How the Prophit Engine Actually Works (Full Walkthrough)

[00:00:00] Speaker: Alright let's go ahead and get rolling and we'll share this. So my goal today for those of you that are joining you, appreciate having you is to give you a sense of how this claim that we're making, that I

[00:00:11] Speaker 3: recognize is a bold claim is sort of possible.

[00:00:15] I. We're gonna give you specific examples of in practice why we think we're living in a world now, where one of our people, this role that we're referring to as the profit engineer has the capability to deliver the work and the outcome efficacy of three people. And how that plays into not just is it about the savings that it offers brands in terms of labor costs, software costs, et cetera, but why we actually think it produces better results and why it is a better way of working than having more people.

[00:00:48] Richard Gaffin: Today's episode of the e commerce playbook is brought to you by charge flow. The world's first automated chargeback recovery and prevention platform. Over 15, 000 brands have already stopped handling chargebacks, leaving it to charge flows platform to win chargebacks for them on autopilot. Q1 is the peak season for chargebacks with first time customers and so called friendly fraudsters who came from Q4 sales.

[00:01:11] They're starting to submit disputes. That's where charge flow comes in. The best part. You only pay them when they successfully win your chargebacks. If not, you won't pay a dime. So here's an exclusive offer for our listeners. Get your first 100 chargebacks handled. Absolutely free. Just download their app by going to charge flow.

[00:01:29] io and enter code CTC 100 after activation that's charge flow. io code CTC 100.

[00:01:41] Speaker 3: And so my goal today is to go through. The setup that enables this to occur and then some of the tooling and efficacy that goes into it. And Luke's gonna share a little bit of some practical examples of it coming to life in terms of what we're seeing be possible in this combination of what is really sort of a technology layer that begins with the database side that has on top of it a set of context related to CTC methodology over many years combined with what AI is in enabling us to be able to do quickly.

[00:02:11] And then the sort of human operator that sits on top of that. And so, I'm gonna begin with what I think is the primary enabler of great decision making and progress towards outcome, which relates to clarity. One of the biggest problems I watch organizations with, especially with large teams, struggle with.

[00:02:33] Is clarity of where we presently stand. And therefore, in light of that present state, what is the right insight or action to take now to close the gap or to improve upon the existing outcome? And that sounds like a really simple idea. But there are a million complexities that get introduced into that process.

[00:02:53] And the more humans that are involved in it, the more complex it becomes. And so what we've spent so much energy around as an organization is this sequence of beginning by building good data structure. So one of the things that we care a lot about is have we absorbed all of this connection between the marketing layer of the data?

[00:03:17] So think of that as the traditional meta Google, TikTok, Snapchat, Pinterest, all the APIs spend, whether you brand as a brand, have out of home podcast affiliate, like everything that exists in the cost structure of the spend side, overlaid with what we would call the customer data. So the order level data out of your platform.

[00:03:35] So all the order history, customer history, new versus returning customer definitions. If you've migrated once, if you have two different CDPs, all the complexity that gets introduced at that layer. And then into the finance portion of it where we have clear and accurate cost of goods, cost of delivery to be able to understand and see where we sit on a contribution margin level.

[00:03:56] How that ladders into a full p and l level expectation of the organization. And then that gets sort of decomposed into the daily expectation based on your marketing calendar and actions. And so there's this giant sphere of information that has to coalesce into one single point. That is a endeavor that for most brands, the actual capacity to make AI or any humans effective depends on the quality of that underlying infrastructure.

[00:04:23] And if I could offer anything, whether it's CTC or anybody else, I would offer you that you don't spend enough of that time on the organization of the databasing of your information as an organization in order to enable fast movement and effective team function. And it's like a problem. Brian, I'm gonna mute you just 'cause I'm getting feedback.

[00:04:41] There we go. Th this is like an underrated portion of the problem, and everybody wants to move fast. Anytime someone hires an agency, it's always like, how quickly can you get up and running? But in many ways it's the wrong question. It's what would enable us to move fast forever? Like that's, that's the actual question that we have to go after, is what would allow us to be fast forever?

[00:04:59] Not how fast can we start today? And so much of that begins with the clarity of the data. So once that exists, we spend a lot of time in the planning process. So planning for us is a process of building a financial forecast that connects to the marketing organization in a way that allows us to understand what we're trying to do every single day.

[00:05:17] So planning for us begins in the marketing calendar. So I'm giving you this view as an illustration. I'll compact this just so you can see what I'm talking about. This is a brand's entire email send schedule for a month. So every blue line is an email. Green is SMS. Yellow lines represent marketing moments that are impacted more broadly.

[00:05:36] So they show up on the website, they show up in other places. So you can see the beginning of how we're going to generate revenue this month together as a group begins here. And this is why I fundamentally believe that marketing has to control forecasting, at least at the revenue layer. Finance in my mind, controls financial forecasting as it relates to cash flow, but the revenue generation is the building block of the actions that we're gonna take as a group.

[00:05:58] And so the planning process has to begin here. And when I think about this data infrastructure for brands we have 200 customers. I have one customer, literally out of 200, one for 200 that has a day by day marketing calendar built for a full year, one out of 200. And that sort of illustrates this sort of gap that I think often occurs for why forecasting is so hard is because the financial numbers exist.

[00:06:22] Before the actions that we're gonna take to create those financial realities. And that's backwards. What actually has to start with is what actions are we gonna take as an organization and therefore in light of those actions, what are the financial realities that they will produce? And so that movement is one of the first things that we're trying to introduce is that, like we call it units of growth.

[00:06:39] The way you make e-commerce is not software. There's no recurring contracts, there's no guaranteed money. Next month, every month we have to go out and do things in order to generate revenue. And if we don't do those things, there will be no revenue. You'd have to send that launch ads, you have to deploy emails, you have to post on organic social, you have to do those PR hits, you have to launch that new product, whatever the underlying set of actions are.

[00:07:01] We have to start to be guil then build this organizational idea that money in the future is a result of actions that we will plan and take. And so the marketing calendar for us is a center point to that. And it's not just for the sake of getting the qualitative expectations down. It's because those moments form the foundation of something we call the event effect model, where we can actually now take all of your historical marketing actions, so everything you've done in the past, and we can ask ourselves what did they do?

[00:07:26] What was the effect of those actions on the future? Or in the past on your revenue, how much existing customer revenue came out of that sale you ran? How much new customer revenue efficiency was affected by the launch of that new product? Whatever it is. And then we can use that to say, okay, as we plan events in the future, how does that affect our revenue?

[00:07:44] So that's one piece of it. Then when that exists, we're able to create sort of this ongoing relationship between your spend and efficiency as a business over time. So we can look at let's. Sorry, let me change

[00:07:58] Speaker: the delta here. 'cause this number is crazy. I'm just gonna go out and do a future, but there we go.

[00:08:04] Speaker 7: Go, go full screen with that.

[00:08:07] Speaker: Okay. Sorry.

[00:08:14] Speaker 3: Oops. Too little. There we go. So the, the next, once we have that sort of calendar piece, now we're gonna answer two questions about the revenue build on exist on new customer revenue. So we're gonna absorb all of your historical ad spend. Every one of these dots on the line represent some spend in efficiency over the history of this business.

[00:08:31] So you can see spend on the xxi efficiency of new customer acquisition. On the Y axis, we're gonna build a relationship between these lines over time to think about, okay, if we wanna spend X, the efficiency will be y, we'll have a seasonality coefficient for that. We'll have an a OV con configuration for that.

[00:08:47] And then we'll work to determine the monthly budget and then the corresponding new customer revenue. So now we have a calendar, we have a monthly budget. We have an expectation of new customer revenue. In light of that. We're then gonna build a returning customer model. This is just gonna be a cohort specific LTV model.

[00:09:03] So the performance of your customers over time, relative to the expectations of the things that are going to occur. All of that gets broken down. And I'm not gonna spend a ton of time into the minutiae of all that. I could go hours into the models and everything and how it happens. But for today's call, I just wanna illustrate that the goal is to get down to this, which is like every day an expectation of every dollar in every channel laddering up to the financial goal of the organization for that period of time.

[00:09:30] Okay? That's, that's what we're after. And the reason this is so important is now I can put together a set of expectations around the month that includes the marketing actions I'm going to take, the financial expectations I, I, I expect to produce, to get to the financial goal that I have as an organization to create the most important part of modeling and forecast.

[00:09:55] Richard Gaffin: If you're a DTC founder who's sick of losing chargebacks, then you'll love Chargeflow. Their AI powered system recovers chargebacks on autopilot, and the best part is that you only pay when they win your disputes. If not, you won't pay a dime. Ready to add bottom line revenue on autopilot? Get your first 100 chargebacks handled absolutely free.

[00:10:17] Just visit chargeflow. io and enter the code CTC100 after activation.

[00:10:24] Speaker 3: Clarity of where I'm wrong. Because the whole point of all of that exercise is that on day one, which let's, looking at the example here, you can see that on March 1st, I had an expectation of sales that was gonna be a hundred, $2,000. I came in at 93, I had an expectation of spend at 26. I came in higher than that.

[00:10:43] And so I'm off course from day one and in every model I've ever built for every brand, for every day, I've never gotten day one exactly right to the dollar, because that's not the point. The point is, can you enable the person in charge of delivering this outcome to know where we have a deficit relative to what we thought was going to occur so that I can quickly change my actions?

[00:11:04] And this is the entire power of the game, is that it is about the ability to course correct quickly that enables you to get to where you're trying to go. Think of it like your Google Maps when you set a map, if I turn off course on right, it immediately builds me a new map back to my destination. The destination remains content.

[00:11:22] The the constant, the path changes all the time. So let's use this example here where one of our growth, our profit engineers is managing this business towards the goal, okay? And you can see in the beginning of the month the, there was a promotion or moment that didn't go quite as well. Now 93 against 1 0 2, still pretty dang good.

[00:11:38] That's not a huge deficit. But if we look at contribution margin, you can see these first six or seven days of the month. Contribution margin was a little light to target, right? But I can go from, alright, now here I sit on March 11th. I can see that I'm 3% ahead of contribution margin. 1% ahead of sales have had to spend a little more to get there.

[00:11:56] So my MER is behind. Now, we don't believe that MER should ever be the governor because the ultimate thing I care about is this financial reality because contribution margin, which we would define as net sales, minus shipping, minus cogs, minus all fulfillment expenses, this is what covers opex. This is ultimately what shows up in your bank account and generates ebitda.

[00:12:12] So this is what I'm driving towards. We call this the scoreboard. It's a hierarchy of metrics. The reason this one's in the top and the biggest is because this is the game that we have to win above all else. Now, it's not to say we don't also care about revenue, of course we do, but this supersedes everything.

[00:12:26] If I have to spend a little more to get to this contribution margin, then so be it. But from here, I have things to do today to resolve still where I want to get in the future, and we're always playing multiple games at the same time, which is that it's not sufficient to say that I'm ahead of contribution margin because I also want to make sure that I'm not, we would call this squeezing the sponge.

[00:12:44] In other words, I'm not getting all of my margin out of the existing customer base. I'm also ensuring the future health of the organization by making sure my new customer revenue is on path. So even though I'm winning the overall game for the month, I can very quickly see there is a job to be done and solved.

[00:12:59] So very quickly, we train our people to isolate. We call it the hunt for red October, right? It's like, where is red and what problem then in light of that, in actions do I need to take today? So new revenue has a deficit, 10%. We're a little bit behind on efficiency of that new customer acquisition. Now that's okay because we have a buffer, because returning revenue is coming in higher than we anticipated.

[00:13:19] Now I've moved down to the layer of the actual efficacy of the media, and now's when I can start to take individualized actions. And there's a couple of really important things here that make this easy and fast to do. Okay? So in the quagmire of information, one of the things that slows down decision making the most is trying to make channel allocation decisions relative to ROAS numbers that aren't normalized against one another.

[00:13:41] Okay? So in our world, we work off of IR oas. So incremental return on ad spend as the normalizing factor to compare the channels. So we do that when you begin with us through a set of benchmarks that are the aggregate of many tests that we've run. But then we start a measurement program for each individual customer to get to the incremental impact of every individual channel.

[00:14:01] And then we use an i factor to compare the platform reported result against the actual incremental return. So what that means is, as you'll see, I'm able to compare everything from Google, non-brand spend to Google brand spend to Facebook acquisition spend to Facebook retention spend on a normalized scale relative to the incremental impact of each channel and where we're coming in reported against that.

[00:14:24] So this allows me to go, okay, as an if I'm gonna allocate across these channels, where am I? Where's there opportunity? Or where am I po potentially overspending? And you guys who probably don't even like you don't work in our system, could probably look at this and go like, oh, well it looks like we're might be overspending on Facebook retention spend and underspending on Google IR os non-brand here relative to the results.

[00:14:45] Facebook, we should probably go tighten up and see where that opportunity is 'cause we're spending to the total target, but we're behind. So like the next set of actions just reveal themselves very quickly as we go. And then we can break this down. Like email, I have an, an expectation of email revenue as we go.

[00:15:02] So I can see every piece of this. And then I can even go further down into the individual day layer of the individual campaign. Now, I'm not gonna click into our, what we call our Facebook trackers right now for this brand because it would give away the data. So I wanna be a little bit protective here, but I'm gonna show you in a second how I can go all the way from the daily expectation of the channel level down to the daily expectation of every individual campaign in the account.

[00:15:27] So the key here is one person looking at that can get to a set of decisions very quickly about what we're trying to accomplish. In addition, the tooling that they, they have at their disposal. You may see this YouTube one here, which is like, Hey, that one feels like the furthest off. But what I know is that we're actually running this is a ch launch of a new channel that we're running an incrementality test against YouTube for them for this period of time.

[00:15:49] And that's all at their disposal. If you go to the study section here, what you'll see is that an individual profit engineer can set up, launch execute, measure, and implement geo holdout studies entirely themselves. So this is something that, you know, if you go and try and hire a house or a measure, you're paying $10,000 a month exclusively just for the measurement process of setting up geo holdout studies.

[00:16:10] That all happens directly within our system where the test recommendations are designed for the regions that are held out. The people can implement the holdouts in the individual channel, the validation and results come back to them to give them, and then they can immediately deploy them as the benchmark settings in their accounts.

[00:16:28] So all of this is how like, one person can set up a test, design the test, execute the test, implement the results, see how it affects the immediacy of the outcome and off, and, and they're off and running. So this is the sort of foundational idea, is that clarity breeds insight and action fast. Okay? And every one of these calendar moments, every email that's planned, every moment, every one of these emails, and I'm gonna show you an email tracker a little later, has an expectation of revenue.

[00:16:56] Every campaign and every channel has an expectation of revenue so that this person can very quickly see where they're at. So the other piece then that becomes really important, I'm gonna bounce to a different brand here because again, I'm trying to protect a little bit of the data side, is that there's a creative function.

[00:17:11] So if we think about these three roles that we say that this person can encapsulate, we would say it's your head of growth, sort of the person that does the financial plan, media allocation, management of day-to-day decisions. They're sort of the orchestra conductor across all of your online revenue. And then the second is the creative strategy.

[00:17:26] Now in, in our world, when I think about creative strategy. The core job that I think is missing from what people define is really what I would call creative supply chain is that every month we need to answer a question of what, how much do we need to produce, focusing on what and why, and where's it gonna come from and when it's gonna go live.

[00:17:46] There's an entire orchestra conduction around that workflow that often involves many different producers. For most brands, the best brands we see are actually using three or four different sources for creative, whether that's a creator program, an influencer program, an agency, an in-house team, customers.

[00:18:02] Sometimes there's this like network of creative that's coming back into the ad account. Well, what we've built is something we call the creative demand plan, and what this does is it takes your financial plan every month. Okay? So once you build that forecast. It's then going to look at your historical media performance, and we're gonna talk about the metrics that it uses to do that, to define how many ads do you need to make this month in order to accomplish that financial goal.

[00:18:30] And then it takes that number of ads and it breaks it down into how many videos do you need to make? How many images do you need to bake? What products should you focus on, and what marketing moments should you connect it to? And then depending on what producers you have in your system, so who are you sourcing them through from?

[00:18:47] How many are gonna come from the client? How many are gonna come from CTCs, UGC? If there was a third party agency, we could add another one. Let's call this like the creative shop, right? Like whoever's contributing to the ecosystem, we can divide up the responsibility of the specific ad expectations by moment by campaign.

[00:19:06] And this builds an entire plan where for the month I can see. For Evergreen, I can see what, how many forecasted ads, who's making it, whether they've been delivered or not, where we're at this month relative to the expectation. And it sets up everybody from the start to know exactly what the job is to deliver.

[00:19:23] And I, as the creative strategist or profit engineer, know exactly who I'm supposed to be getting ads for, when and whether they're there or not, and what products that they should be focused on. And the way that this works is it analyzes these five metrics to determine how many ads you need to create in order to get where you're trying to go.

[00:19:40] So you can see that it's connected to this scenario plan. So when we build the financial plan for the organization, how much money I'm trying to spend on meta, it generates this output based on these five things. One is the first metric is what we call zero revenue rate. So this is the percentage of active ads that never converted during the measurement period.

[00:19:58] High values indicate many ads aren't effective. So what happens a lot of times is brands have lots of ads running in the account, and what we can see is how many of them get zero spent. So this is sort of this classic thing that we all know occurs, is that eventually media spend consolidates into the best performing assets.

[00:20:14] That's a good thing. We want that to happen. But the signal of this tells us how many ads, if we launched a hundred of them, how many could we expect that the spend would concentrate around? And you can see that this percentile shows where this brand, this is Nike strength, they have a really good creative score because a lot of their ads do get spend.

[00:20:32] And you can see that only 38% of ads don't get spend. That's actually like really high. Oftentimes it's more than that. Then there's this next metric, which is ad concentration. This is also a metric that Meta uses a lot in their own creative reporting, which is the percentage of total spend concentrated in your top five ads.

[00:20:47] So when we think about how many ads we need to create in the future, part of that is portfolio risk. How much concentration of ad spend do you have into a small subset of ads tells me how much risk there could be that if you lost one of those ads, you would lose a large chunk of the spend. And so you can see that only 22% of the spend for this customer is concentrated on their top five ads.

[00:21:07] That's really low concentration. That's pretty dang good. We want a diverse spread. Like the best case scenario is you have lots of ads generating lots of spend, so that at any one moment, if you lost that ad, you wouldn't lose the account, right? So really important thing to be aware of because just like your own portfolio of investing, if you have all your money in one stock and that stock collapses, the whole thing's gonna collapse.

[00:21:28] The second, the next metric is what we call ROAS degradation, which this illustrates the change in ROAS after an initial week of launch. So when you launch an ad, do you usually see that it increases in performance over time or it decreases in performance over time? This brand, one of the best, sees a 58% increase in spend after the initial week of launch.

[00:21:47] Okay, well, why is that important? If we're gonna model the efficacy of every creative that we launch, to think about how that campaign's gonna perform in the future? Well, if we know that every time you launch an ad, it actually starts out good, and then it collapses, then that needs to be baked into the modeling of the performance of the channel.

[00:22:03] If in this case, we see that as you actually spend optimizes over time, it tends to improve in performance. That also gives us something to model. And then the last metric is what we call evergreen share. This is the percentage of ads that have been running consistently for 30 plus days. Right? And Nike is the best in our entire portfolio at this.

[00:22:22] So 61% of the ads live in the account have been live and spending for more than 30 days. That's the, that's the single best performance of any brand. And so what does that illustrate? Well, the higher that you can get this number. The less job there is of replacement for ads the next month. So the creative supply chain needs become less, right?

[00:22:41] What is actually true for most brands is that very few ads run for longer than 30 days. And so imagine this metric is at 15%. 15% of ads have actually had duration and durability. What that signals is you have to do a ton of ads. And so the creative needs of every organization relative to their own ambitions are different.

[00:23:00] And people, this is like a, a, a concept that most brands are wholly unaware of. Most brands come to us with some pattern that is like, every Friday we launch 10 ads. It's like, well, why 10? It's like, well, that's what our system is currently just designed to do. Is that sufficient for where you want to go?

[00:23:14] I don't know. And the problem with that becomes that a media buyer stuck inside of that system only has one lever to solve their problems. It's budget up, budget down, and they have no choice but the scale on the assets that they have available to them. And it tends to leave to over performance on bad ads.

[00:23:30] So this, this is where this idea that Nike only needs 58 new ads this month to get to their spend goal comes from, is that they have really good evergreen share. They have a great base of existing assets. So 58 ads. I don't know if that sounds like a lot, like raise your hand if that sounds like a lot.

[00:23:45] 58 ads to produce in a month. Okay, Hannah. Okay. What I'll tell you is that's like one of the lowest numbers that we see. Like most of the time. In order for people to get to where they wanna go, it's north of a hundred plus ads a month. Is the actual expectation based on what the data, if they wanted to ensure that they had a chance of achieving their objective.

[00:24:02] So that is like 58 is like pretty, pretty light work for, for what we would expect for most of the brands to deliver. But this is all set up now. Again, I'm sitting at the center of this. I'm a profit engineer. This is all there for me. I can share this with the customer. I know who's deploying everything.

[00:24:18] And now the next thing we had to figure out was like, okay, well, but how does that same person actually manage the ad account? And when we built when we started looking at like the biggest hindrances to people doing work, we found that over 50% of a media buyers time in our ecosystem was spent building ads.

[00:24:36] Like the literal trafficking of an asset to uploading the account, to setting up the campaign structure. That whole workflow, one, like asset trafficking, is like this hidden nightmare in our industry where every one of you run a different creative process in terms of where does an asset go. Once it gets created, it goes into a separate process.

[00:24:55] It's named differently. It exists here. There are folders that are sometimes this way, sometimes that way, like it's all over the place. So what we created is what we call push to build. So what this enables is if I'm one person, I don't have time to go build all of these individual ads into an ad account there's no way that that would be sustainable.

[00:25:13] It would completely break, and it's actually the lowest value task that you could get anybody to do. And so what we created is we, for every customer that comes in this is now a different brand. I'm looking at Sunday Red here. So they come in and when their team has an ad completed, they just drop the folder link here.

[00:25:29] They come in, whether it's a Google Drive, a Dropbox Link, they tell us the go live date that they want, if it's connected to a marketing moment any specific text that they want and they just hit submit and they can upload 50 ads at once through one Dropbox folder if they want. All of those then get created instantaneously for me to the profit engineer based on the outline of all those subheaders.

[00:25:51] And I can just simply click, deploy, add. This gives me a little opportunity to make sure that I'm confident in doing that. I click to deploy the ad and instantaneously the ad will. Be live in the ad account. So I'm just gonna show you so you can see. So there's the ad Right now, they're all pushed at, paused.

[00:26:11] So we we're getting more and more confident in sort of the right capabilities to feel like we could just publish and go live and track that. But this ad right here is the one I deployed. It gets published and shows up in the ad account in the right spot. And now I can just turn that on. So again, this whole idea that like what makes a person be able to do the work is this idea that there's a set of tooling, whether it's deploying a incrementality test, whether it's launching a Facebook ad, whether it's building a financial forecast that is predicated on a series of longstanding methodology for how we do things, the technology to make it possible, and then the ability to go from I as the individual know what the gap is what we saw in these cases was that the general workflow would be if there's multiple individuals, is I, as the head of growth, recognize a deficit. I go, okay, there's a gap here to some result. And I go off and I send a message, Hey, media buyer, we're behind on meta by 14%.

[00:27:11] Can you take a look and let me know what you think? And then, because that's a human that isn't just sitting waiting for your Slack message, 38 minutes later they respond and go, Hey, I see that this campaign is behind. I'm waiting on two ads. Can you check when creative will have it? And they send a message to the creative strategist and they say, Hey when is those next set of ads gonna be ready?

[00:27:32] They respond, I think by Thursday we'll have them. And you can watch this. You can like go on your own Slack channels and just look at the gap between message. Go look at the email deficits and watch it play out. Like, and this isn't 'cause your team's bad or broken. It's 'cause this is how human communication works.

[00:27:45] It's not, it's instantaneous. It's this slow translation of information from the person who first found the insight to the actual person who has the capacity to execute it. And this is what like we think teams are. It's like someone ascends to this role of manager or leader and then they get to the responsibility of determining what we're all going to do and then passing away the responsibility to the execution to then be reported back to, to then make more decisions to then pass back the sets of actions to then receive back the information.

[00:28:16] And I'll just tell you that world is gone and a lot of people think that. The way to think about this is that like, oh, the task doers get faster. But what I'll tell you we actually believe is that I want the smartest person, the highest level person in your organization to actually be manned with the capacity for execution.

[00:28:31] I don't wanna try and bring strategy down to new people. I wanna bring execution up to smart people. And so the, the, the question is, how do you get the highest leverage person in your organization? This is why you see the Toby Luki of the world are coding now. It's because the, the barrier before was, it would've take, going to build in the ad account would be a waste of my time as a head of growth.

[00:28:52] That would be a bad use if it took me four hours to build in the ad account. But if it takes me four seconds, then there's no reason I shouldn't go from instantaneously recognizing a problem to instantaneously executing that idea. 'cause any lag between that is just wasted loss. So, that's like an example of every day what our people are doing is they're just hunting opportunities and we have other views which which like show them their portfolio mix, where every brand they're working on is behind or ahead, what the, and then AI recommended actions for exactly what they should do in light of that.

[00:29:23] All of those different things exist. And then there's this whole other layer, which I'm gonna hand to Luke now to talk about, which is that inside of our stats interface is just one example of the workflow. But the real power of what we're starting to see in this moment with AI is that the underlying database where all this information lives.

[00:29:41] So let's say we, we've absorbed all of your marketing data, financial data, order history into a single database instance. Then we do some transformation. So now there's all this forecast data that includes context of expectation of every individual metric, and then we've done a bunch of incrementality tests and had a bunch of calls and Slack messages.

[00:29:58] The organization of all of that data into a single database is actually an asset that enables something totally new, which is in this clawed open claw world that we all now live in. We only as an organization, design and build an HTML. So like we have eliminated decks. Like we don't build Google slide decks anymore.

[00:30:16] We just build live things in HTML. So if a customer says to Luke Hey Luke, I'm curious about the volume of ad creative that we're producing and whether it's enough or not, and how to think about how many we should produce. Well if we don't have a prebuilt dashboard for that. The old process of trying to answer that question was probably 30 hours of multiple people's work in order to produce the analysis and the display of the information in a way that can answer.

[00:30:43] But I'm gonna let Luke show how the underlying database connected to Claude and Open Cloud now allows us to create instantaneous dashboards and analysis connected to our methodology instantaneously saving tens and tens of hours across the organization. So Luke, I'll hand it to you.

[00:30:59] Speaker 8: Yeah, so I'll, I'll bring something up and we'll keep walking through a visual with a bunch of data like we, like we've been doing.

[00:31:05] But before I get there, so just sort of like, take a step back, everything that Taylor's walked through, the profit engine, the profit engineer, and what that enables. We we, we will do this thing where we started analyzing the, our Slack conversations and calls across the organization and assessing how much time is spent discussing different topics within the organization and your organization.

[00:31:27] 80% plus of your time is being spent on diagnosing what the problem is or what the problems are to solve, rather than net new insight generation or action. Like on average, it's gonna be 80% plus. And that's, and that's what we see across our data set, is like the conversations and the amount of time that we all spend.

[00:31:43] And so what's going on? What's the performance like? Is this the right data? Should we look at this report as well? Let's look at this different attribution model and trying to uncover what the action is to be taken accounts for so much of our time. And so everything that the profit engine and profit engineer is compressing the time from insight to action, one capable operator to handle all, all of those roles, which, which enables this workflow, this output, the efficiency.

[00:32:07] And then the thing for us to think about is like, how do we expand on this, right? Like how do we, how do we build on this to produce the net new things that are actually gonna continue to add incremental value now that we're not spending 80% of the time trying to figure out what's going on and what to do, right?

[00:32:20] We're spending 10, 20% of the time, right, and can spend a lot more time on this. Sort of set of actions. And so like Taylor alluded to, what we see as our unique position in providing operating leverage to you all as brand owners, operators, marketers, is at the intersection of the data, the database, the methodology, and then the capable operator enabled with an ai.

[00:32:43] And so those sort of three layers and the data piece Taylor's talked about all the data is aggregated for your business at the order level. Finance, marketing, cost level, the data lives in the context of targets and your business forecast and historical performance. So it's not just like data in a sheet, right?

[00:32:57] It lives in the context of your actual business expectations. And then it's further informed by our context of a larger data set of hundreds of direct to consumer e-commerce brands. So that that has to exist, the data within that structure and within that context. The second piece is the methodology, which is our unique place in the world of sitting at the intersection, the data of of, of billions of dollars of GMV across hundreds of brands.

[00:33:20] And and and what we're able to do is that we're able to benefit from the pattern recognition across those, that data set, right? So we're able to see what's working well, what's not working well. We're able to see into the individual streams and workflows of all the different D two C brands, how they run meetings, how they run their organizations, the tools that they're using, how their media buying methodology, their creative vendors, and resourcing.

[00:33:42] And then we build our methodology around what we're seeing, working well and not working well across the, the broader data set. Right? And then the final piece is a capable operator enabled by ai who has what we like to call clarity, accountability, capacity sitting at the intersection of all the tools and everything Taylor talked about.

[00:34:00] And that allows us to focus our time on the highest leverage, impact, insight, generation and action. So data, methodology, and then a capable operator enabled further by AI are like the three pieces that help to enable this, this vision altogether. And so an example of this is like, okay, we have the data, we have the methodology, and we have the system built out.

[00:34:23] And so we're able to uncover these insights. What does that actually look like? And so the example Taylor gave is a good one. I'll pull up my screen here and share an HTML not, not a deck around analysis that we did for this, this topic around creative volume needs and creative and creative performance.

[00:34:42] And this is likely one that a question that you all have circled around in, in different spheres at a different time. So, so probably beneficial in terms of the framework, but the starting set of this is, okay, what is the necessary creative volume for this brand? And what is the, what is the right allocation of asset type, right?

[00:35:01] Video image creator content, low five versus branded. What, how should we, how should we think about that? So, step one, the database exists, right? We have all the data, the stat lists. MCP is something that is coming where we have all the data connected into the, the models that we're able to leverage through Claude, et cetera.

[00:35:18] And so that exists. Then we have the methodology of how we think about the different ways to look at creative velocity and volume needs. And we'll look at that across four different dimensions here. The first is spending power, the second is the outlier engine. Third is creative rotation, and then fourth is activation economics, right?

[00:35:38] So these are four different sort of like when you think about methodology, when we, when we look across the data set, these are four critical ways that we see our creative volume and velocity needs are being informed. So to start with the first spending power this is something connects to the, to the spending r the spending power model at Taylor Walkthrough as well, which is at the highest level.

[00:35:58] If we are making adjustments to our media mix and producing additional creative volume and diversity of our creative content, all the things that we're doing creatively. What we should see is that it increases our spending power over time. First, at the highest level, it allows us to spend more dollars and press against the degradation efficiency curve that exists for us, right?

[00:36:18] So for each one of our business, the spending power model that highlights that, that as we spend more the degradation curve, we start to hit that ceiling. And so improving our spending power is all about how can I spend more money and press against the, the ceiling of that curve, make the curve flatter of what my spending power looks like, so that my CAC and a MER efficiency doesn't degrade as fast as it would have historically, right?

[00:36:41] So for this business, there. CAC over this time period was stable, basically flat year over year and spent 34% more in, in total ad spend, right? So like that's what you wanna see. At the happening at the highest level, we're able to spend 34% more drive, much higher new order volume, and our CAC remains consistent.

[00:36:59] It's a really good signal as it relates to spending power and sort of this first bucket of the creative methodology and what we wanna see happening. That said, there's all sorts of factors that pull into this, right? Which is like the new product launches, the competitive space, et cetera. So we wanna start here, but then we need to get to a level that's much more specific to the actual ads within our ad account and the behavior of those ads.

[00:37:22] Which brings us to the next section here, which is the outlier engine. We have a series that Taylor did called Outliers that sort of builds around what we've seen over across our data set over the years in terms of the behavior of ads within our ad account and. What that is, is that across every ad account, a tiny percentage of ads drive the vast majority of results.

[00:37:42] This is what a power lies. This is how Meta's algorithm works. So we define high performing ads within this ecosystem as ads that spend greater than equal to one standard deviation above the mean. So in your ad account, if you took all the active ads over a time period and then all the aggregate spend in your ad account against those ads and divided those, you'd have the average spend per ad, right?

[00:38:06] So on average is how much spend we get per ad. There's a wide there's a wide distribution of where that's coming from, right? It's not an even distribution of, okay, on average, this is how much spend is gonna come through each ad. And so what we do is we take what the, what the mean is what the average is.

[00:38:23] And then we increase one standard deviation above that to signal ads that are getting one standard deviation above the mean in terms of their spend distribution, our signal to us as high performing ads or your outliers, right? So, what what this does is it helps us to understand what percentage of our media spend is being driven by a smaller subset of by a smaller subset of of ads ads.

[00:38:48] Camille to your question, ads in this case is ad id, so unique ad ID is within the, within the account. So, so as it relates to the outliers, so what we can see here is for this brand, three and a half percent of all ads are outliers. So 194 of the 5,595 ads, three and half percent are outliers of the total ads in the ad account.

[00:39:09] Those three and a half percent of ads account for 66% of the total a spend in your account. 3.5% are outliers. Those three and a half percent of your ads account for 66% of the total spend. And you can see that this is, I'll scroll down here real quick. Like we have some benchmarks for other brands where you can see this is really normal, right?

[00:39:29] This couple other brands here, 5.7%, outlier percentage, 4.4%, 3.7, 1.8%. This is typically what we see is single digit outlier percentage, somewhere in the range of two to six to 7% right on the upper threshold. This is how, how it works, and you all have some experience of this, right? Where like your, the, the three ads in your ad account that have been running for the past two years still account for 70, 80% of your total spend, right?

[00:39:55] And so we can see how that plays out here, where in this chart, here's all the ads that were launched in every, any single month over time. And then this dotted line is the outlier threshold. So the mean plus one standard deviation. And you can see all the ads launched over time, and the ones that are above that threshold are sort of signal the distribution that are becoming that are becoming outliers within that, within that time period.

[00:40:17] And there's, in terms of discovering outliers, there's a pretty wide distribution in terms of the time necessary to discover them. It's, it's not like we launch an ad and then within two weeks we know for sure. It's also not like we launch ads and then it takes six months to find outliers, right?

[00:40:32] But the distribution the fastest quartile, like the fastest 25% of ads that you launch will become outliers within 18 days. So that's the fastest you can expect it to happen. On average, it'll take 39 days. So a little over a month it'll take to find outliers. And then it, it takes up to 81 days.

[00:40:50] So the, so, so the last sort of quartile in terms of the distribution of these ads it's gonna take 81 days to find ads. So what does that mean? We can't call it early. A two week, two week timeframe. Is, is too slow. And then volume ultimately is the, is the core thing that's necessary in order to be able to discover these outliers and be able to see them play out over time as well.

[00:41:11] Okay. So spinning power outlier engine, and then third creative rotation. Right? So there's this other piece, which is how how quickly are your creators rotating out of your top 10 in your account? Are you seeing the same ads in your top 10 quarter after quarter, after month, after month? What does that look like?

[00:41:28] So for this brand, 70% average quarterly turnover. So, 70% of the ads in the top 10 are replaced every quarter for, for this sprint. So seven outta 10 of the ads are gonna be replaced, which sort of signals to us what the necessity is to replace those over time, right? And so we can sort of see how that sits on a, on a monthly basis, 60% turnover, 70%, a hundred percent turnover in this quarter.

[00:41:52] And, and sort of hovers around there. So what this frames up for us is this last piece of how we think about creative volume and and the necessity to replenish around activation economics. So ads don't last forever, right? So we see this happen where ads die off and stop producing in different timeframes.

[00:42:13] And so there's a replenishment need in your account. So in addition to like discovering the outliers and, and identifying what those ads are, we need to replenish the sort of baseline of our account. 'cause there's ads that are dying all the time. And so in this case for this brand, 79% of ads don't activate.

[00:42:31] And this, in this case, how we're defining activation is they reach one K in spend or more. So 79% of all ads launched never spend a thousand dollars. So don't activate 21% reach greater than one K in lifetime spend. And so actually out activate. And then three point a 5% again is the act, is the outlier rate.

[00:42:51] Of of, if when we launch ads, what percentage of them are going to become outliers? And this is sort of a visual of that happening where the gray and the stacked bar chart, the gray section here is all ads we're carrying over into the future. So they're sticking around with us and we're building our foundation of creative.

[00:43:06] The blue section is the new ads within within that specific time period. And then the red. The red line is the ads that are dying off each month. And so what we wanna see is something like this where we're stacking net new ads on top of each other, and we're creating this foundation where ads that we're carrying over into future, the future stick with us.

[00:43:24] And we're replenishing enough ads in the account to take the place for the ones that die off, while also creating enough volume to where the outlier the outlier rate becomes more of a math problem than a gambling problem. And that's, and that's sort of what we see as it relates to creative, right?

[00:43:40] With like, as we zoom out, spending power outlier rate activation, economics. Ultimately there becomes a level, this is sort of a simulator for like, at different levels of spend. What it looks like ultimately becomes the, a level of the amount that we're spending or the amount of creative that we're launching where because of how the math works, the, the outlier percentage and activation economics, it's going, we're, we're playing a game that is much more similar to gambling on the creative than it is to like working a math problem and making the probabilities work for us because the volume's just too low to get there.

[00:44:13] Right? So like for this brand, when we're launching 50 ads a month we're we're going, we're gonna be very lucky if we find one outlier breakout within that subset, right? Just based on how the probability and that outlier percentage look. Looks like. Whereas when we increase the volume, we can start to see, okay, if I launch 200 ads this month, 46 are going to activate and eight are gonna be expected outliers.

[00:44:36] And that's the game we're playing, right? Which is not, let's launch a hundred ads, 200 ads, 300, 300 ads, and they're all gonna produce, at some level, there's gonna be 80% of them that don't activate at all. So 20% of them are actually going to activate. And then of those, a few percentage points of those ads are actually going to become outliers that become the outsized growth engine for your account over time.

[00:44:58] And that really becomes the framing for how we think about creative volume and then necessity to be able to drive the outcome that we're all after. And so this, again, taking a step back, this is sort of how we think about, okay, we have the data infrastructure in place, we have the methodology and the view of what we see across our data set, and then the operator enabled by ai, we're able to spend more time.

[00:45:21] Creating net new insight and and, and creating action off of creating action off of those insights informed by the dataset and the methodology as well. So that's sort of framing up this piece.

[00:45:33] Speaker: We move it all the way to the right just to see. Camilla wants to see what what happens when you wanna spend infinity

[00:45:37] Speaker 3: dollars.

[00:45:39] There you go.

[00:45:39] Speaker 8: 35 outliers.

[00:45:41] Speaker 3: Yeah. And that's what it would take, right? Like this, this is, this is what people just don't often struggle to fundamentally grasp about media creative is that we all want the idea that we all know how to make hits every time, but it's just not how it works. Like, that's just not the reality of making ad creative.

[00:45:52] But Luke, lemme ask you a question. Do you know how to code?

[00:45:56] Speaker 8: Do I know what to cook,

[00:45:57] Speaker 3: know how to code?

[00:45:58] Speaker 8: Oh no, I do not know how to code.

[00:46:00] Speaker 3: So you don't know how to code anything? How are you as a designer?

[00:46:04] Speaker 8: Designer I plugged around in Canva. Here or there?

[00:46:07] Speaker 3: Yeah. Yeah. Like you could screenshot things,

[00:46:09] Speaker 8: I could mix those, mix some pixels around.

[00:46:11] Speaker 3: So like I, I could show you some Luke's decks from back in the caveman days, like six weeks ago when we had to just make them in Google slides with screenshots

[00:46:18] Speaker 8: like a lifetime ago.

[00:46:19] Speaker 3: And, but he built that by himself, chatting with an interface on top of our database. He built an interactive h TML dashboard to make a point built on top of methodology and like, that's the world today.

[00:46:29] It's just like so different and faster in terms of getting to the insight that, that, to create that, to actually end up with that end asset, like a front end interactive ui. On top of that data analysis would've taken like five people a hundred hours. It's just like it. So, so, so this idea of what the biggest struggle I have right now is like, what are my people supposed to be po like, capable of?

[00:46:50] Like what, what's the actual boundary of what we should expect from people to be able to produce? 'cause it's there, there are people that feel like they're playing limitless, the movie, you know, like in inside of the organization in terms of what they're able to do. But I wanna show you like why I'm gonna make a, this is, this is a bit of my like selfish sales pitch of like, why I think, okay, so cool.

[00:47:09] Taylor, I'm gonna go do this myself, but I, I, I just wanna show you something. And this is, so this is my, I run open claw. It's like my, my, my AI instance of how I interact with the world. But I wanna show you how the sequence of interaction often works without this context layer. You're gonna hear this phrase a lot as it relates to the AI context layer.

[00:47:26] So while we're sitting here, I took this screenshot, okay. From stats, and I gave it to my claw and I said Hey, build me an interactive HTML analysis of the right actions to ensure we hit our contribution margin goals for the month. How should I prioritize the actions? Okay? And it starts to go through.

[00:47:42] And what it first tells me is that the problem is almost entirely Google spend is up 51% revenue down 45% row is cratered from 2 6 5 to two 18. That's roughly blah, blah, blah, blah, blah. But when I go and look like Google's actually way ahead of its efficiency target. So it just like fucked it up. It just like hallucinated like that.

[00:48:00] This is the problem, right? And then it tells me, I was like, Google is ahead on spending efficiency and it's not the problem at all. Green is good, red is bad. It's like, oh, you're right. I misread the color coding, right? This is like this constant thing that you deal with. But I wanna show you what I did. So I have this video I, it's called the Hierarchy of Metrics.

[00:48:18] Okay. And it's what our dashboard was built around. It's built around this idea that the way you should sequence through information is contribution margin is the most important thing, and then there's layers of things that work below it. And I said, okay, hey, dumb dumb bot watch this video and apply the knowledge to your analysis of the screenshot.

[00:48:35] Okay? So, this is methodology on top of data, on top of context, right? And so now it goes, good, I watched it. Your per Pyramid of success framework changes the entire analysis. Let me redo this properly top down. And I'll copy and paste this in the chat so you guys can like read through the distinction between, okay, what was the analysis that I got the first time absent?

[00:48:54] A contextual understanding of methodology for analyzing e-commerce data. And what did it get when I gave it that context? And how different is it, right? It's like two 75,000 characters too long. I can't even paste it in the chat, but you get the idea of what I, what I'm, what I'm getting to. And then so I asked it like, alright, now rebuild that in light of that.

[00:49:12] So let's see, let's see what it did while we were sitting here. Okay. So while we were sitting here, I was like, all right, in light of that, here's an action plan off the back of where we're at. So Pyramid of Success, four layers is the only thing that matters. So it created sort of this pyramid view of it created a deep dive, broke the expectation, created a diagnosis, made a recommendation, maintain spend, close the app through a OV and retention protects future while hitting the scoreboard.

[00:49:39] Here are some pro prioritized sets of actions, blah, blah, blah, blah, blah. And like, again, this was three seconds while I was sitting here, while Luke was talking. And so, so now could I go through and probably find some, some issue with some of these, but again, like interactive changes to, if I can affect these things by this much, off we go.

[00:49:58] This is just sort of the world that we live in now where suddenly our people can take everything that CTC has intended for them to understand how to deploy. We can provide them with this really well structured database that we've taken from you guys and organized it in a way that makes it actionable.

[00:50:14] Eliminate all the API risk associated with trying to interact with all these things, gave them the tools to move fast. And now it feels like there's like super humans everywhere. And I think the thing that's really hard is that this public messaging around like, you know, making one role combined into three, or like, I don't know how many of you guys are playing for triple whale in house and all the different tools all together at once.

[00:50:35] It's a big leap to go like, oh wait, one person, these three people, like the three people on my team work hard. They're smart, they're capable. And it's like, yeah, you're absolutely right. They are. This is not about someone becoming sort of suddenly more intelligent than others, but it is to say that like for 12 years we spent as an organization, like 10,000 reps at trying to build a system for this, for forecasting effectively, for building methodology, for thinking about how these things might occur for building a creative demand plan that would help a system to engage, to organizing Dropbox links that make people able to move fast.

[00:51:05] So the, the, the delta between how an internal thing might function around what you guys do, which is one brand, one instance of data to, you know, all the things that go into it. It's just a, a byproduct of messing it up a lot, learning a lot along the way, but the organizing and compounding of the information, and the other thing that, that, that affects it is like, okay, so Luke did this analysis.

[00:51:27] Well, what, what now is that in, in the old world? Like that deck sits on G Drive and like every organization has this thing where like, we record every meeting and they're available if you wanna read 'em. And it's like nobody has ever watched back a meeting in their life. Like that's what a nightmare to watch a meeting.

[00:51:40] I wasn't even in, let alone read meeting notes or whatever exists or read the deck that Luke created for some client meeting. Like, but what instead happens is the template for that analysis can now instantaneously become a report and stat list and be published through everybody. So, so the way that you can compound the knowledge set is that every time one person has a good idea.

[00:52:01] It can affect everybody and the persons, right? In the old days it was like, Hey, I have this idea for a report. Create a product ticket, put it in Jira. The dev team will prioritize it never, and maybe eventually it'll get built, right? So when all of a sudden everybody can build the analysis that's in their head for smart people and then that can compound across an organization, it all just moves really fast.

[00:52:24] So that's a, a little bit about what we're trying to create and give to you all to help you do what you are doing. And really what we wanna give you back is like the dream state of the best relationships we have is that there's a lot of things that we're never gonna do for you and we're not going to build your product development roadmap.

[00:52:42] That's not what we do. We're not gonna help you organize your supply chain. We're not gonna manage your cash flow. We're not going to define the big marketing moments in the future for your business. The ones that are gonna break the model that are gonna tell stories and do that big influencer partnership and all those things.

[00:53:00] But the best partners we have work on these counter cycles where, okay, C, D, C, your job is to, we're gonna agree on the expectation. We're gonna have the marketing calendar, we're gonna trust you to execute tomorrow and the next day and the next day, and deliver to this outcome and hold you accountable to that outcome.

[00:53:14] Meanwhile, we're gonna go build the transformative plans for distribution, expansion, and product development and big marketing moments, and we're gonna work off the counter cycle. That like, 'cause the reality is that's what growth primarily comes from in our industry is that it's not this idea that you're gonna make the next great ad.

[00:53:30] Like that happens sometimes and it's awesome when it does. But primarily the best brands that are growing are developing new products, expanding channels, finding new stories to tell that unhinged the data from the model. They break the model, they reach a new tier of performance and we aren't set up to accomplish that for you.

[00:53:47] And so this is what we think we can be is like the best execution engine of your realized business. So of the products that exist, of the stories that are capable of telling, of the creative that has been produced, how do we ensure that every day the allocation of those efforts produce the most consistent pliable or possible outcome?

[00:54:05] That's us. That's what we're gonna do for you. And it's why we're able to forecast so accurately. 3% to target across $3 billion is because forecasting is an exercise in execution. It's working to make it right. It's not guessing. It's how quickly am I off course and can course correct. So if you wanna build that level of confidence into your own business to give you this, the freedom to step back to go do the bigger things, that's what we think we can create a great partnership around.

[00:54:29] So. That's our a little bit of what we're up to. We're hoping to add more and more things to this. Right now Google still happens as an individual buyer. It's not built into the system yet. There's some complexities around the deployment that we're working through email and SMS. We have a service offering that's still just people designing emails and deploying it.

[00:54:45] But all of that is coming. Like, the reality is, is that it is just way too it is way too capable and way too fast and way too good for us to stop or to continue to hand. Humans whose brains aren't designed for perpetual data analysis. Like the reality is, like, that's just not what we, our brains are designed to do, is to monitor data 24 7 and make decisions.

[00:55:06] It's just not the thing. So, we would love to do a demo for you guys of this specifically. The easiest entry point we have is what we call our profit system, which allows us to basically build all the infrastructure out for you, hand you the analysis of what we would do, and then either give it to your team to run or we can do ongoing service from there.

[00:55:23] The other thing we're really confident in is all of our pricing will include some sort of outcome-based obligation that sort of ties us to doing what we said we're gonna do because we believe it has the capacity to do it. So, that's what we're up to with the profit engine. That's what we mean.

[00:55:37] It's a combination of tooling and engineer together. We like to use the Ironman metaphor. It's a bunch of Tony Starks combined with the Iron Man suit that make it possible. We have awesome, hungry, intelligent people that are AI native, that are excited to use these things and then we give them the 12 years of history and all the, the tech to let them play with and set 'em free on your behalf.

[00:55:55] So appreciate you being here. 12 o'clock on the dot, have an awesome afternoon.

[00:56:01]