Episode 48: Understanding the Facebook Learning Phase
In this week's episode, Steve and Stewart break down the Facebook learning phase and how to use it to better understand your results and optimize your ads.
00:01 Announcer: You're listening to the "Spend $10K a Day" Podcast brought to you by the performance marketing experts at MuteSix. This is your source for cutting-edge insight into the world of online advertising from the team with more Facebook case studies than any other agency on the planet. Here are your hosts, Steve Weiss and Stewart Anderson.
00:25 Steve Weiss: Hello, everyone, and welcome back to Spend 10K a Day podcast. Today, I got Stewart Anderson with me, our CMO, as well as myself. We're gonna be discussing a new Facebook ads algorithm update, this new amazing thing called Facebook Learning. Actually, if I were to say it's amazing, I'd probably be lying to you.
00:46 Stewart Anderson: It's gonna be an adjustment, I think, is more... What it's gonna be more than anything. It's, I think, made for more beginner to novice Facebook ad campaign people, people that are newer to the platform. For people that are more experienced on the platform, I think this is gonna be more of an annoyance than anything else.
01:04 SW: Yeah. It all goes back to Facebook methodology of they don't want you turning on campaigns and off campaigns. The way to optimize ad sets and campaigns is never to actually turn stuff on and off. They don't want you to do that. The new Facebook Learning update pretty much says that Facebook cannot optimize a campaign properly until it gets the minimum of 50 conversions. Until Facebook gets 50 conversions, it's in this "learning phase" where Facebook is just trying to figure out which audiences to open up your ads to within a given ad set.
01:40 SA: Yeah, that sounds... It's certainly a bit of a departure. As I said before, it's gonna be an adjustment for people that are used to the platform the way it is. And if you're doing performance marketing the way that we do, having any sort of adjustment is gonna be... There's gonna be a little bit of pain in the early parts where your performance is gonna be as good as it used to, or good as you're used to seeing it. Steve, if I was to... If I'm an experienced campaign manager, what would I do to best adjust to this kind of new reality, the new way of a Facebook doing these early stages to campaigns?
02:15 SW: I think it all goes back into ad set optimization. I think that turning on and off ad sets is never gonna be the recipe, so when you have an ad set that's not converting really well, lower the volume on it and keep filling more creative into that ad set. What I would say is to start optimizing the targeting in an ad set so you're out of the learning phase and then keep the ad sets live that have 50-plus conversions already in them. Because if you don't have 50-plus conversions, and then you're gonna have to go back to square one, and now Facebook's taking your bid again to the auction and you're gonna be back in square one, bidding against all the other hundreds of thousands of advertisers.
02:55 SA: Yeah, that makes a lot of sense. I can't tell you how many times on calls with clients, potential clients, when they talk about strategy and they always wanna test things really, really quickly. They wanna launch new campaigns and newer ad sets every day, they wanna switch from one to another, and the tough thing to tell people is one of the most important things about having success in the platform is having patience and not making snap judgements based on early results. You need to actually let things run, and Facebook is really emphasizing that here. What would your advice be to, let's say, newer advertisers? People who are just getting started launching a store or some sort of an online business, they might struggle to get 50 conversions if a conversion is a purchase or a lead or something like that. Would you advise those newer businesses to target something higher up in the funnel so they can actually fill that conversion pool?
03:52 SW: Yeah, that's a really good question because obviously, you wanna give Facebook enough data to optimize. You wanna show what a conversion is and have given up data to optimize. I would actually stay still optimized for conversions, but maybe start off with more ad sets and lower budget. Maybe you start off with 50 different ad sets or 30 different ad sets when you're starting, maybe do a lot lower budget. There's really two ways or two different strategies to look at it. If you know your audience and you know your demographic and you have some experience with Facebook, you know your look-a-likes ones of your buyers who are gonna be the best converting audience, maybe you wanna up your daily spend caps initially and buy the data and really optimize a lot quicker. Strategy number two is, "I don't really know who my audiences is. This is a new product, this is a new audience. I might not even have an email website custom audience," so what I'm gonna do is I'm gonna do a little bit of precise interest targeting, a little bit of look-a-likes off my website, and I'm really gonna mix and match and then create a bunch of ad sets with lower budget.
04:49 SW: Those are the two main strategies. I think the underlying thing is wanna get over that 50 conversion mark to get some stability in your ads, and when you have some stability in your ads, Facebook is gonna stop spending rapidly and they're gonna really understand your audience, understand how to actually drive a conversion or a good ROI.
05:08 SA: Yeah. You know, Steve, I think you started talking about this a little bit earlier, touched on it a little bit. What does this tell you about Facebook's approach from a product standpoint, to the platform? What are they trying to do here and what does this tell you that they might be planning for the future, if anything?
05:26 SW: Well, they're trying to tell you where in the process your campaigns are, as far as how quickly you are to reach your goal. More and more advertisers are entering the auction and what they're trying to do is they're trying to tell advertisers where you are and how close you are to reach your conversion goal, and I think more and more, Facebook is preparing to really serve the needs of a multitude of different types of advertisers, so I think that they're really thinking wholistically. They're listening to the pain points of advertisers 'cause there's really no insight into how the auction works. There's the Delivery Insights tool, which shows if you're winning auctions, first impressions, how many new people you're reaching. But at the end of the day, Facebook really wants to give advertisers the ability to understand where their campaign sits in line, and how quickly they are to reach their goals.
06:14 SA: Yeah. That's really, really good points. The more I think about it, too, it's maybe... And Facebook's obviously got some really, really smart people working for them. Maybe their growth team, their data analytics team, is looking at advertiser drop off. Where in the process does an advertiser who's starting to spend money on the platform reach a point and say, "I don't wanna do this anymore." And they're addressing those concerns now with this new feature. Maybe they're saying, "Hey. People are starting to spend money. They're in this no-man's land of... Maybe they have 10 or 15 conversions, or anywhere up to 50. Before now, they didn't have any visibility into how it worked, as you mentioned, and maybe Facebook is saying, "Because these people don't know how it works and they're not seeing results, if we maybe game the system a little bit more so it's more user-friendly at the early part of the advertising journey, maybe we can keep more advertisers spending on the platform." And again, that's just one theory I'd probably have to think a little bit harder about specifically whether that's the case or not, but that could be an explanation to why a future like this has come about.
07:22 SW: Yeah. It could be a mechanism for a form of retention of advertisers and to keep them spending. Obviously, they're very adamant that it's this 50 conversion mark. I think our Facebook team is really clear in that you need 50 conversions. So they're kind of quantifying what you need to be successful, which is interesting. I don't really... There's really no playbook on, "Alright. Once you get 50 conversions, now this and this is gonna happen." There's no real clarity, especially from a product perspective, of what actually goes on once you've reached these 50 conversions. We're just on this podcast making assumptions of what goes on.
07:57 SA: Yeah.
07:58 SW: But as far as what goes on behind the scenes and how the auction works and the methodology of what happens to a campaign when it's out of learning phase, we don't really know about. What we're trying to do right now is we're trying to speculate, we're trying to use the data that we have from our current campaigns, from managing $100 to $150 Million a year spent, to really share that intel with other people who are experiencing similar problems and really prepare other listeners in this podcast or audience for what's coming out of the chute next. And I really want everyone to really understand what this new Learning algorithm means because, the holidays are coming up and everyone is under the ball and having pressure to really scale campaigns and scale campaigns profitably.
08:39 SA: Yeah, absolutely. Do you have any other major points or takeaways that you would want maybe a newer or intermediate advertiser to take away from this new Facebook Learning feature?
08:51 SW: Yeah. I would say... That's a good question. I would say don't get into the phase of doing one strategy. I think that there's so much discourse and so much new stuff coming to the auction that what you were doing three months ago, two months ago, one month ago might not work like it did at three, four, five months ago. Obviously... I'll give you an example. We're dissecting a campaign that's spending about $7,000 or $8,000 a day, and we saw that, at-scale, precise interest audiences were working a lot better than look-a-like audiences. Well, why is that? I don't know. I guess when you spend at scale and you have all these other byproducts that are going into what makes a campaign profitable, that interest-targeting still works. And that goes against everything that Facebook's reps tell you, everything Facebook tells you, but you still see pockets of success. Obviously, this podcast is specifically talking about the Learning algorithm update, but I really wanna stress to our viewers and the listeners of this podcast to consistently be creative in the way you're thinking about your campaigns. When nothing is the way it is, it's always gonna be the way it is. 'Cause remember, each time that you actually go into the auction, it's a new bid and a new journey. What might not have worked two weeks ago might work now, and that's the thing about Facebook.
10:17 SW: There's really no rhyme or reason to anything that works. And there's no background, historical data that says, "This is going to work." I stress to everyone to really go in with an open mind to the holiday season and start using this Learning algorithm to really understand where Facebook's product is going. They want high engagement ads, number one. You know that they want people to engage with your ads. They want the user experience to be paramount. With that said, start trying to build ads, and then when you put those ads into your ad sets, be a little patient. Maybe start off... If you wanna get the data back quicker, start off with a high budget, but don't kill that ad set. Start rotating in newer ads to that ad set because that ad set has conversions and history. Just be very thoughtful about how you optimize your ad sets.
11:02 SA: Wow. That's really, really, really sharp. Anything else?
11:09 SW: Let's see... What else? Instagram. Using Instagram... Top of funnel has been working really well, I think Instagram is obviously becoming a new route to actually drive conversions.
11:21 SA: Yeah.
11:22 SW: I think every month, we're seeing a pick-up in Instagram conversions, literally every month across the board. A lot of people like to take Instagram and put it off on its own into a new ad set. We're still seeing a lot of success on direct response taking Instagram and doing placement objective.
11:43 SA: Yeah.
11:44 SW: There's no right or wrong way to do that. There's always two options: A, split it off, have specific ads for Instagram. That's logical, rational, makes sense, probably helps conversion rates. But at the same time, allowing Facebook to really placement optimize your ads also has been profitable for us. Just having an open mind and running split tests that actually make sense, that are actually gonna drive the bottom line. And that's also something I think about on a daily basis, is running split tests that have a positive impact on the bottom line of your business. One other thought is remarketing versus prospecting, really understanding the bottom-line impact remarketing has on your funnel. I think that's another topic for another podcast, but really, doing conversion with tests and seeing remarketing as a new customer acquisition, remarketing to people who've been to your website but haven't bought. Really running a conversion lift to really figure out, "How much of an impact does it have running it to this audience? Will they purchase on another channel or will they purchase here?" But that'll be another podcast.
12:48 SA: Definitely.
12:49 SW: Stay tuned for for the cool areas of the nitty gritty conversion lifts. But appreciate everyone, another awesome podcast about the new Facebook Learning update. Just a quick recap: Facebook came out with this Learning update. Now, you need 50 conversions on an ad set before Facebook will actually show your ad to the right people. Before you hit 50 conversions, you are in Learning mode, where they're just spending budget, trying to figure out what to do with your ads. Once you get out of Learning mode, now you have what Facebook calls a little bit more consistency as far as hitting your conversion goal, especially on manual bidding. Don't expect to hit your conversion goal on manual bidding while you're in your Learning phase.
13:35 SA: Sounds good, yeah. Thanks for joining us today on the Spend $10K A Day podcast. Again, it's Stewart Anderson and Steve Weiss. We're looking forward to seeing you with another great episode coming up very soon.