← DTC Podcast
DTC Podcast
Ep 554: Q4 Meta Strategy in the Andromeda Era: What’s Changed and How Smart Brands Are Adapting
00:24:20
Loading summary
Transcript38 lines
- [00:00]
A
What is Andromeda?
- [00:01]
B
It's a lot of things, but essentially what it is, it's Meta's next gen, like ads retrieval engine. And it means a lot of things that we're going to get into. At the very base level, Creative and ads are now leading the way.
- [00:15]
C
We've always started with the angle or your core message being the most important. From an order of operations perspective, we still focus on the angle itself, but what we're doing is we're not going out and testing every single text idea we come up with for an angle or going really broad with like 25 statics that all just focus on slightly different text variations. Now we're focusing more on. I think one example of something I've seen change that's really interesting is.
- [00:51]
A
It'S all killer, no filler. I'm Eric and I'm here with Pilothouse's Meta brain trust, Jacob and Taylor, to talk a little bit about Andromeda. Now, let's start by prefacing this. I'll preface it with a David Herman tweet. Tweet that I thought was great at the end of last week, which is Andromeda is not new. Retweet and save lives. So what's happening on Meta with Andromeda is not a new thing, but I feel like this Q4, we're really starting to feel its effects and so people are talking about it more. Jacob, what is Andromeda?
- [01:22]
B
Yeah, I mean, it's a lot of things, but essentially what it is, it's Meta's next gen, like ads retrieval engine. And it means a lot of things that we're going to get into. But at the very base level, creative and ads are now leading the way in how ads are being targeted, who you're hitting. And it's a system that narrows millions of possible ads for a person down to just a few eligible ads. This means a lot of things when it comes to your Facebook structure, your account structure, how you're actually approaching your targeting, if starting with things like Personas and whatnot versus, like, you know, let's go and target these people. So, yeah, it's. There's a lot of hardware and software behind it. You know, the Nvidia Grace Hopper super chip, Meta's MTIA chips, all these things that sound very advanced. And we're going to kind of break it down in layman's terms. I like to think of it as kind of two parts. There's. There's Meta's Lattice, which they recently released in the last six or so months. And I like to think of that as the library and then Andromeda as the brain that searches that library. And Meta kind of uses the same metaphor on their articles. But it used to be that when you were out there running ads, there were segmented libraries by industry angle. And when I say this, I'm referring to Meta's delivery engine, how your ads were being delivered. So you had to work to find the right library and pair it to the right person and then find the right book for that person. Now with Lattice you have this big library where people are walking in and Andromeda is searching that massive library and finding them the perfect book immediately. So it's unlocked a lot of developments in AI delivery. Less work being put into some of those things on the front end now and more work being put into like creative strategy. So if you haven't adjusted, you know, your strategy and whatnot yet, you're, you're definitely behind. So I'd say that's kind of my high level summary.
- [03:36]
A
It's, it's definitely the trend. I just saw Elon tweet the other day about essentially doing the same thing for the X algorithm. Basically saying instead of doing it through all these like man made steps or whatever, these categories, he was just like, I'm just going to let Grok look at every piece of content and have it determine who it should go to, where, where it should go. And I feel like this is a move to, to that sort of like mentality when it comes to analyzing creative. Super interesting. What's gem?
- [04:02]
B
GEM is the super brain. So basically just machine learning model trained on thousands of GPUs, optimizes your results. It's similar to what I just explained with Andromeda. But GEM is actually what is fueling it. It's increasing conversions like it is getting your ads in front of the right person quicker. Essentially it catalogs, analyzes and connects all the info from your ad. So like the creative piece, you know, the content, the copy, the headline, the visuals and the creative everything that we don't even have control over. And what it does is it like categorizes this has its algorithm set up and it'll just provide the ad to that person at the right time with low latency. So that's GEM is just another piece of this like overarching thing. And there's, there's kind of four pieces. There's gem, Lattice, Andromeda, and then there's, there's the whole sequence learning side of it where basically it's enabling the ads to consider the sequence of actions people are taking. Not just, not just one action, but down the road. Oh, there's always going to be a click, a card edition, a purchase and meta starting to take all this into account with how it's also delivering.
- [05:19]
A
So that's super interesting. And this has been maybe, Taylor, this has been in effect, like, how long has this been? The, the modus operandi. I know this was introduced at the end of 2024. Has Andromeda been running the show this whole year?
- [05:32]
C
It has, yeah. So it's been. It's been an interesting thing. And like, as with anything the, A lot of these shifts, there are gradual tweaks and refinements to the engine behind it. But as far as everybody knows, it's been basically since December 2024, it's been active then.
- [05:48]
A
So what sort of changes has it led to us making in how we service clients, how we. How we deliver ads?
- [05:54]
C
I think it's really. And the reason I think this has become such a prominent topic, like right now as we're kind of gearing up for Black Friday Cyber Monday this year, is everybody's thinking about their creative strategy and like what their Black Friday is going to look like now that we're in a crunch time and we're a month out. So what it. How it's translated downstream to the brand side is I think a lot more excitement about leveraging a diverse range of creative and then being able to focus on marketing fundamentals like aligning your content more to Persona rather than trying to game the algorithm. Or spend your time figuring out like what targeting and all the nitty gritty details and instead thinking about the bigger picture and bringing that to life through your concepting approach and, and how you, how you translate all that into your ad account.
- [06:41]
A
What does this update do to. In terms of creative redundancy? That's something I'm hearing a lot in the space, is that it makes a lot. I think a lot of approaches in the past would be like, you get a winner and then you create 10, like slight variations on that winner. And what I'm. What I'm kind of understanding is that it really behooves advertisers to be taking bigger swings rather than taking making smaller incremental tweaks to ads. Because of the way this Andromeda works, it kind of makes things that are similar often redundant. Is that, is that accurate?
- [07:15]
B
Yeah, basically similar creatives now would almost be bucketed like under a similar bucket. So like as they're delivering you're not getting that exposure to, to the probably Personas you're trying to hit. So definitely more variety is more important now. As you find a winner, you're going to want to think on why that's a winner. And then ideally you're going to introduce variety around that concept, you know, rather than, hey, now let's get eight variations where we change the font or change the color. You know, there could still be value to that, I think, but at the end of the day, you do need 5, 6, 7, 8 varying, you know, contrasting ideas within each campaign or ad set now. So, yeah, you definitely want to take bigger shots. And because the creative is leading that targeting in the sense of, like, how this delivers. If you have some, like, scrappy ads in an ad set with polished ads now, the delivery on those are going to be separate anyways. So you kind of want that variety to be.
- [08:19]
A
To be honest, I know, Taylor, we've been running things like the pilot test or talking about this idea of, like, clearly running five separate angles, seeing what works best, what has changed in how we think about, like, angles, because I know, I think one of our first podcasts ever on all killer, no filler, was about angles. And thinking about that, like, how is it just a matter of going deeper from angles really into Personas, making sure that the whole purchase funnel is congruent. What specifically is different? Maybe about how we test creatives and contrast test creatives with Andromeda versus how we did it before.
- [08:50]
C
Yeah, it's a great question. I think it's interesting because both. A lot has changed, while not a lot has changed at the same time. And I think anybody who's on DTC Twitter kind of gets that sentiment from this discussion and what, what it looks like we've always started with the angle or your core message being the most important thing that's going to drive people to take action. At the end of the day, every, every marketer, advertiser knows that that typically is the case. So, like, from an order of operations perspective, we still, we still focus on the angle itself, what's actually going to agitate and then drive interest and then drive the action. But what we're doing is we're just, we're not maybe going out and testing every single text idea we come up with for an angle or going really broad with like 25 statics that all just focus on slightly different text variations like we might do at points. Now we're focusing more on. Let's look at the groups of concepts and like, let's go angle to concept, angle to like, and then that concept style. So it might be, we come up with a concept around a message for an angle and then we create a video, a static and then we go move on to the next one, move on to the next one and go through it like that. So what it's resulting in is I think more quality, less quantity and more range rather than iteration really. But to be fair, like, I think Meta's kind of primed everybody to be aware that this is where the platform's been going for a long time with broad targeting and all the very, the focus on consolidation and all those kinds of things. And now is where, you know, we're, we're expecting it to, to ramp up and hoping to see some good things for brands that are, that are opting into that.
- [10:36]
A
I feel like we've been saying creative is the new targeting for at least three years, at least since iOS 14, that update. But it's. You're right, it's like they were priming us for this environment because that really is now the case more than ever. Is that what you're seeing as well, Jacob, where we're just generally going broad and then letting the creative do the, the targeting?
- [10:55]
B
Yeah, definitely. For a sense, like structure is simplifying. Like, definitely would advise against, you know, running too many campaigns for testing reasons, things like that, because, you know, you can get a lot of that data quicker and more accurately now by just sort of bucking it under, under one campaign or ad set. And then with how Meta's, you know, introducing all these AI optimizations at the ad level as well. One ad might become 50 to 100 ads because they're doing different overlays and delivery, you know, changing the headlines and all these things. So if you have all those enhancements enabled, you do kind of want to simplify and then allow that variety to happen at the ad level again to that, that same point and having, you know, all these, these enhancements and all this stuff, it's happening at the ad level. Meta's not saying let's build enhancements to add 30 campaigns into the account. It's all just ad specific stuff. So it's all pointing towards that. And now this, you know, Andromeda push, the past, past year here, six to eight months is, is really echoing that. And results are starting to show that like, to Taylor's point, it's some of the traditional methods are still outperforming. So it's not like just go in and switch everything over, but they are leaning this way and all their releases are pointing this way. So you're kind of battling against an uphill battle if you're, you're just stuck in your traditional method, if that makes sense.
- [12:26]
A
And so what, I guess it varies by account, but are we tradition, Are we in an era of testing campaigns and scaling campaigns and basically maintaining those two campaigns, adding to the full broth in perpetuity?
- [12:39]
C
I would say from my perspective, it's definitely more blurred than it used to be. I think those fundamentals are still there, but now there's more emphasis on this creative flywheel and being able to feed something that keeps going. And then there is an element of control that you still want, right? Because there are certain media buying, I think, tricks and tips and things that won't go away in all of this and that like, and a lot of them are rooted in fundamentals, right? Like you want social proof buildup so that you have a lot of great content for people to engage with underneath the ad. That involves a long standing ad. That might involve moving that ad to different places to see it catch momentum. But I think what we are starting to see is the system can handle more ads than it used to be able to. And so being able to have a level of consistency in your account rather than trying to game certain times of day or things like that, like, that's where a lot of the shift is happening. So it is naturally becoming more simplified. So it's account to account. For me, I've got some accounts where I'm launching ads in scale campaigns and then breaking them out into additional audiences as supplementary plays. And it kind of depends on how we're seeing that account react to the content based on the Personas we're targeting. But I have other accounts too where I'm still running test campaigns and graduating them up like you would see. And I think now it's kind of an open playing field. It's a matter of everybody kind of has to test and see what works best. But generally staying more simplified works better. From what we're seeing, you teased a.
- [14:08]
A
Little bit there, but art, like what I remember, like, because I come from the Facebook ads era of Tim Burd's the Shotgun and the Surfing method and all of these fun methods and hacks and trips. Like, are there still any of those? Like, what's an example of a hack? Or like, you know, I always think of the way that you're sort of forcing meta's hand in a way that they might not. Their algorithm might not want to act, but you have a reason, whether it's a promotion or something or a creative test that you want to see done that that might not work if you just let the algorithm. Algorithm take over. What's an example of a hack or one of these things that still works?
- [14:41]
C
I think one example of something I've seen change that's really interesting is we used to like even last year around Black Friday, we would day part more and there was this pattern you would see in a lot of ad accounts where Meta wouldn't spend through in the final two or three hours of the day at the same level it would earlier in the day. I've seen more consistent spend through all hours of the day at points this year than I had in years past, even when we're looking at hourly breakdowns. So that's one example of how the system is starting to shift a little bit. And I think that's really interesting. But I mean there are still components at the end of the day. Meta is typically a game of momentum for brands. And so being able to time your evergreen launches with when you think they're going to catch and start building social proof. And so that might be during a key moment so that you can build momentum up with them and carry it all the way through. Those kinds of things I think are still valuable and still in play. But generally speaking, we're moving away from the bully method and all those other kind of.
- [15:40]
A
The bully. That's right.
- [15:41]
C
I forgot about all those classic media buying tactics.
- [15:44]
A
Is it still, like you say, the more things change, the more they stay the same. You're still going to have winners that make up 90% of your budget when you have these winners going. But in a perfect world, it seems like you should be able to make a very diverse array of creative. And because you've got this library concierge worker who's just finding the right piece of content for the right person, you should be able to to get more diversity in your creatives that that win. In a perfect world, are we seeing that or is it still just as consolidated with the few winners as ever?
- [16:15]
B
I'd say more winners now and less ads like hogging the spend within each ad set kind of conversely to what it used to happen because that's what everyone's worry is, is like you're running, you know, all this variety in a campaign, an advantage plus campaign. And two of the ads get spend and the rest just doesn't. So that is starting to shift a bit to the point of just how Andromeda works. And if you're putting in good quality creative that is aligned to your Personas that you need to hit. Ideally, all that creative is going to start, start spending and you're going to uncover more winners. But it's just that fundamental shift is still happening. Like Andromeda's updates are still coming out, so there isn't even like a best practice per se yet. Aside from try to consolidate, introduce creative variety and make sure you do that back end rough work on really doing the research on your brand and like where are the gaps? Do the SWOT analysis, find out who these people are that you need to target.
- [17:21]
A
Okay, let me get some live ad consulting because we're just redoing a bunch of the ads in the DTC account and so we're not an E commerce brand. So our consideration cycle is much low. We're just trying to get someone to subscribe to the newsletter. And so what we've done is we've taken all the icp, all the data we have on our ICP through all our various surveys, pilot houses, surveys, and I pump those into a GPT and I'm having them basically create the words that have been reflected back to us from our ICP about the challenges they're having. Then I'm writing an ad text that's about how the newsletter has content that solves those specific issues, which it does. I'm using sticky notes, which I know are a testy issue, but it's basically my idea is that if we just create enough of these sentiments, they're going to resonate with one person or another having thought or felt these things and they'll click through. Is that a good use with Andromeda?
- [18:12]
C
I think it's aligned with the direction. Yeah, that's spot on in a lot of ways. And then as you go, you'll just want to watch to see if the data supports that. Because that's the other thing that I think has been missed in some of the communications from Meta too is they brought back a lot of the old breakdowns and things that kind of went away with iOS 14.5 and are really valuable. Like you can go back and look at 28 day click performance to understand the consideration window a little bit better. You can break down by your demos and stuff like that. So I would just encourage you to look at that as you're pulling the data in from this test.
- [18:43]
A
How are things looking? We are in Q4 firmly. We are in the ramp up phase. What are the early signals about what we're expecting for this? This Q4 on, on the full Andromeda system.
- [18:54]
C
It's been an interesting start to Q4. I like to think of this time of the year as like that weird point between you're going from short hair to long hair, you're growing it out and you're kind of in that in between phase where you maybe don't want to be going and showing it off right now, but you're, you know, you're working towards something good. That's kind of what this part of October typically is. So I would say like general sentiment from my end has been pretty solid. From the like the paid side, I think from the consumer side, we're seeing maybe a little bit of lag in intent. This is kind of that period where like, you know, Everybody's excited about Q4, but like black brothers around the corner, everything hasn't quite settled in yet. I'm noticing even like as of this week starting to see some shifts in like things picking up a bit more. So I think we're about to hit the ramp up point but. But it's maybe been a little bit slower than some brands had hoped for. I think that's kind of the general sentiment from what we've seen and a lot of brands are in that point of just trying to fill the funnel and get ready for the big show coming up next month.
- [19:54]
A
Very cool. Taylor, just in some of the notes for this episode, we didn't dive deep on this, but I think it'll really benefit people when we talk about distinct angle. Like you wrote some like actual numbers down about the way people might want to be thinking about distinct angles, creatives, variations, all those things. How would you. I know this is just a rule of thumb, but how would you maybe lay that out for people? Curious.
- [20:17]
C
Yeah, I think that's a, that's a great call out. Usually from that we kind of, we try to put. Promote a bit of a structure with it so that it's clean and easy to kind of think about how you can go about it. So the easiest way I think if someone's looking for a bit of a framework for how to work with angles and concepts is almost like a 3, 3, 3 where you, you look at three distinct angle concepts. There are three pretty clear levels of the funnel. So your top of funnel, your middle of funnel, your bottom of funnel and you can basically start with an angle or message for each stage. So you have angle for top of funnel, middle of funnel and bottom of funnel and then come up with a few ad concepts. Three ad concepts for Each so that you can look at three contrasting ideas and then from there you can go with format. Right. Like for some brands, depending on your resourcing, it's a lot easier to work with statics and get those ready and come up with three different concepts for statics. Some brands will want to mix static and video, some will want to focus on video. You might want to do a catalog, a video and a static or something like that. But that's a pretty simple framework for being able to go through and come back with some clear outcomes and learnings at each, each stage of the funnel based on your research and the, the marketing approach you're taking with it.
- [21:37]
A
And then from there, what would a refresh cadence look like with that trough of creatives?
- [21:42]
C
Yeah, so from there usually, I mean, it really depends on what you learn from it because like there are lots of different scenarios that could result as the outcomes. But typically, like if you, if you start to find winners with that, like you're now with this, you know, you're, you want to balance creative velocity with quality and contrast for Andromeda. So typically from our end, that's usually coming up with new concepts and launching them weekly and then being ready to kind of keep pouring in and make sure that your budget supports it too, because obviously that's a big part of it, but I would say roughly weekly. The cool thing is from what we've seen Meta over the last few years, they've gone from it being common to have three to five ads in an ad set. Then they started testing with the Advantage plus window up 10 to 30 and then 30 plus. And so we're starting to see the system be able to handle a lot more. And so if you do have a lot you want to work with and you're seeing at work, you just, it's worth continuing to just feed that and, and keep feeding the machine as Meta is building this to, to handle more and more ads across the ecosystem.
- [22:46]
A
Well, we've got, we've launched now a couple of different podcasts with Meta themselves, so. And I know we have another one coming up, so I'll make sure I can, I'll dig deep with them on, on the company line about Andromeda. But it's, it's overall, would you say it was a pretty like to go to like a whole new paradigm of ad serving in the background? Like it's pretty incredible that there haven't been massive performance lags or performance hasn't fallen off a cliff. Like it's been pretty seamless overall.
- [23:13]
C
Yeah. And I think again, it's really been brand dependent from what I've been seeing across the space. I think really dependent on how everything was set up to handle it from our end. I think what I'm most excited about is this feels like a deeper and deeper shift into true marketing and less gaming the system, media buying. We're just moving further away from that. And even as a media buyer, that gets me, gets me stoked because I think the, the marketing strategy and all the, all that work is a lot of fun to dive into.
- [23:42]
A
Very cool. All right, well, thanks for coming on again, Taylor. We'll have you back again soon to get an update on Q4.
- [23:47]
C
Amazing. Thanks so much, Eric.
- [23:55]
A
Thanks for listening to today's episode. If you're not getting the D2C newsletter, you can subscribe for free at Direct to Consumer. And if you want to learn more about Pilothouse's all killer no filler services, take off to Pilothouse co. I'm Eric Dick and this has been the DTC podcast. We'll see you next time.