Facebook Targeting: How to Exclude for Better Performance
Published: May 30, 2014
Author: Zach Greenberger
Great performance with Facebook marketing is all about targeting. If you’re not speaking to the people who matter to your brand, your message will fall on deaf ears. But sometimes telling your story means knowing when not to tell it.
Many ad networks offer impression and frequency capping, letting advertisers set how often each user sees their message. For example, a site that gets a lot of impressions might let an advertiser buy five impressions per day, per user. Once a user saw the advertisement five times, they wouldn’t see that ad again until the next day.
Up until recently, Facebook didn’t offer anything like that. They’ve started testing a new Reach & Frequency Campaign, but that’s a tool for enterprise media buyers looking to reach a very specific audience. Most advertisers will leave it up to Facebook to determine who sees their ads, and how often. Sometimes this results in negative comments about users seeing one advertiser’s posts all the time. In reality, Facebook only serves paid posts to users once per day, per advertiser. But the power and impact of News Feed ads can lead to user fatigue with an advertiser’s message.
So how can advertisers take some control over their messaging, without reducing their reach?
One solution can be found using Facebook’s new Website Custom Audiences. While traditional custom audiences are static, built from data files uploaded manually, Website Custom Audiences change dynamically based on user behavior. As users visit different pages on an advertiser’s website, they can be moved into, or taken out of, different audiences. Further, each Website Custom Audience has its own Retention Window. Defined in days, the Retention Window lets advertisers set how long each user is remembered by the audience. For example, a Website Custom Audience with a Retention Window of seven days will forget any audience members who were added eight days ago.
Facebook proposed a great use-case for this functionality at a recent training event. Let’s say your e-commerce site sells something that people want to buy, again and again, but not every day. Maybe something like Gummy Bears. Of course once someone’s had some Gummy Bears, they’re eventually going to want more. So a smart advertiser will remarket to past purchasers. But if someone has just purchased a 10lb bag of Gummy Bears, they’re not going to be ready to re-order for at least a few days.
To avoid bombarding them with ads that have a low chance of resulting in a sale, Facebook recommends creating a Website Custom Audience of past purchasers with a Retention Window of a week or two. By excluding that audience from all targets, an advertiser can be sure they’re reaching fresh users, hungry for more Gummy Bears. A creative advertiser could target a different campaign to those recent customers, helping them better enjoy their purchase. For example, they could promote a series of posts featuring innovative Gummy Bear recipes, or little-known facts about Gummy Bears.
How can we adapt this strategy that limits over-exposure to past purchasers so we avoid excessive frequency to all audiences? It’s as simple as changing our strategy a little.
Instead of creating audiences with short Retention Windows just for purchasers, let’s create them for all site visitors. But let’s also reduce our Retention Window down to just a couple days. With a window of one or two days, we’re limiting our ads from appearing to users who’ve recently visited our site. This eliminates the “stalker” factor we sometimes hear about when users are retargeted with a company’s ad immediately after visiting their website.
But how does this affect performance? We’re excluding people from our campaigns who we know are interested in our products. That can’t be good, can it? The only way to find out is to test!
For one of adMixt’s clients, we ran identical audiences, one where we’re excluding visitors to the site from the last 24 hours, and one where we’re not. Otherwise the targeting and the creative are identical. Here’s how it performed.
This is a small spend of a few hundred dollars, and it hasn’t been running long, but the results are amazing: Adding the exclusion gives us a 19% increase in CTR, giving us a 13% reduction in CPC and a corresponding 14% drop in CPA. Based on these results, you can be sure we’re testing this across more clients with different-retention windows. Your mileage may vary. This client has a high level of site traffic, so excluding frequent visitors has a big impact. We’re guaranteeing their ad campaigns are only served to users who have not yet clicked recently.
Let’s take this one step further and think about how we can use Website Custom Audience to get even more sophisticated with our messaging. If we create an audience with retention windows that get larger and larger, and include and exclude them strategically, we can “storyboard” out a detailed communication plan.
Include: Visitors in the last 180 days
Exclude: Visitors in the last 30 days
Include: Visitors in the last 30 days
Exclude: Visitors in the last 21 days
Include: Visitors in the last 21 days
Exclude: Visitors in the last 14 days
Include: Visitors in the last 14 days
Exclude: Visitors in the last 7 days
Include: Visitors in the last 7 days
Exclude: Visitors in the last 2 days
By featuring different creative in each of these campaigns, we insure our audience sees fresh creative every week, until they click through to the site and the cycle starts over. The great thing about this type of schedule is that you don’t have to constantly manage your ad rotation. Each user ends up on their own schedule based on when they last interacted with your site. It’s a simplified automated marketing campaign that can easily be customized to exclude users who convert, for one or two weeks, or forever.
Special thanks to Or Flakov and GoChime for inspiring this post. GoChime offers realtime management of static Custom Audiences by integrating with email service providers like MailChimp. With their service, you can build similar dynamic audiences and include and exclude users from your targeting based on their email newsletter subscription status, or email open rate status.