While it has plenty of features and tools to play around with, one thing you should always be doing in Facebook Ads (and for your digital marketing program overall) is run A/B tests. A/B testing helps ensure you’re maximizing the performance of your budget by constantly trying new ways to get better results. Emily Lewis shares her years of experience in Facebook Ads with her tips and tricks for A/B testing.
Learn more about Fujisan Marketing here or email your questions/comments to contact@fujisanmarketing.com.
Full episode transcript:
Preston Parshall 0:13
Welcome to the digital marketers guide podcast where we break down digital marketing topics into bite sized episodes that can help make your marketing program and you a little better. I’m Preston Parshall, the founder and CEO Fujisan marketing. We’re here today with Emily Lewis to discuss Facebook ads a B testing. Emily has been in digital marketing, and specifically social media marketing for the last nine years. And she has a deep understanding of Facebook ads, as well as other social media ads platforms. And she’s been on our team for the last two and a half years and is our paid social channel manager. Welcome to the show. Emily.
Emily Lewis 0:55
Thank you. Great to be here.
Preston Parshall 0:56
So to start off, why is ABX testing on Facebook important?
Emily Lewis 1:01
Well, as marketers, I feel like everything we do, it’s really important to be continually testing. And so it’s basically the core component of everything we do on paid ads too. So when we’re testing, like, it could be a geographic area, it could be an audience, it could be a creative type. We’re learning something every single time we’re running one of those tests. So basically, it’s just the fundamental part of making optimizations when we’re running ads.
Preston Parshall 1:30
Emily, can you tell me about a time where you set up a test for Facebook and AB test that was successful, or one that interested you?
Emily Lewis 1:39
Yeah, well, every month, we’re running a variety of different tests. I think one that would be really interesting is, let’s say it’s for a product for a client, it’s a, it’s a retailer client, and they sell to men and women. Obviously, most retailers, ecommerce brands will segment clothing by gender. So we would typically separate those out in our ad sets as well, one time, we had a product set that was pretty gender neutral. And we decided to still segment the audience by gender. So we would have one ad set for men and one for women. And given that the majority of this audience was men, about 80 to 90%, we expected most of those purchases would be coming from men. But what we found out was that women were at a much higher purchase rate, they were higher engaged, and they had a lower cost per click. So that test told me that women were actually an untapped market for this particular client.
Preston Parshall 2:41
Why do you feel ABX testing is important on Facebook,
Emily Lewis 2:45
Facebook is amazing, because we can segment by so many different things. And we can actually measure things, apples to apples, versus some other platforms, like Google ads, for example. So with Facebook, we can identify smaller segments that we want to test against. So for example, we’ll test age groups. So for one of my clients, we’ve been testing like a younger audience versus an older audience and seeing which one is more engaged. We have segmented that 25 to 45, and then 46 to 65 plus. We’ve seen that the younger audience who is their target is much lower engaged than their older audience. So one of the things that we’ve been able to do with this is say, Okay, it looks like the current creative we’re using isn’t as engaging for the younger people. And we did kind of a creative pivot in order to engage with them.
Preston Parshall 3:42
How did the client accept that news?
Emily Lewis 3:45
Well, they were expecting the older audience to be a higher, you know, engaged audience, because that’s who is purchasing their products. But, you know, they, they were kind of surprised, to be honest, because they feel like they’re creative is really interesting for a younger audience.
Preston Parshall 4:03
Testing, testing, testing. All right, have you run any tests that have surprised you?
Emily Lewis 4:08
I feel like I’m continually surprised. I always like to go in with expectations of what I would anticipate, especially if it’s for an audience that I’ve been running ads for for a while. But I think one of the ones once again, it’s a gender test, but it was a client believed that their target market was men in kind of an older range 45 to 65 plus, but what it turned out to be is that women were the ones that were purchasing this was for a specific event. And women were purchasing at a rate like three times higher. So that was shocking. Additionally, some of my clients come in to us with like persona work done, they’ve had work from another agency, and they come in saying, okay, these are our four key audiences that we want to target. And then I will set up very segmented audiences on Facebook ads, because we have that capability, and I run these tests and we don’t really see as good of results. This happened with a client, not too recently, it was like maybe six months ago. And then I proposed, let’s broaden this out. And when we broadened the targeting further, we were able to see much better results.
Preston Parshall 5:17
Yeah, that’s interesting. So a lot of companies, I wonder if they’re using research or research companies in their personas that they’re building nowadays Because sometimes it really doesn’t seem like that, like they’ve skipped a step or like they don’t want to spend that money.
Emily Lewis 5:31
Yeah, I would agree. I also think that when it is an agency, or a different company compiling those results, it might not be as reflective of what what we’re seeing in Google Analytics, for example,
Preston Parshall 5:44
what things do you consider when you’re setting up an A B test?
Emily Lewis 5:47
I consider all sorts of things. So this really is part of like, the conversation that we have with clients is what are the things that they what are the assumptions that they may have? And can we kind of build upon those so things that I consider, I would want to ask them what kind of creative they have. So the best opportunity is I love to test different creative, I want to test a static a video carousel ads, all of those are features that we can run on Facebook, if they have the ability to create video in multiple formats. So that square size, the landscape and portrait, the portrait size being great for reels and stories, we can test placements as well. So we can see do reels do better than the traditional feed. It really depends especially because we want to make sure we’re testing basing on what their objective is, if it’s to get brand awareness, then feeds and stories, the combination probably would be great. But if they’re trying to make purchases, then we might need to pivot and just test creative type. So once I understand their objective, then I can kind of give them recommendations. Another one is also just audiences. So whenever they have an assumption, I always want to either be able to back that up or test it. So like I’ve talked about a few times throughout this, we have a variety of different assumptions coming in. I also like to always test prospecting versus retargeting audience audiences.
Preston Parshall 7:19
And if it’s the same type of creative, say it’s static, or video, or carousel, what are individual elements that you’ll test within those? And that’s a huge question. And you can answer however you want.
Emily Lewis 7:30
Yeah, that’s a good one. And that’s one of my favorite things to test. So I like some of my clients, they do not have the availability to create video, or we have a quicker turnaround, so they just have static, and that’s perfectly fine. Something I like to do is say, let’s just test two very different static images. So maybe it’s like for an E commerce product, it might be like that product, the hero image of the product versus a lifestyle photo of the like a user applying it or using it or wearing it in some way. So I like to do features versus benefits. So features being more like technical benefits being what is that user going to get from purchasing XYZ. Another thing that I love to test is ad copy. Now this is something that’s not as, although I found some ad copy tests to be really beneficial. It’s not the only thing that will move the needle. So for Facebook ads, I will do short form versus long form. So long form, meaning you have to press that see more to read the whole ad copy. I like to use emojis sometimes versus no emojis. And that’s also really relevant on like, if the client is a really professional company, they might not want to use emojis. But otherwise, it’s great for social ads.
Preston Parshall 8:47
What would you not recommend? Do you have any horror stories?
Emily Lewis 8:51
I don’t really have like horror stories. But there are some things that I avoid. So sometimes a client will want to Well, in a true A B test, there really won’t be more than one variable that’s changed, right? It’s just the one variable that you’re testing to see which thing did better. And then you can turn off the bad thing and then go with a new test. But sometimes clients will take that to like the next level. Headline tests aren’t really they don’t, I haven’t seen headline tests be that dramatically different. Just because it’s such a small element and on Facebook and Instagram, not all placements will be able to show the headline. So if everything else is the same, it’s just the headline that’s changed. In some places like the search placement or Instagram explorer placement, those aren’t going to be visible and therefore it’s not even going to be a test. Another thing is assumptions. So like I said, clients go in with assumptions we go in with assumptions before we run a test. And sometimes we’ll make an assumption like this time of year we’re just not going to get any for this particular product or this particular service, it’s always a good idea to test it, even if it’s a small test budget, it’s worth kind of exploring, unless, you know, it really doesn’t make sense for that client.
Preston Parshall 10:10
That’s why I love digital marketing. So what tests do you decide to pick to run? I mean, you have a limited budget. And we have limited resources. And so how do you decide what test to run and win?
Emily Lewis 10:23
That’s a great question. And this is something that I think a lot of times we need to talk to clients about a little further to make sure that they really understand what ones are our priorities, something I talk a lot about is statistical significance. Facebook actually has like a specific AB test opportunity, you can leverage it or you can just manually do it, which is what I prefer. But it’ll tell you when your test is complete when the statistical significance is final. And what that basically means is like the here’s an example, you have two ads running ad A has a click through rate of 5%. Ad B has a click through rate of 1%. You would think okay, ad A is the winner. But what if I told you that ad A only spent 25 cents and ad B’s spend $250 That wouldn’t be statistically significant. And so we want to make sure that we have enough budget in order to really get the results that say, Okay, this one is the winner. My rule of thumb is I like to have at least $30 per day per ad. So depending on how many different ad sets meaning audience and placements or different ads that you have running, you have enough in order to see those results. There’s also calculators online. I think Neil Patel has one, but they’re all over the place. And I would highly recommend plugging your numbers in to make sure it’s statistically significant before you make that call.
Preston Parshall 11:50
Thanks, Emily. Thanks for listening to the digital marketers guide podcast. Be sure to subscribe to stay up to date when a new episode is released. If you have any feedback or topics you’d like to see covered, email us at contact@FujisanMarketing.com or learn more about us at FujisanMarketing.com. See you next time.
