He Calls My $10M Creative Process "Brilliantly Foolish." Here's Why. transcript

Jess Bachmann is a three-time returning champion on the Andrew Ferris podcast talking as always today about creative for e-commerce brands, how to make ads that sell products for e-commerce brands. And today we're going to talk about some end of the year wrap-up. What happened this year? What do we notice in in Meta and any other platforms that we're working on? But specifically, Jess came today ready to tell me why the creative process that Patrick and I are building at AJF Growth is not going to work and why it is.

I think he said I think the quote in the Slack DM was brilliantly foolish and it's just it's everything I'm looking for in a comment from Jess to me about my approach to advertising. As always, Jess and I have plenty to disagree about, but I think also a lot that we enjoy talking about. Let's get into it. talking about creative ads, all those things in the world of AI and e-commerce and all that stuff. How do you make ads and sell products for your e-commerce brand?

Let's get into it with Jess Bachmann from Fire Team. >> Howdy. How's it going? >> Great here. >> Yeah, >> it's I'm I was really looking forward to this. I was I I always look forward to talking with you about stuff and uh and this is this is a this is this is going to be exciting. So, tell do you want to get right into it? Tell me why I'm stupid. >> Yeah. Well, uh, I came to deliver and it's not just you. It's Patrick is also stupid in this [laughter] regard, too.

So, let's let's share the blame around here. >> Yeah. Yeah. Okay, great. Great. >> But, >> I will say I I love what you and uh Patrick are attempting to do. Um, with >> Do you want to summarize that for people who don't know? >> Um, so you have this uh ambitiously named Prometheus grief coowriter tool. Um, and from what I gather, it has information about what makes good ads and you can put some inputs into it and supposedly it it spits out a very competent brief.

Is that Am I missing parts of that? >> Uh, well, it's more than a brief. It actually uh spits out a script. Um, full it actually does >> spit out uh I I guess there are elements of it that are brief because it'll have some creative direction for for editing and things like that too. Although although we take those a little more with a grain of salt, it's sort of a little less tuned to do that specifically. And and I by the way, we haven't talked about this yet, but Patrick named it Prometheus himself first of all before Bezos named the company Prometheus.

And I think without realizing that he was a little bit stepping on the fire creative territory, so or maybe it was a maybe it was a conscious nod to you guys. I don't know. But Prometheus of course being the mythological figure, he steals fire from the gods and brings it to humans. So yeah, that's right though. Basically, right? Right. So, the idea is like it's a co-writing tool that we use that is loaded up with critical context on both the individual brand as well as more broadly what we think makes good ads in in GTOC.

And uh I I could I could nuance that more, but I'll just leave that there. And and we're sort of always working on it, always trying to make it better. And we give that to our creative strategists as like a core part of our process. It does some other things, too, but it's a core part of our process for actually generating scripts that we think will sell products. So, yes, that's the basic idea. Right. And so I I certainly love the attempt and I think anyone, any agency, any in-house brand needs to be putting some effort into systematizing this brief for script creative process with AI.

It is certainly the future. However, um I do I do feel Thor Patrick because I have gone down the exact same rabbit holes that he has. And what he's done is built some very elegant workarounds within Claude to get around Claude's limitations. And so one of one of the limitations in claude is that um you you can it can suck in a lot of context but it has this sort of lost inthe-middle um attribute of it which uh a lot of AI has where it sort of it understands the top stuff and it understands the bottom stuff but the stuff in the middle sort of gets lost and um that can be sort of problematic.

So what he's done I believe is like chunk things up until they're very small. So there essentially is no middle to these various documents and Claude can read them all uh etc. And so that is a a sort of mad scientist approach to getting around what one of the major limitations within running cloud native. >> Yeah. So uh before you say anymore, I'll I'll say a couple things. One of them is I will link the episode of the podcast in the show notes here that Jess is talking about uh where Patrick walked through the architecture of of the information and the context that Prometheus has in Claw to accomplish this.

And so anyway, that's in the show notes if you're wondering what that is. We really tried to go out of our way to show the work to try to be as helpful as possible. I literally worried on that episode while talking to Patrick like, are we going to need to delete this because it give away too much of what we think here? Does it sort of make us not useful? We don't think it does. We think uh for a bunch of reasons, but anyway, that's the level of detail we tried to provide.

We really tried to leave with value. So, there's that. Secondly, yes, you are correctly explaining a lot of what has happened here. And I think calling it a limitation of Claude is one way of doing it. I actually thought about bringing Patrick onto this episode for this reason. But I would say another way of thinking about it is sort of trying to more deeply understand beyond just like AI is cool and makes reasonable scripts. trying to more deeply understand >> how Claude thinks bas so to speak and how Claude works including using things like Claude skills which has been a really critical part of this really specifically actually you mentioned this problem very specifically of of Claude reading the top and bottom of your context but not the middle there are a number of reasons for that at a technical level that I actually don't fully understand but Patrick has attempted to solve uh in some key ways um so yes you're right he is definitely trying to say like if you framed it as a workaround for limitations I would frame it as really trying to deeply understand the technical makeup of the tool and then use it within its grooves as best as possible. >> Right?

And the reason why I am so fascinated about this is because we are doing the same thing. I've gone down this I've hit up against the exact same limitation and solved it in an entirely different uh and better way which which I can explain. I would love to. >> One of the one of the issues that I I think you have like all these skills that trigger different parts of the text. So if you if Patrick's like this is going to be an objection handling script or something, it'll load all the information for objection handling and use that to sort of write the brief.

Is that sort of right? >> Yeah. Oh, that's one skill only. Uh so it's all >> right. But you have lots of skills. Yeah. Okay. All right. >> No, it's all it's all done through one. The skill is actually called Prometheus. So uh yeah, that's that's what Prometheus is. So yeah. Yeah. >> All right. So, >> now it it does that skill has a bunch of information for sure about sort of different categories of ads and our intention for the future.

In fact, a lot of where I tend to put my time in the next year is essentially in identifying it's it's like I'm a bird watcher and my goal is to go look and and identify different birds and describe them in nature. Except in this case, birds are ad formats and styles and like script styles and script formats, stuff like that. >> And I don't think I'll ever get to the end of all the birds. I think there's going to be more of them all the time.

But this analogy is going to break down soon. But the that's the basic idea is that my goal in the next year is to sort of go and say like what works well and then how do we sort of connect all of those up? >> And this is the exact reason why we need AI because there's so many birds. There's so many birds that exist. There's so many birds that you find in the attic account every week. And there's going to be so many more.

And the human brain cannot be an encyclopedia of those and call them up as easily as an AI can. So this is why this is sort of essential. And so I I'll go into a little bit what we built and how it's and how it's different and why yours is problematic. So >> um >> Okay. Yes. Please. Yeah. I I need you to tell me more of why it's dumb. Uh but yes. >> Yeah. So I started in Claude and then I moved to um N which is a platform where you can piece piece a lot of these different parts together and you can expose a lot of their settings and you can allow that to get around some of it and also improve the the quality of certain settings which are actually just kind of like locked up in clawed and not exposed.

But um the the gist of what we're trying to do is like we have a lot of information about what works on various clients and what works across clients and information about the clients like the offer, the bundles, the products. So the AI has what it needs to write all the particular scripts. What we don't have what we are not is very prescriptive as like what yours is like this is an objection handling and then go do that and then it triggers very specific things.

We have a system where we can put in a very rough idea like oh let's do something funny about X and then it will then search through all the files. It will chunk them up very specifically. It will rerank them to what's most important and then load those very specific pieces of information into the model essentially. And so it's it's very different than me being like this needs to be objection handling. We could do that as well.

And we and we have a document about like what proper objection handling is and all those different formats. But I think it's very important as we're developing this stuff not to be so prescriptive and allow the AI to search like this is what the AI is good at reasoning and doing this type of stuff. So the more you can push that reasoning into the AI itself, the better. We actually do some of that. So it probably is a little less prescriptive than you think.

So, one of the things we'll do particularly for like long form video scripts, which is like a format that we are very good at right now, like long form explainer videos. Um, and you know, you and I have talked about this before, Jess, but particularly for those, one of the things that we do is like we'll sort of say like here is the angle that we want to work on. Okay, let's just say health brand, new year, new you because that's the time of year we're in, right?

So, something like that. [clears throat] Here's the product. Here's the value the product provides. here is the the particular angle on that value we want to produce, right? You could say anytime that the product makes you um I don't know recover faster from your workouts, but you specifically want to put that into a new new angle. So let's we go through that and we said this is and so the first thing we do in our process is define the message.

That's the core thing, right? Which is you know message angle basically trying to talk about the same idea and then and then we'll go with work with Prometheus and say okay I want a long short and mid-length version of these. There's going to be different formats, different placements, different whatever. here's who the talent's going to be for this because we know that ahead of time. But then what I'll do is I'll say to I'll say to Prometheus, suggest to me the right framework for that, right?

So like one of our frameworks is like the time traveler warning framework which is like >> you know this is like in six months you'll be one of two people, right? Which is like perfect for new year new you. Um and so so Prometheus will say >> well you're doing new year you new new at the beginning of the year and you want three long form videos. You should try time traveler warning here. It's a perfect way to bring storytelling device together with the angle that you've suggested at this particular moment.

Now, I c like you said, we can be more prescriptive. Like, for example, I just think negative hook ads just work. I just try to prescribe some of those as part of the process, right? But then there are definitely times where I want Prometheus to do the thought work alongside me. And and of course, sometimes I'll outvote it and say, "Ah, that sucks." Like, I see why you got there, but whatever. But we'll do both. And I'm my hope is that the the more birds we identify that this becomes an even more useful part of the process. >> Yep.

And so another issue is like it to me having him describe it with like these 50, you know, dozens of 50line files. It just seems like this like huge filing cabinet. It just seems like a really complicated uh and a lot of work. And you bring in a new client and it's like I need a dozen more weird chunked files and stuff. That's not the case if you're having the AI do the type of searching and doing that stuff for you.

And so one of the big differences which I found when I rebuilt my system within N is you can change a lot of the settings within the model. You can change the temperature. You can change how much it chunks or the the reading depth and all these things do actually have an impact on the quality of the output. And the biggest variable is changing the model entirely. Changing from claude to to Gemini or whatever. And I've run my exact same system on multiple models and I've found like the model which is the best.

Do you want to know what that is? >> Yes. >> Why would I tell you that, Andrew? [laughter] >> Why would I? >> I have a guess. I have a guess about what it is. >> What is it? What you say? >> Gemini. >> No. Gemini was actually the worst of of of what I tested. >> Interesting. The only reason I guess that is because I I've just seen a lot of people talk about how Gemini is overrated. So I I haven't tried it specifically for or excuse me, underrated underrated.

Uh so uh so I haven't tried it specifically for this kind of writing task. Although what you're saying like the the actual Prometheus in its infancy was actually built basically as straight into Poppy. And this is one of the things I really liked about it is that you would like you would be able to just be like I don't like that script. What happens if I just go from claw to GPT, you know, or whatever? Like, so, uh, yeah.

Interesting. Are you going to tell me? Are you going to actually tell people or do they have to just get that sauce by hiring fire team, which firete is, by the way, link is in the show notes to go work with Jess and his team to get them to unleash all of this creative uh power and AI power on your on your brand. But yeah, tell me. >> It's actually um it's GBT5 uh. Pro is the most creative, the most natural sounding, the funniest.

Claude Opus is very good at sticking to instructions a little bit better than the GPT, but this is like a this is a scenario where if we're making these leaps and jumps within the technology and you are locked into one and like it's a race of whose brief writer is better than someone else, uh you're going to want to have that sort of flexibility be able to built in. >> I think that's a thoughtful objection to the task.

I think it's a good point and something I'm excited for Patrick to listen this point and see what he thinks. You've heard me talk for a long time about more staffing, how much I love working with people at more staffing to get me the very best Filipino talent working on my businesses. And that's what I've done for a while both in my uh e-commerce brand that I'm starting will be staffed basically entirely with more staffing people.

And then my agency also right now has people from more staffing and from sister businesses of theirs. Just a huge fan. The the value proposition is really simple. your dollars go really far attracting the very best talent from the Philippines in your business and you can get incredible people working on your business hiring at the top of the market in the Philippines versus at the bottom of the market in the US for the same cost.

So you just get really really good talent and more staffing is a great reliable agency to help you do that. Like you can get everything done, somebody on board within two weeks. They give you a one-year guarantee on new talent they hire. They coach and train people on top of the recruiting help they do. Just really really good getting you high quality resumes across every part of your e-commerce business. But actually, it goes even further than that now, and it's really cool.

They've actually now more staffing now added part as their service. Part of it is is a basically forecasting for you, giving you a 12-month road map of the actual cost structure of your business. Not just staffing, but the real cost structure of the opex of your business to help you think about where are you fat, where are you lean, and where can you be thinking more intelligently about how to apply good quality leverage across your business to get you real operational efficiency. with a combination of that forecast plus figuring out how to help you leverage AI and then incredible offshore talent in the Philippines.

They'll give you a road map to think about where your business can use a little more help, where it is cost effective and where it's not and really help you think about that. I get questions about benchmarks all the time. Am I spending too much here? Am I spending too much there? What they'll help you do is actually do exactly that. Like really give you some analysis of where that is, forecast the future and and apply with a combination of AI and offshoring those things to make your business leaner and more effective and [music] more efficient.

That is the promise of business right now and it's what e-commerce businesses should do. They should be making a huge impact on their P&L by running a really lean opex really effectively [music] and really being careful about their applying operational dollars. More staffing can help you do it. So go check it out right now. Staffing.co. Really get deep with them. Really look at the whole business. It's going to help you a lot.

Staffing.co. I have one thought rattling in my head and I think it's related to the reason that I joyfully bring you alongside AJF Growth with clients and I don't feel threatened by that at all. Why I like working on clients with you. It's related to that thought and it's related to this thought in a second. So I'm going to footnote that and say I'll come back to it in a minute to see if you had anything else you wanted to say about the architecture and all this stuff because I think by the way I I think this is actually extremely helpful information for people.

Jess, so I'm really glad you're I'm really glad you're you you're taking the time to yeah tell us about what we're done. It's useful for the others out there. >> Well, I'll just say for you and for Patrick, moving to an open platform like that has been like a 2x game changer and I'm I'm more than happy to like run Patrick through what what we're doing very specifically and stuff like that. But it definitely seems like the future is more flexibility um within these systems.

I think yeah I think it's it's a it's a really it's a really good thought because one of the problems that we have right now and this gets to the thing I was going to say earlier >> is that we get a lot of things that sound kind of the same over time >> and I think that's partly the deterministic structure of our approach I think there's some other reasons for that one of them is that we need to identify more birds but that is one of the concerns that I have is that one of the concerns that you would have with our approach >> it really depends on the documentation that is being loaded in there and that is that is like the secret sauce.

That's probably like 50% of the game for all these implementations is the quality of what you're loading in there. And if you I guess in the future if you just like download that off the internet, it's going to be the same as someone else's. But from from our experience, every client is marketably different in terms of what works and what what's formats are, you know, seem to be effective. And so you need to be building that in yourself inhouse.

And I I imagine ours would be very different than yours. So that might lead to some of the differences. >> Yeah. So it's funny you say that I this is this is actually why I didn't fear to to in the end to release the episode that I released because uh like Jacob Posel great AI person thinker uh who's done a lot in ecom is partnering with Shireen Also great ecom thinker delight of the internet one of my favorite follows on X. the two of them are partnering together on some kind of agency work and they haven't been super public about it but in fact I didn't even know about that he was working with her until Jacob just tweeted at me so I don't mind saying it because he did it publicly about it but you know Jacob is in my view almost certainly the most advanced like AI technical person that I know in e-commerce like he's gone very far down the true developer AI pathway and he understands it stuff extremely deeply I've had him on the podcast talking about it I don't think we're going to try and compete with him on that you know I don't think think we're going to try and like out technical Jacob, but I have 11 years in e-commerce, which Jacob doesn't.

And Patrick also has a whole bunch of years in e-commerce, 10 or 11 years or whatever, including building and selling a business, uh, a brand himself, all of those things. And so, what I believe is that the actual secret sauce of all of this stuff is actually in exactly what you just referred to, the context documents and the information that we're feeding it. And the quality of that information to me is the most important thing and is the thing that I think we can differentiate on.

I actually fully believe that there's some worldview and like human philosophy things that come into play here that are going to be more right or more wrong. What makes what motivates humans to make decisions and what do you think about people of different ages and different genders and all all of these different things is how you really think about >> um how you talk to them about products etc. and and what you want to say and what you don't.

And I, you know, that's that's all stuff that that the bet that I would make is that at the creative level that stuff's going to come through and now it's a game of like how do you get all of that into your context. In fact, the whole reason to date that I have not part of why I haven't tried to scale an agency is that a lot of those thoughts have been unscalable before. They like they would require so much work. Whereas my my goal in the next year is to do less actual client work, >> quote unquote actual client work, but like actually to spend much more of my time really carefully honing and refining our thinking that goes into our context for AI because that's actually the way to make the biggest impact on the most clients is to brush up my research to do that.

So anyway, I just completely agree with what you just said a second ago. That's where the actual magic is. And again, you saying yours will probably look different than mine. I bet you have insights that I don't have on a lot of these things and that's okay with me. like I think like you're going to do that and deliver it in formats that you're better at and all these kinds of stuff. >> And here here's the thing we have to remember as we're like arms racing to the most powerful script writer ever is that the script is just the start of the creative process.

You need excellent freaking uh creators, editors, like the whole thing. And so there's no one who's gonna get like this superhuman AI brief writer and then take over the industry if you don't have those other parts, you know, locked down as well. >> Yeah, I agree with that. Yeah, I'm curious to hear you talk more about if you're willing to say anything more about sort of what goes into your context like you said at the individual brand level or total level like how are you thinking about sort of the mix of broad principles that you believe in? >> Yeah. like let's just call it like advertising doctrinal statements you know like this is what works in direct response right and you have to have some point of view on those things to even make AI work for you right that versus sort of individual brand level okay this is different for a supplement brand versus an apparel brand versus whatever you know >> right so what we have is we have two documents for each client one document is uh general related to it's justformational about the brand and their products and if the AI needs to reference product ingredients or list them out or whatever.

It just needs that information. It can't find it on the web. That's not the point. So, that's one document. It has and and when you or when a client will update their create a new bundle, like we need to put that information in there. So, now it knows that the new bundles are the same thing, whatever. And then the other document is what works specifically for this client from our own creative testing. You know, there might be like 20 to 30 different rules that it can sort of draw on.

For example, we have one where it's like using the product while talking has been very effective. It's like a spray product. You spray in your mouth and you and you say something, spray and then you keep talking and it's great for keeping people engaged. It will sort of insert that into most of the briefs without us telling it to do that. Something like that. And then agencywide, we have a lot of learnings which are which we have found to relate to all clients pretty much universally.

Uh and it will sort of fold those in. And then we have very specific areas where there are we have a formats which has a format document which is a list of all the different types of formats like scroller or all these different some of the ones that that you named as well and and it will pull from those if it finds something that fits the idea that we're looking for. Um, we have a document that's just based on humor with examples of specific humor within DTOC because this information technically is in the model training data somewhere.

But having these context documents is like putting it front of mind uh and very specifically and that and that goes the same with like you know very specific you know aida and PAS and all those types of things which are technically the model already knows but having your take on those and putting that front and center in the context document is going to be important to focus it. >> Has this changed your volume output at all?

One thing I've known about you in the past is that you guys are beating the volume drum a little less than a lot of people. Actually, I don't want to put words in your mouth, so I don't I don't know what you think and what you used to think versus now about creative volume, but is this changing at all sort of your strategies around volume? Because I know you have a very high view of quality like just it really important to you that the ads are good.

Actually, I have a follow-up question, but let me leave it there. Has it has changed your approach to volume ad volume? >> Yeah, it does. Because at some points we do have to be a volume shop just because of the way our scopes of works are structured where if the client is spending more money on our ads, we are then making more ads and the cycle continues and then at some point you do need a system to create uh 20 ads a week or something like that.

We are not in the like ridiculous hundreds of ads a month per client scenario. Oh, I don't I'm not sure anyone. >> We We We are on some of those clients, but we are not providing those hundreds of ads. >> We are. That's definitely what we That's definitely what we're doing. Hundreds of ads per months for sure. Yeah. >> Yeah. >> For each client. >> Yeah. >> Uh Yeah. >> Okay. Wow. Yeah. >> Yeah. Anyway, there's a lot more you could say about that, >> but I I'm curious.

Why not? What is stopping you from doing that? Why don't you do it? I mean it sincerely. Why Why don't you do it? I think it's one of the reasons is a lot of clients come to us because they sort of have the lower stage volume game figured out. It's a little bit easier or their internal people can do it or whatever and they're coming to us for outside the box ideas or things they haven't seen before. And I think it's probably impossible to provide a hundreds examples of things you haven't seen before, you know, for each client per month. >> Why not do both?

Uh because it pro it's probably not impossible for you guys. This is why you're dumb, Jess. This is what I'm saying. Uh, you already told me why I'm dumb. No. Uh, why not? Like, here's an example. Like, you've made some ads for some clients that we work on together >> that strike me as so deeply replicable so fast that I don't know why the account isn't just like flooded with them alongside the bigger swing kind of things.

And that's certainly how we do it. We actually have now done something where we're trying to say like we we we have complexity waiting to all of our all of our ad creation. And the idea here is like there's going to be some low complexity ads that we're going to spin out at a whole lot of volume and they work like an old example this would be the post-it note ad like we're not really making a lot of those or anything like that this but the part of the point of that is you can make a lot of them really really fast.

That's part of the reason it works. It's really fast and cheap and nobody thinks that that should be your entire account all post note ads. It's just the idea that like >> it just there's just outsized value in creating a whole bunch of them, right? But alongside that, you know, there's like the three minute explainer video with like a whole bunch of custom animations and stuff like that which takes us a lot longer to make.

Well, we just try to do both and try to create a volume man quality output and think about them even even in terms of like who's working on them and and all that kind of stuff as sort of just >> slightly different tasks. Uh but yeah, I I'm curious why you don't think that way as well. Maybe it would be margin profile or something. Yeah. >> Well, it's interesting because actually some clients we will create something like uh a scroller like image scrolling text on screen like a long story of narrative very easy to replicate.

We'll find some winners. We will replicate. We'll do more of those and the clients will be like, "Thanks. We'll take it from here. We got that." And so, um, >> yeah, we we've replicated that style, that exact style in other accounts because we've seen you do it for that exact reason. It's like, "Oh, that was so easy to do so fast that we're just like, >> yeah." Yeah. And so I actually I love it when clients will copy us and then put that in the account because I think a really good relationship is a good like sparring partners where it's like you can you can copy us, see if you can do it better and then you do some better and it's like guess what?

We're going to try and frigin top that what you just did. And unless you have these two competitive forces in the account pushing each other, that's ideal. But if you're saying like >> don't do that, don't do the easy stuff, then I I think that's sort of a wasted opportunity on the client side. I think in general like we have we have a team of 13 people. Um I we just can't do a hundred. I'm not sure I know or want to scale the agency to 10x that type of of headcount.

So that's why I am exploring AI because I think the quantity side of things is easy to do. The quality side I'm I'm really trying to push the quality as much as possible, extract as much as possible out of there first and then maybe there's some quantity benefits once that's sort of locked in. >> Yeah. Yeah. And I I think that yeah, like I said, we're trying to sort of think about how to build both, how to build in a way that makes sense so that our margin works for both and and all those things.

And and to your point, like this is actually the thing I was going to say earlier is one reasons I'm sort of comfortable with the idea that we're going to exist in claude for a while is that we know Claude can do the kind of ads that we like to make that we are really good at. It's okay with me if we're sort of stuck there because it definitely can deliver those things and I would say excellently it can deliver those things.

Yeah. Obviously like all AI stuff, you know, there's a human in the process like making sure it's right and all that stuff. But one of the things I believe is that like uh and this gets to why I'm so comfortable partnering with you guys is like you guys just make stuff that I will never make. Like I just don't and I I think that's really good. And what I want us to do is say like there are certain kinds of ads that we are incredible at and we'll make a whole bunch of those and we can probably get you to a few million bucks and spend at least making that kind of stuff.

And then like you said, people bring you in when they're ready for the next stage of creativity or something like that, you know. And not that we can't push people past a few million in spend. We do. And like that's part of our process as well. But it is harder. You hit some kind of a ceiling, a local maximum or something like that. And so you have to figure out what what is the next stage. There's all kinds of answers to what the next stage is, but one of them is like more creative diversity.

And and one of the things I just think is that like sometimes brands think in terms of oh, I'm here. I need to shave my opex or replace this creative agency with another. And my response is don't do that. like if you like our work then keep us and then add Jess as well and look what'll happen. He'll make really different kinds of stuff that's that's an extension of your team and he'll deliver it at a really high quality and that's the way you actually win over time, you know, is like you create more and more of those and eventually you get to where like you're producing thousands of ads per month or whatever as your as your account is spending as you find good reliable partners who do very different kinds of things all working together. the same way you would then go seed influencers and get their content in there as well, which is not scripted content and etc.

You know, so to me it's like I'm really comfortable with the idea that we're going to get incredible at making a few ad types that we are especially awesome at. And right now that might mean I say like no to brands to work with that I don't think those ads are like really the best use of their time, you know, >> right? >> Yeah. >> Yeah. On the the local maximum stuff is very interesting because a lot of brands will be like, I can't get above this amount. uh maybe my I've reached my TAM or whatever and it's like I was like your audience is the amount of people who are on meta who are converting at your very specific CPA target like that's not the big TAM this is like the tiny little TAM that you are that you are working in and in order to get above that it's like you want to make a an impression on a a customer in real life and you have like a tie and you have a like a suit jacket and like the tie and suit jacket is the ad and you've done very good on it, but you've also you're wearing like sweatpants and Crocs and that's like the landing page and maybe the offer as well.

And so there's a lot of things that your CPA is a condensation of like the entire funnel including your brand all into that number. But a lot of people only look at it as sort of, you know, a creative or a media buying thing because it's really the the media buyers who are like tasked with dealing with it. >> Right. Right. you know, and so those are the tools that they have, but man, there's so many other options before swapping out a partner that you have. >> Yeah, it's funny.

I this this gets to like Taylor Holidayiday's perspective that like um they don't call themselves a growth agency and he doesn't like the title head of growth because as he says, growth is the CEO's job. And and it's getting at a very similar concept to what you're saying, which is like if you think it's the media buyer's job to grow your business, there is a reality earlier on in businesses, I think, where there's some truth to that, where they have a particular skill set that can be deployed at a specific stage of business.

But there is a point, and I watch this happen a lot with brands where they kind of they hit a ceiling of some kind, and the way they're going to break through that ceiling is going to be well, I mean, it could be all kinds of different things. And it it may be add another creative agency like get Jess to come alongside Andrew and get them in there so that you can add more reach in deer adap or something like that but but there is not like a tweak or media you know something like that.

It's like no maybe you do have to build a creative process. is I think of like Jordan Manard here where it's like our creative process outputs for us like >> thousands of ads all of the time at massive volume to hear him tell like I don't exactly understand it and that's the way they do it but partly because they just are like brute forcing a probably a pretty low rorowass on >> uh on a product with monster LTV and it's just the game is just like how much stuff can you get in there because there's outsized value in doing that you know that's one way to do it but there's also all kinds of other things like that I just know that at each stage the job gets a lot harder you know, as you go >> to each revenue growth stage, you know. >> Yeah.

Process process is an interesting well because people are people are trying to deploy the tools that they've already used to get to a certain stage and and they just don't even have the headsp space to think of those particular tools. >> Um, but one of the tools that like >> no one really talks about and I absolutely nerd out about is your internal process of getting things done. Um we have made absolutely huge changes this year on that and we actually do quantify it now but it definitely feels like our effectiveness is like 2x what we were um you know seven eight months ago. >> Can you talk more about that?

What what what are the core things that you changed? >> So we moved to a a project management system that no one has ever heard of called fibery. It's basically notion on steroids with a lot of AI sort of mixed into it. at every different stage. We looked at a way to because we're not a quantity shop per se, but we still need to output like a hundred new ideas and 100 new projects a week. And so, for example, I'll just give you a brief rundown of how an idea runs through our system.

So, we have two calls a week where we're discussing new ideas. We're discussing stuff that's working currently. And then the other call is where we're discussing inspiration that we see around. And from those we on the call we say like project name whatever and so it's in the transcript and then we take the transcript and we run it through a cloud project which will pull out the little bit like we call it a gist and it'll create a title for it and then it'll it'll pull out like 30 or 40 of those.

We'll take that list and we'll put it into Fibery AI and it will create projects for each of those things that have a template for the specific client with the gist and then that just goes onto the brief writer. And so that happens about hundred times a week. And so like before we looked at every like even cutting out button clicks at every different any if you do something a hundred times small little changes like that can make a huge difference.

And Fibery has allowed us it's so customizable. It has allowed us to just cut out like hundreds and hundreds of clicks and and you know hours and hours a week. It's pretty crazy. >> Patrick is going to salivate over that because that's the kind of thing he loves to dig into. And yeah, I it's really interesting. So you you think the main value there like is it about the reduction in button clicks or is it about or is it about like uh something else in the process the way it streamlines things like like not just a matter of clicks.

It's everything. We used to like create the projects manually in notion during the call and so we thought we were smart because a lot of times you like you'll brainstorm and then you'll get to the end of it and it's like oh what are the ideas we had and you end up with like three or something but we were like creating the projects on the call and we thought we were smart to do that and we'd create like 20 to 30 projects out of a call and now we're just like speaking them aloud and allowing the AI to do that's work and we're creating you know 50 to on a like a 60 60-minute call and like doing like a creating an idea per minute is pretty insane.

So >> um >> and so Okay. Yeah. So then repeat that back. So it so then it it then writes essentially briefs right away off of those ideas. >> It it writes the we call it a gist. It's like a pre-brief. It's like some quotes about the idea, what we were talking about, what we might have been referencing, and that information is given to the brief writer, which is myself or a few other people, and they take that and turn that into a full brief and script or run it through the AI, and it will do that.

Yeah, it's this is interesting. So much of the ability to generate profit for yourself and value for your clients, all those things is about the ability to turn ideas into action and >> and to not let ideas just go away like this right now. But this is a great example here. You and I are talking about this and I'm thinking like should we do that? If we do that, like wouldn't that make us better? Okay, but the the the danger is this dies.

My thought here dies and I never actually do anything about this, you know. Um, but this is a good idea that you're talking about and turning that into action would really help us and really help CL. So, yeah, it's it's interesting like I see what you're saying. Essentially, if you can increase your action to idea ratio, you know, which is essentially what you're describing, like you could just accomplish a lot more things.

And if you can do that without it also feeling like chaos, but actually it feels in some ways uh like a reduction in chaos at the same time, that's like it's glorious. >> Yeah. And so those those ideas will get into the turn into projects and then our account managers will look at the pipeline and then slot them in to where they need to go and then it's off to the races in terms of getting that stuff produced. >> Interesting.

We um we have a very different way we approach like which briefs we're going to work on than that. And we don't do a lot of like creative meetings in that respect, but I it's it would be interesting to do that, especially if you can continue to increase the creative output on the back end of all of that. You know, if if there's tools you could use to, >> you know, format styles and things like that that you could use.

This is one thing we're after all the time is like formats where we can test a message at a very low cost and very fast turnaround. uh like we love we love you know you mentioned the scrolling text ad scroller whatever like one of the reasons I like that format and integrated it into our work uh stolen from you is because of exactly that it's just so easy to make it fast that if you're like I want to just see if this message will play I know it's not going to be the best version of the ad I'm ever going to make I don't I don't think that it's just but it's like it's just sort of like an MVP version of it you know and then you can go and hire talent and like build long videos and build custom you know all that kind of stuff but so the idea that you could combine that kind of thinking, low cost, high volume output, fast turnarounds actually all the way down to the idea level with this kind of tools is really interesting to me.

I think I think it makes sense that that's like a great use of AI and actually it really reinforces the idea of AI as a leverage creator where it becomes a way to like hire more people not less, right? Because like it's like yeah uh like it suddenly is like now everybody becomes more valuable because they could because all more of their ideas become become actual actions. You know, >> you need help in your Google Ads account.

You need help not blowing money and you need help making money. And you could do both of those things with Zetto marketing. It's amazing how much I see brands blow money, just wasted ad dollars. I bet right now you watching or listening to this episode, are losing money in your ad account due to shoddy work in the Google Ads account, whether it's by you or an agency. I see it over and over and over again. Branded search terms slipping into non-brand campaigns and being really deceiving about money you're making.

Just spending too much on brand campaigns in general. no use of Google shopping in ways that still really work, still generate value, no use of YouTube, those kinds of things. There's all kinds of ways in which Google Ads accounts are massively underoptimized. It's tricky. It's hard to get right. And that's why you should work with Zetto Marketing, Zatomarketing.com, led by Kirk Williams, longtime Google Ads expert. Zetto Marketing is a boutique, Google Ads agency.

I think they have like six people on their team. A lot of them have been there for 10 years. like it's like a really really good hightouch Google ads agency that can do a great job making sure that they are in the details understand what's going on making you money not losing you money all at the same time if you just need reliable good quality Google ad support if you're spending any money on Google ads in your ad account just have them take a look have them give an audit I bet you there's stuff that you can be doing better I see it all the time from brands especially [music] like seven figure brands early 8 figureure brands it's just a really underoptimized channel and it can really matter you can spend material material money there, which means you could be losing material money or making it.

Go to zettoarketing.com, get a call going, fill out their intake form, see if they can be some help to you. We partner with Zato on a bunch of clients, like just it's just my go-to reference all the time. So, uh, so they think similar to the way I think, big fan of theirs, highintegrity people, zatomarketing.com. Go check it out. You also suggested to me that you have a take right now of the importance of the non-media buying account role at an agency.

Can you tell people what you mean by that and why you're hot on this idea right now? >> Yeah, so the the account person, so if you're talking about like a media buying agency, the account person is generally referred to as someone who does the media buying. We don't really do media buying but we still have account people and their job is to be client-f facing for the most part and understand the client's needs and their priorities because these things change all the time.

I think when you first start with a client, you can make a bunch of like hot creative and like let's maybe you know maybe it works, maybe it doesn't or whatever, but as you start working with a client, their priorities shift and they're dropping new product all the time. And if you don't have someone who is getting that information from the source and bringing it back to the agency, it's just going to you're going to be like a month late to every product drop. you're gonna see stuff in the account that was working and then you're gonna iterate on it, but like it's going to be like two months down the road like it's just gonna lengthen the iteration time of everything going on.

And when you hire an agency, I think they are a very potentially effective resource, but you have to like use them properly. And so if it's on the media buyer to like come up with ideas and prerogatives on like a monthly call, like the the iteration time on that is like in measured in quarters, which is insane, >> right? >> And so um like on clients where who spend over a million dollars a month, our account people sit on their internal meetings with their internal people and absorb the information because like codes are freaking changing all the time and stuff.

And so I'm not sure if this is the case at other creative agencies. I don't I don't work there. I don't run those. But it's definitely not something I've really seen talked about a lot. >> Yeah. I think what I hear from that is like so I think you're right that there is a really this is actually very similar to the thing we were just talking about with like sort of idea to action or action to idea ratio. Like it's basically the same problem which is that things are happening and somebody's not finding out about it.

And so there's like ideas and initiatives happening, but they're not turning into actions and they're not turning into actions fast enough and therefore like somebody's left out. And what it makes me think about is another problem with like sort of the process of how work gets done and and the local maximum problem and the scaling problem is like for growing e-commerce brands like let's call them like early mid eight figures trying to get to like true mid eight figures, right?

Call it like 15ish trying to get to 50 something like that. That is, I think, a lot of what like a true CMO's role is is figuring out how to leverage a bunch of high quality resources and get them all do making the best use of their time and money and like sort of putting them on the right things because >> like what the agency has to do as you say is like have somebody who's responsible for trickling that information to the to the the team and then what the the team has to do is figure out how to make sure that they are actually sort of using all those people correctly and even compiling the information well enough that that that's even possible for the agency to accomplish, you know.

Um, but it is a real sticking point because it's a very hard skill and particularly for I think if you come from a sort of more individual contributor background and become a CMO or something like that like it becomes really hard I think to uh to to do this very well because it's just a very different skill set. >> Yeah. And you can, it may seem logical that like if you have good performance, you'll the account person will have a good relationship with the brand, but I swear it's absolutely the other way around.

Like if you have a good relationship with your point of contact, good performance will flow from that. >> Yeah. Um, I know for like we were working with the Wooles and their director of paid social went on maternity leave and like we the fire team subbed in for that role because we were just so integrated and so like that is a level where you're running the the resources that at your disposal at 100%. And like if you have a resource that can run at 100% and do very well, you definitely want to push that as much as possible.

Yeah, it's really good. Uh, tell me Jess about some kind of win you've had recently. Just tell me something some way in which you guys have done something good that you can talk about publicly. An ad that popped off and took over an account, a client that just smashed. Like, give people one really awesome fire team victory from from recent times. >> So, I'll I'll go back to the scroller because I love that ad. And you've seen some versions of it, but we have we've been deploying it on other accounts and some larger accounts.

And you would not believe the level of iterations we have done on this scroller. Every like we've done like it's a like a Facebook comment, you know, that's sort of scrolling down or we've done with like voice over and like we have like split the atom on this particular type of creative to find every possible um permutation. I find that so fun and I find that so fascinating. But again, it's it's not something that's going to happen if I am getting notes from the media buyer on a monthly cadence about what's working, you know. >> Yeah, it's good.

Uh, okay, Jess, tell people where to go work with you and tell tell people who should go work with you like what what is the ideal fire team client and uh and then where does that ideal fire fire team client go to go to work with you? >> So, firete.is is our website. Everyone loves it. Says it's says it's hilarious. I would say our ideal client >> it is it is in fact a great website. Yeah. >> Oh, thank you. >> Yeah. >> Um I would [clears throat] say our ideal client first and foremost would be really solid good A+ people and operators like no no um >> glorious >> weird drop shipping nonsense or whatever.

You can absolutely do great work with with other great people. In terms of like general client size, I'd probably say like 200k a month is the floor of where we start. Certainly like >> 200k in revenue >> in ad spend. >> Okay. An ad spend. Yeah. Thanks. >> Yeah. Um certainly if there's a lot of promise there, we love to work with brands. We have like three year-long relationships growing brands um much larger. And then and then brands who you know get up to like 800K or over in a million a month.

I think we're uniquely adept to filling a lot of holes and breaking through a lot of plateaus that seem to plague those particular brands as well. >> Yeah. Yeah. Great. Okay. Awesome. Go check that out. I have I have worked with Jess and his team on multiple accounts as we said. They've delivered great work uh every time. And so go to firete.is to get started there. Don't forget to follow uh Jess on X as well. Hiref Fireteam.

Great follow. One of my favorite follows on X. Uh good chance that you see him call me dumb there as well at some point. Uh so uh so if that's if that's if you're into that then then that should be good. Uh Jess, thanks for coming on, man. Well, go ahead. What were you say? >> Just kudos for you for putting it out there because this is how in in regards to like Prometheus because this is how we all get better, you know, and a lot of people keep their little secrets, but uh if you're sharing your secrets, it's only going to improve. >> That's the hope.

All right, man. Thanks. Appreciate it. >> Thank you. Peace. Thanks for watching or for listening. If this episode was helpful to you, you should do two things. You should subscribe first of all and you should send it to somebody for whom you also think it will be helpful. And uh I would love to also hear from you. Podcastfrowth.com is the place to email me. [music] If you are interested in working with me and the AF Growth team, we're taking on just a few new clients in the new year.

We're going to be a little selective about it, but if you're the right fit for us, we'll know that by you telling us a little bit about your brand, by filling out the intake form on my website and just letting us know. Afgrowth.com place to do that. Tell me a little bit about your business and we'll see. Maybe I can poke around your ad account, give you some thoughts. Even if you don't want to end up working with us, it might be helpful to you.

Big thanks to my sponsors for this episode. That is more staffing. Go to more staffing.co to get a conversation with going with them. And Zo Marketing, I love ZTO Marketing. Um, big fan of them. So, zettoarketing.com to get your Google Ads needs met there. Thanks again for watching, for listening. I'll see you next time.