Episode 6: Finding the Right Audiences for Your Facebook Ad Campaigns
Without the right audiences, your Facebook ad campaigns are headed for trouble. In this episode, we talk about the various targeting options available.
00:01 Stewart Anderson: Welcome back to the, Spend $10K a Day podcast. I'm Stewart Anderson, the Chief Marketing Officer here at MuteSix.
00:09 Steve Weiss: Steve Weiss. I'm the CEO. Welcome back. I'm gonna talk about Facebook ads. Another area of Facebook we'll be exploring today, this podcast is totally dedicated to audience testing.
00:20 SA: Yep.
00:21 SW: Finding the right audience for your product, understanding how to come up with messaging for the top of the funnel, what the top of the funnel means, what the mid-funnel means, and what the bottom-of-funnel means. Setting up your messaging around specific audience segments that you're gonna be marketing to.
00:36 SA: Yeah. Very, very important. Honestly, obviously a lot of attention gets paid to the creative on the advertising side, but if you don't have the right audiences, and I mean the right audiences, your ad campaigns are doomed to fail.
00:51 SW: Before we go into it, as you guys probably know, you guys probably are very... Some of you might be experienced Facebook marketers, others are rather brand new. There's quick information on the type of audiences on Facebook. You have your precise interest audiences, which is I'm targeting somebody who's liked a page. This is the old school way of marketing on Facebook. You have your WCA audiences, which are your website custom audiences, which is I visited your website and I'm gonna get re-marketed with an ad. Or I've just visited your website, and I'm gonna be a lookalike of this audience.
01:28 SW: But WCA is Website Custom Audience. You have your lookalike audiences, which is Facebook creating a set of audience that's very similar to a specific audience that already exists, aka lookalike. Then you have your email audiences, which are email lists that you upload into Facebook, and building audiences off those. So those are the four core audience types. There's other audiences which are very, what we call "open audiences" of people who just live in Beverly Hills, or people who just live in San Francisco. But for more intense purposes, those are the four main audiences that you're gonna be targeting on Facebook.
02:12 SA: Yeah. Absolutely. So Steve, talk to me a little bit more about how you like to first approach building audiences for a new campaign. If you're a completely new client, or you're starting a new project, what are your first things that you're gonna build an audience on?
02:28 SW: If the campaign has no history but they have a lot of website traffic, the first thing I like to do is I like to build a lookalike off the people who visit your website, but also the people who visit your shopping cart. I like to build a lookalike off people who visit your shopping cart but don't convert. Then I also like to build a lookalike around people who purchase your product. That goes along with building a website custom audience around people who visit the shopping cart, people who purchased, and people who've just been to the website.
03:00 SA: Yeah. Really, really important ones, and obviously some of the testing you do between those gets... When you're really sophisticated, you start seeing tweaks there, you might not expect. But another good source that I've found, and Facebook has some good tools in there, and there's some other good tools out there for syncing audiences, is email lists. Depending on what email provider, I think they have a direct connection with MailChimp for sending custom audience data over, but you can also use your email list as a custom audience source as well.
03:31 SW: Remember, for Facebook, to connect a user to Facebook, they have to be logged in to Facebook at the time of making a purchase. Not everyone is logged into Facebook when making a purchase. Most are, but what you really wanna do is you wanna upload the email list along with building a website custom audience, just in case that person is not logged into Facebook.
03:54 SA: Definitely.
03:54 SW: So three pillars of our audience segmentation strategy is separating our audiences into different buckets. There's the top of the funnel, there's the mid-funnel, and there's the bottom-of-funnel. Each audience that we're marketing to falls into one of these funnels. And the funnels dictates the type of ad units for those audiences. Usually when you're doing top-of-funnel, aka prospecting, you wanna use a lookalike audience and you wanna exclude all your current users, or people who visit your website, people who've purchased. You wanna exclude those users 'cause you want them to get a custom ad 'cause you wanna bring them into the funnel.
04:39 SA: Yeah. And I think one of the things I usually tell people when they ask me about prospecting, about building this top-of-funnel audience is, I think a really good takeaway is the lookalike audiences you use for your prospecting, the sources of those should be your actual audiences from the bottom-of-funnel stuff. So if you're targeting people, for example, who have purchased and you're trying to upsell, or they have added to their cart but they haven't bought, those are perfect bottom-funnel audiences, and you should be building lookalikes off of those, and then using the lookalikes top-of-funnel. That's a really great way to get started with lookalikes.
05:12 SW: Yep. And also you could build lookalikes based on percentages. I'm sure you saw this in Facebook where you could build lookalikes 1% all the way up to lookalikes 10%. So what lookalikes 1% means, it's the most exact people to your audience. So lookalikes 1% is Facebook saying that, "I'm gonna build a lookalike to the best percentage of your audience." Then if you want a bigger audience, they have lookalikes 10%, which is the farthest... Not the most exact, but still close to who your audience is; aka user point, I upload an email list of 10,000 people, I can build a lookalike 1%, which is Facebook saying, "The most exact we have of who looks like these audiences, then 10%, not the most targeted list of people that look like this. So what you always want to do when you build lookalikes, is you also want to exclude your lookalikes 1% from your lookalikes 10%, 'cause you don't want to be targeting the same people.
06:11 SA: Yeah. That's a really, really good point. A lot of people forget that when they're building lookalikes. A lot of people think that when you're building lookalikes, you're like, "Alright, here I'm done. That's good. Created a lookalike audience off of this other audience. I'm ready to go." You have to remember to make sure there's not overlap in your audiences, because lookalikes are big. Especially when you're doing a 10% one. A lot of people.
06:29 SW: Exactly. Our segmentation strategy, high level, is splitting people off into different funnels, and then targeting users based on where they are in the funnels. Obviously, people will get moved between different audiences. So we target... Prospecting, we exclude all of our WCAs, all of our email lists. Mid-funnel, you might wanna exclude just the buyers. Just people who bought the product. You don't wanna exclude people who visit the website, because they haven't bought product already.
06:57 SA: Yeah. Marketers have different opinions about what the right size is for creating lookalikes. The source audience. Some people say it's as low as a few 100. Some people say it's 1,000. Some people say it's 5,000. Steve, what do you think? What is the best number to say, "Hey, this is probably a big enough audience to build a good lookalike off of?"
07:21 SW: For e-mail, you need at least 10,000. For website... For people visiting your website, you need... Or for transactions, you need at least 1,000. You want sales of a specific product. For website, you need at least 20,000 users that Facebook is cookie-ing. Anything else is just not statistically relevant. I like to say that's the early... Early on the smaller amounts, but you could build it off of anything really. I'd say just starting there is probably the best point. Just start off with... Try and get 1,000 people purchasing, try to get 10,000 email list or 20,000 people on your website.
08:01 SA: Definitely, definitely. I think we've gotten a pretty good read on lookalike audiences. One thing I've found as well and Steve you've probably used this a lot in your campaigns. When you're testing lookalike audiences, especially when you're spending at scale, you spend a lot of money, and one little squeak here or there can make a big difference. Sometimes what you would expect to be a really effective audience, a lookalike based off of purchasers, sometimes you're not getting the right balance of volume and CPA. I don't know about you guys out there, but sometimes you might want to try testing that against a lookalike audience based off of people who've added to cart. Sometimes for whatever reason, it could be Facebook's algorithm, it could be whose composing that source audience. But sometimes I actually get better results using a lookalike off of add to cart versus purchases.
08:52 SW: So remember one thing, Facebook is never gonna to tell you how they build lookalike audiences and what variables. We don't know. One of the biggest factors of serving ads, and this is the reason why Facebook is so difficult to optimize sometimes, because a lot of your success is based upon people being on Facebook to see an ad. So even if both the add to cart audience, and regular lookalike are very similar, you're gonna see different conversion results because some people are online and other people aren't.
09:20 SA: Yeah. I think the rule of thumb here is as always with advertising, it's test, analyze and optimize. See what's working, test a bunch of different things out, figure out what the data is and then just continuously improve.
09:34 SW: Exactly. We talked a little bit about lookalikes. We talked a little about about setting up your funnel. Mid-funnel is targeting users. Just how we do audience testing. Top-of-funnel, mid-funnel, bottom-of-funnel. We target mostly users who haven't purchased with a unique piece of content. Mid-funnel. Then bottom-of-funnel, we're usually targeting people who've already purchased and both people who've already visited the website with a purely direct response focused ad to make an action. Some type of scarcity mindset, just something that will try and get a user to make an action.
10:05 SW: And then we're always split testing different segmentations of audiences. Both on prospecting, both on website custom audiences and both of our buyer audiences. Just remember, there's no... You can break the rules here. You could be very very creative in figuring out which lookalike to build. Build lookalikes of your email lists. Build lookalikes of your fan... If you have a large fan base on Facebook, you can build a lookalike of your fan base and you can even exclude all your fans if you want to get totally new users coming into your funnel. This is the best part about Facebook is because you're able to be really creative with your audience test.
10:42 SA: Yeah, absolutely. Well, there's a lot of information here that people can go off of. I think building on your last point, you're never really done figuring out audiences on Facebook. It's just... There's so many different combinations. If you have any questions about any of this stuff, feel free to email us. Questions at MuteSix.com. You can e-mail Steve and I directly. We love talking about this stuff. We're always trying to learn what everybody else out there is testing. Yeah.
11:12 SW: A lot of information. Don't want to overwhelm you guys, but feel free to e-mail us questions at MuteSix.com. We'll reply with a very educated answer.
11:21 SA: [chuckle] Have a good one.
11:23 SW: Thank you.