Tuesday, October 16, 2012

Optimizing Facebook Campaigns For Performance

Image representing Facebook as depicted in Cru...

To get the most out of your Facebook campaigns, you need to devote time to evaluating their performance — not just after they’ve finished running, but while they are still active.
You aren’t best served by creating ads, running them for several weeks, and then checking them after they have finished spending. You will see much better results if you’re willing to optimize on a regular basis during the lifetime of the campaign.
Two Ways To Optimize 
There are two key ways to do so. First, you can spend more against what is working, and less against what is not. This concept seems obvious, but how do we determine what is working and what is not? Use the metrics and values that you’ve set for each campaign as a benchmark to evaluate them. For our sample CPF campaign, we’ve determined that our main metric is cost-per-fan, and our goal cost is $1.25 per fan.
We can then look at the data we’ve generated and find campaigns or ads that fit those parameters. We can choose to bid up individual best-performing ads so that they’ll serve more in Facebook’s auction environment, or we can increase the campaign budgets so that campaigns will spend more.
The other means of optimization is creating new ads based on what we’ve discovered is working. That means looking at the data on a deeper level and evaluating which individual elements are the most successful in terms of your KPIs. Start by focusing on each element separately. With body text, are there visible trends? Is it apparent that ads using the “Direct Like” text category have CPF much lower than any other?
Do the same thing for images, headlines (if applicable), and any targeting parameters. Hopefully, you’ve set your campaigns up in such a way as to make this data collection easy. Isolating one variable per ad is the best way to see what drives performance.
After understanding performance of individual elements, combine the best image, body text, headline, gender, age, and interest targets, etc., together to create an optimized ad set. We assume that if we combine all the best elements into a single ad, that ad will perform better than the elements did on their own.


No comments: