Facebook values relevance, but tells advertisers NOT to focus on the audience?
Balcom manages nine different Facebook pages for our clients, and we’ve dealt with all the algorithm changes as the social media giant strives to prioritize content users actually care about. Those changes make it tougher on us, but for the most part, they’re good – because ultimately, they push us to make better content.
But one of Facebook’s ad policies, however well-intentioned, seems to be having the opposite effect.
Over the last few months, our social media team has run into problems getting Facebook to approve seemingly innocuous ads. An example:
Lactose intolerant? Find low-lactose recipes and tips for living with lactose intolerance on our Pinterest.
Facebook rejected this ad on these grounds (emphasis added):
Your ad wasn't approved because it doesn't follow our Advertising Policies, which apply to an ad's content, its audience and the destination it links to. We don't allow ads that use profanity, or refer to the viewer's attributes (ex: race, ethnicity, age, sexual orientation, name) or harass viewers.
How to fix: We recommend focusing on your product or service, rather than the audience, and/or remove the profanity from your ad and/or destination it links to.
If you think your ad follows our Advertising Policies, you can appeal this disapproval.
We did appeal the disapproval. In fact, we went back and forth with the Facebook employee responsible for reviewing the ad, and it was still rejected.
Another ad rejected recently, even after appeal:
Are your young athletes tired? Sore? Can't pull ahead in the weight room? It may be what they're eating.
What was so offensive about these ads? They suggested the audience might be lactose intolerant, tired or sore. According to Facebook’s ad policy on personal attributes:
Ads must not contain content that asserts or implies personal attributes. This includes direct or indirect assertions or implications about a person’s race, ethnic origin, religion, beliefs, age, sexual orientation or practices, gender identity, disability, medical condition (including physical or mental health), financial status, membership in a trade union, criminal record, or name.
On the surface, this seems reasonable. If you met someone on the street, you wouldn’t say, “Hey, you’re a bankrupt/ex-con/[insert personal detail], here’s a product you might be interested in.” That would be presumptuous and rude. You might even inadvertently talk about some attribute that they’re not ready for everyone else to know about. And with the recent noise over alleged privacy violations with Cambridge Analytica, plus the issues Facebook has had to deal with regarding discriminatory targeting as it relates to the Fair Housing Act as well as racial slurs accidentally being offered as targeting criteria, Facebook naturally wants to err on the side of caution.
But by erring on the side of caution in this particular area, Facebook is inadvertently making ads worse.
Great marketing is all about the audience
Marketing starts with understanding the issues the audience is dealing with, then offering them solutions; not pushing them to buy stuff they don’t want. It starts with listening, empathizing, then meeting them where they are with a relatable message crafted in their language. Facebook’s admonition to “focus on the product or service, rather than the audience” makes that kind of empathetic marketing very difficult.
There’s a reason Facebook employs real humans to review these things; there are subtleties an algorithm can’t differentiate. I’m hoping that in the future Facebook can allow those reviewers to be a little more liberal in their judgments – particularly on less sensitive topics and when the ad takes the form of a question. Here’s why.
- The question format is more personal. It’s not just about pointing a message at someone, it’s about showing them that the product was created specifically for people like them. Instead of “Buy our stuff” it says, “Have this problem? Here’s a solution.” Also, the question format asks rather than implies. Saying “Are you X?” instead of “Hey X-person” serves to further target without being presumptuous: audiences can answer “no” and ignore the ad.
- Facebook is not “the street.” The on-the-street comparison I used earlier doesn’t really fit, because Facebook isn’t a stranger making assumptions based your outward appearance; Facebook knows what content you like and ignore, and what groups you identify with – because you voluntarily tell them. Also, it’s not really public: Unless you leave your Facebook logged in on a MacBook at the Apple Store, you’re likely to be the only person who sees which ads Facebook serves you.
- People already know they are being targeted. Part of Facebook’s objection to mentioning personal attributes seems not just to be because a person might feel embarrassed, but because they might feel creeped out that Facebook knows them so well. But you already know that the ads you see are served up based on your search history, content interactions and demographics. Is an ad that asks if you have a certain attribute any creepier than the ads that merely mention that attribute?
- The hushing only adds to the stigma. Ever heard one person hush another for talking about some attribute you have – an attribute you weren’t embarrassed about? Did it make you self-conscious? Did you question your confidence? If we make it taboo to talk to people about their common conditions, we’re turning something normal into something strange, alien and “other.” Hushing things up can make them seem shameful – even if they aren’t.
Three ways Facebook might fix the problem
Facebook’s appeals process and move toward more transparent policies are already helping. Recently, a similarly structured ad (“Feeling overwhelmed? You're not alone. Talk with one of our behavioral health experts.”) was approved after we appealed its rejection. But another ad on lactose intolerance was rejected around the same time. A couple of suggestions for the future:
- More liberal guidelines for ad reviewers – at least for less-sensitive topics that a person usually makes public anyway. People who are lactose intolerant, gluten-free or vegan are usually pretty open about it.
- An algorithm change that allows attribute-related ads to be targeted to people who have expressed affiliation with that attribute multiple times. If a person is open about being a Christian or vegan or union member by joining related groups and liking related content, isn’t it safe to say we can approach them on those attributes?
- More human review, less artificial intelligence. We need more human reviewers – ones who are native or fluent speakers of the language used in the ad, so they can understand context, tone and idioms.
Fortunately, Facebook has already made some strides toward improvement, including a commitment to hire 1,000 more ad reviewers in the coming year. Hopefully, they’ll continue to listen to feedback and find smart ways to keep both audience and advertisers happy.