Today’s post is by Steve Palombo, Sr. Account Executive at Triggit.
I often read and hear skepticism from industry pundits, conference attendees, and even potential clients regarding the viability of the format of Facebook Exchange, or the Marketplace ad. I’m asked, “What can Triggit possibly do within the confines of the Facebook ad unit? And how do you do it in 90 characters or less? Who reads that stuff, anyway?”
At first glance, the image space looks tiny, and the text limitation is like trying haikus for anyone writing Facebook ad copy. In fact, if this paragraph was the body copy of an ad on FBX, we would have already tripled the character limit! However with a little outside-of-the-box thinking and adept manipulation of first-party data, the traditional Facebook ad can become an incredibly versatile and potent ad unit.
With our initial clients in FBX, we immediately saw that users were drawn to dynamic images of products they’d just clicked away from, even if the image is confined to the seemingly microscopic 99×72 ad and description. From our initial data, we found that moving to dynamic from static creative with product name and price had massive increases in click performance of over 500%.
Although the performance was already excellent for dynamic ads, we decided that it was time to do the next logical thing: A/B test dynamic ad copy to the ground. But with 90 characters, how creative can you actually get?
Dynamic ads — the copy test
We had an international client that was a perfect match for this test. They had thousands of product items to dynamically retarget site visitors with from all over the world, and had lots of creative parameters to play with including product name, price, category, and description. After quickly uploading their entire product catalogue into our system, we started off with a baseline model and generated our traditional dynamic product image with a title advertising a big “20% off” deal in the title line with sparse body copy that included price and the product name.
We then pitted that ad unit against an identical one, with the exception that the body copy was replaced with about 80 characters of copy related to the product image, followed by an ellipsis and a call to “…learn more!”
The results defied our expectations. Conventional display advertising wisdom says that less is more – tight, price-sensitive copy and flashy images and calls to action were thought to draw more people’s attention. However, after letting the test run to five million impressions, we found that the text-heavy, product-related ad copy without prices was converting much better. Click-through rates were about 25% better and click to conversion rates were 50% higher, despite the fact that we were paying identical effective CPM costs for both ad groups.
Although we had clearly found a successful winner in this A/B test and proved that users can be moved by more than just price in certain verticals, we also could take a bigger logical leap: people aren’t just using the social network to look at vacation pictures or gloat about football victories; they’re also using the social network when they’re in the mood to buy, even taking the time to read the longer, more informative ad copy. Tests like these are starting to provide more light into why we’re seeing higher than expected conversion rates on Facebook Exchange compared to traditional display channels.
This is a huge win for advertisers, especially for e-tailers working with FBX partners like Triggit who are able to test and optimize different dynamic creatives. Finding a way to get 50% more conversions without doing any extra technical work is a huge windfall we’re happy to put to our clients’ advantage.
– Steve Palombo is a Boston native who graduated summa cum laude from Wake Forest, is known as Triggit’s in-house data scientist, and has spent countless hours testing data sets while managing millions of dollars in spend for several dozen advertisers.