00:48 AM

The Flaws In Facebook And ComScore's Social Ad Study

Facebook seems to have succeeded somewhat, in holding back widespread concerns about the effectiveness of its advertising platform with a recent paid-for study by comScore, in the wake of GM's announcement that its Facebook ad campaigns don't pay.

That's a damning announcement by GM and it's not something that any company does in such a public manner. Many seemed to dismiss this event as GM being clueless about how to run social ad campaigns. But that's a poor argument because GM, like all large corporations, knows very well what works and what doesn't, it has sophisticated processes in place to track and measure every dollar they spend. Car companies are among the largest advertisers, if people think they aren't on top of their massive ad spend, they are dead wrong.

Andreas Ramos, a digital marketing consultant, makes some interesting observations about the comScore/Facebook report, which published case studies on Starbucks and Target's Facebook ad campaigns that appeared to show traction for their ad campaigns. But a closer examination shows some glaring problems with the published data:

Important data is missing: ROI, CPL, CPA. ComScore does not state cost-per-lead or cost-per-acquisition.


The lift in control groups also increased. Why would this happen? The control should be flat. The document does not explain this.


If you carry out a massive campaign, there will be lift, even if it's 0.58%. After all, FB itself showed a billion ads to get 0.038% engagement for their privacy vote.

He shares some of his own data.

I have data for a $2.7 million 60-day campaign for a large nationwide bank. Ads were placed in Google, Bing, Yahoo, and Facebook. Over one billion ads were shown.

The CPLs: Google at $5.78;
Yahoo at $4.20;
Bing at $3.94;
and Facebook at $57.06.

There were 2.26X ads in FB over Google (383.7m vs 169.8m ads), but Google's CPLs were within the target CPL and thus profitable. Google's CPLs were 10% of the cost of FB CPLs. FB's CPLs were not profitable. If the money spent in FB had been spent in Google, it would have produced 19,700 more leads.

Foremski's Take:

OK, we don't have a control group here other than competing ad platforms, but these are interesting results and underline the type of data that companies such as GM are looking at on a minute by minute basis.

Looking at competing ad platforms is a far more important measure of Facebook's performance than control groups. Control groups don't compete for advertising dollars.

Facebook needs to figure out what does work. It's not making money in mobile, and on the desktop its ad performance is weak. It can still make money but can it make enough to support its share price?

This is a very serious issue for Facebook because there are a lot more companies that have suspended their Facebook ad campaigns without having issued a media alert as GM did. And there will be plenty more companies that will find reasons to delay their Facebook ad spend.

Facebook has been trying to manage the financial expectations of Wall Street analysts, it now needs to manage the expectations of its advertisers -- and help them with tools, shared knowledge, best practices, etc.