okay. Tell me why in the comments below. I respond to all.