Many advertisers have been unhappy with Facebook over its rulethat an ad image must contain no more than 20 percent text. Inside Facebook recently sat down with a member of Facebook’s policy team to gain a better understanding of the company’s controversial guidelines.
Facebook told the blog that the company uses both manual and automatic review procedures to see what kinds of ads make the cut. All image ads go through Facebook’s scanning tool, and then a select few are measured by humans. As Inside Facebook’s Brittany Darwell notes, this is why it seems like the rule is inconsistent.
Text on the actual product being advertise doesn’t count toward the 20 percent text rule, but the computer systems might not be able to tell the difference. Unless this kind of ad was among the hand-reviewed ones, the computer might automatically reject it.
To help advertisers figure out whether or not their images are legal, Facebook said it is developing an external version of the grid-based text detector it uses. Facebook started testing this on Friday, and will likely be rolled out to Power Editor users eventually.
Inside Facebook also included examples of what’s OK by Facebook and what would get rejected: