Data + Methodology + Operator: How We Build Capacity

[00:00:00] Luke: So once you have this data infrastructure, then the next thing you need, you can't just be like, cool, I have the data and I can just pour it into all the tools and like, great, I'm gonna have, I can 10 x my capacity. Right? Like there's, there's another step before getting to the individual tech enabled individual, each one of us being able to unlock maximum capacity using these tools.

[00:00:19] And that is the methodology or the context layer that actually sits between the data and then the tech enabled operator. Or the profit engineer 

[00:00:27] Richard: Hey folks. Welcome back to the E-Commerce Playbook Podcast. I'm your host, Richard Gaffen, director of Digital Product Strategy here at Common Thread Collective. Now I'm joined today once again by Mr. Luke Austin, who's our VP of E-Commerce strategy here at Common Thread. Luke's been talking to me for the last few episodes just about our new profit engineering system or our profit engine system, and, specifically, we've talked a a little bit about kind of exactly how it works, how it provides the sort of functionality that it provides to clients, why it's this such an incredible system that puts all of the different roles of the e-commerce team into one person and actually does it more effectively than a traditional e-commerce team does.

[00:01:04] So we've spoken to all that a little bit, but what I wanna talk with Luke about today a little bit. Is the infrastructure behind the tools. So we've talked a lot about like, hey, we give you access to this tool stats that aggregates your data, puts it in one place, and then we build some data models out for you.

[00:01:20] And then our profit engineer swoops in and is able to press some buttons, pull some levers, and provide you a growth strategy that's better than anything you're working with already. But what we haven't really dug into is exactly what those things are and why they work. So I think what we're gonna dig into today is.

[00:01:37] The data infrastructure underlying what we do. So why do we have this capacity? Why is the AI tooling that we use better than just plugging whatever into Gemini or Claude? Why are, do the tools that kind of undergird this thing make this so effective? So that's enough outta me. Luke, I'll turn it over to you and let's talk a little bit about the infrastructure underneath the profit engine system.

[00:02:02] Luke: Yeah, so we've, we've been thinking about this a lot as we've been building in this, in, in this ecosystem, and even over the recent months with the progression of the AI and autonomous agent tools of thinking about like, okay, what is the, what is the infrastructure? What is the foundation? N needed that, that produces the best outcome and allows each of us in our roles to be able to have the higher capacity output.

[00:02:29] Right? And so sort of specifically in the microcosm for like our profit engineers, what, what that needs to be, and then for each of us individually. And so if you're thinking about that for yourself or your team, this, this maybe a helpful point of view as, as we're just sort of in the, in the journey of this together, of what we've seen as being necessary.

[00:02:45] And it's sort of like a three, three layer. A three layer thing that starts with the database layer aggregated data across a wider dataset than just your business as well. The second layer with is, with is, which is a methodology and context layer. So this is this is how do you approach the data and how do you think about things based on, based on that.

[00:03:05] And then the third layer, which is the individuals and the tooling and the AI connection, like tech enabled individual that is then leveraging that foundation for. Increased capacity and output for themself, right? And so to parse each of those layers out, the, the first and foremost is the, is the database layer.

[00:03:26] So first what. Is the starting point is that you need a database layer that's aggregated for your business across order level, finance, marketing, and cost metrics. Like you need all of that together in one place and aggregated. And in most cases like that, for a lot of brands, it doesn't exist even as the starting point, right?

[00:03:45] Like different platforms, different systems, some spreadsheet, and we have NetSuite for this, and we have Shopify for order level. So the data has to live. Together in, in one place from on the order level, finance, marketing, and cost to be able to have a central place to pull all the things, pull the things from.

[00:04:02] The second piece that's necessary within the data layer is the data has to live in the context of targets or expectation. So when you have data that just. Lives somewhere. The only context you're gonna get around it is like historical performance. Like how does this past month compare to the same month that year ago?

[00:04:21] Whatever, right? Like, but the data lives outside of the context of your business expectation, right? So you need on the database layer, you need to have your targets and your forecast to live at that layer as well. And the context of these metrics, right? So you have order level, finance level marketing, cost data, all aggregated, and then you need to have expectations for what those things need to be.

[00:04:41] Otherwise, like the data's living outside of context, outside of just historical performance, which isn't isn't helpful in and of itself. 'cause we can say like, okay, we're pacing 12% year over year on this metric for this time period. Well, does that get you to your business goal? You're trying to grow 23% year over year.

[00:04:58] Like what? You know, which targets are off, which metrics are on or off. So, and then the third piece of the, the data layer is that you also. Ideally want it to be informed by the by the context of a larger data set, right? So benchmarks, trends, industry. Specific insights that allow you to, outside of the context of your business forecast and targets, also understand what's happening in the larger, in the larger space.

[00:05:26] That helps to give context to the data. So the, the database layer, those three things. We have seen as being really crucial in building any of this which is to have all the data aggregated in one place for it to live in the context of your business expectation and your forecast, and then to be for further informed within the context of a larger data set.

[00:05:45] Richard: So, yeah, so I think been then to, to summary, summarize your summary, so to speak. There's, there's the aggregation of all your data in one place, which, okay, so on one level, that's, there's plenty of platforms that do that. But then there's also the, the element of. Like the modeling piece, right? Which is the idea that we could take that data and then from there to put it in the most sort of basic terms, extrapolate what ought to happen or what should happen if business as usual continues, right?

[00:06:13] So not only do you have that data aggregated, you also have some sense of on a day-to-day basis, what status will show you is, let's say your contribution margin over the. 15 days or something like that was, let's say 50 grand or something, whatever. Let's just throw a number out there. What that'll, what the status dashboard will also show you is that whether or not that 50 grand is a header behind what was expected to happen over that 15 day period.

[00:06:37] Which is again, something that is unique to status or is like the thing that kind of helps drive the value of status. And then the final piece too is that like. In status. We also have visibility into the entire industry. So there's also some sense of whether or not overall trends are playing a role in your, the sort of success or failure of your business against that forecast.

[00:07:02] Right. So in, in other words, this is, that is just all a long way of saying this. The data that is aggregated by status. It contains a, a, like you were saying, a point of view. I think that's really important. It's like, what does all the data mean for your success? And that's explicitly stated. And then it also puts it in the context, not only of your historical performance, but in the historical performance of everybody out there in the e-commerce, or at least in our, in our data set.

[00:07:28] So that's, is that, that feels like a

[00:07:30] Luke: Yeah, that's right.

[00:07:31] Richard: Yeah.

[00:07:35] Luke: Anyone that's, that's tried to aggregate data for themselves across disparate data sources, I think can un, can appreciate that. It's more challenging than it may seem to be. There are, there are API endpoints and webhooks for like a lot of the stuff, right. But even just taking Meta Ads Manager, like to be able to get the access necessary to be able to port into an LLM or other analytics tools or whatever.

[00:07:59] AI tool you're using, like you have to create an app, then you have to create a, then you have to create a key, a system, user key within the platform, right? And then assign the assets. And it's like, and then you have to do that for like, all seven of your things where like, and, and especially the cost data like that tends to be the more challenging one where like all of that living in one place.

[00:08:17] And then the next step is, okay, how do we provide, how do we provide access to that and, and API endpoint, right? Or like be able to pull in that data. And so that's, that's the thing that we're in the process of now. And more to come on that related to the SATs, MCP and all that data being available outside of sort of our system and being able to utilize, but for, for broader, broader group of folks.

[00:08:36] But yeah, that's right. Getting all the data into one place, having it live within the context of your business expectation targets as well as the broader industry. Consumer behavior dataset and then being able to have access to that in a simple way. To be able to have that at all time, to generate all the reports and analysis and forecasts and all the things that we, we want to do more and more.

[00:08:58] Richard: Yeah. So then let's talk about how that, the, the sort of, what that enables. So, the original thought behind what we were gonna make, the topic of this podcast was capacity, which is to say like, how is it possible that one profit engineer is able to do the job of, you know, five people or whatever the case may be.

[00:09:15] And so this is a, this is a huge part of it, right? The data enables the, well, so not only does it enable the forecasting piece, and it enables sort of all the things we've just mentioned, but like you were kind of alluding to the data that we have access to also feeds the AI tools that we have so that the ones that you use within our system are much more informed, not only on your data, but also the set of the entire, of kind of our entire data set, right?

[00:09:42] So there it's ability to make, let's say, informed. Choices, the AI is greatly increased over and against just like asking, again, like asking chat GPTA question. Right? So let's talk a little bit more from your perspective about like what capacity enablement looks like given this data infrastructure.

[00:10:01] Luke: So once you have this data infrastructure, then the next thing you need, you can't just be like, cool, I have the data and I can just pour it into all the tools and like, great, I'm gonna have, I can 10 x my capacity. Right? Like there's, there's another step before getting to the individual tech enabled individual, each one of us being able to unlock maximum capacity using these tools.

[00:10:20] And that is the methodology or the context layer that actually sits between the data and then the tech enabled operator. Or the profit engineer and the, the methodology and context layer is crucial because here I'll, I'll give an illustration. So. Many of you might be familiar with our stats dashboard and how it looks from some of our other materials, but our stat homepage dashboard, it's formatted in a way where you look at the dashboard, it'll show you month to date against your, against your business forecast for 35 critical.

[00:10:52] Business and marketing metrics, and it'll show you red or green, how you're pacing against each of those metrics, right? Starting with contribution margin at the top, then revenue spend, MER, business level metrics. Then it gets down to customer cohort metrics. So new and returning orders, revenue, a OV, then down into the platform metrics, meta spend, meta, iro os, meta i, revenue, Google, apple oven, et cetera, right?

[00:11:12] And so the dashboard has every single metric, red or green based on the pacing month to date against a specific target. So we can see we're off course. So, if you were to take a screenshot of the status dashboard and just drop it to Claude or whatever pick your tool and say, what's going on with the business?

[00:11:27] Where am I off? Fix, fix this, improve my business. You know, just, or like, just gimme insight into what's happening. Like whatever the generic, prompt, prompt is that you're working on solving the problem or increasing the forecast spacing, it'll give you a very unhelpful. Output. It'll, it will be it'll be all over the place.

[00:11:43] It won't read the data correctly, like it doesn't have any context into what's going on or how to think about analyzing the performance of an e-commerce business. And, and what's most important. Now, if you go to. Our blog or some of our materials and and pull anything, YouTube, video or a blog related to the hierarchy of metrics, which we've talked about a lot, right?

[00:12:04] Right. There's a hierarchy of metrics in how we look at and execute and operate a business on a dayday basis relative to those with contribution margin being at the top of the pyramid and then it get into business metrics and it get into platform metrics, right? So, and this is how we think about making decisions on a day-to-day basis, is maximizing the contribution margin against.

[00:12:22] The goal, everything is subservient to that as we go down the pyramid. And so if and, and so that, that contextual methodology method, methodological layer, if we layer that in to whatever the tool is that we're using, this is how to think about analyzing the performance of a D two C business and what the objective is in terms of what we're after in maximizing contribution margin.

[00:12:44] Understand and learn this context machine. Then you give it the stats screenshot of the homepage dashboard with all the targets. The output is wildly different. Wildly more helpful relative to understanding the context layer that sits in between the data. And we can give like multiple examples of this like creative strategy and creative output.

[00:13:04] We've done a lot of we've done a lot of work on and have different materials on outlier methodology and how we think about creative outliers and how for most brands it's between three to 6% of your total. Add output accounts for 70, 80% of the spend in the accounts, right? And so what you're after is there's, it's a probability game of finding those outliers that are gonna drive the outsizing outcomes in your ad account, et cetera, et cetera, et cetera.

[00:13:29] So, hey tool or LLM, or. Individual. Think about how many, how much creative output we need to be able to hit our hit our spin goals. The output's gonna be very different when it's informed by a contextual layer around outlier methodology that's been informed by hundreds and hundreds and hundreds of brands informed by billions of dollars in GMV, right?

[00:13:47] That's where this, this, where this stuff comes from, which is like. We see across hundreds of brands that the outlier methodologies is what happens creatively. And these are for different industries, how it shows up, right? And so what's important here is the methodology and context layer is crucial to be able to unlock the capacity at the levels that we're talking about and, and in terms of what we're after.

[00:14:08] And that. And that's, I think something that we haven't talked about much, but is really critical to think about that the database layer is unique as it relates to all the data being in one place and living in the context of targets and fork and your business objective. And then the methodology and context layer on top of that allows us to be able to.

[00:14:28] Parse that data in a way that's informed by a much larger data set, right? Because, because again, the methodology and context is informed of what we've seen over 12 years across billions in GMV, right? And so we're able to see these things happening and then apply the methodology in context to businesses that we start working with in month two, right?

[00:14:44] Or whatever the timeframe is. And that against the data layer is actually what provides the insights and the outcomes that we're looking for, that then a tech enabled, AI enabled operator. Like the profit engineer is able to leverage to create to create the the increased capacity at the level that we're, that we're all after.

[00:15:05] Richard: So it's I mean the long story short maybe is that like. None of this works if, if just a human being's doing it. We need, it needs to be tech enabled, but that tech needs to be in turn enabled by something else, which is like the unique access to data that we have. The sort of tooling that underlies some of the more maybe flashy stuff.

[00:15:24] Like I, to me, push to build is flashy, but I, I don't know if that counts, but let's say some of the AI features that the creative piece has. So yeah, so I, I mean, anyway, I think that's a good encapsulation of like, there's this, it's not just that we're rolling out some sort of like AI thing for e-commerce, whatever.

[00:15:41] It's not just AI enabled, like everything's AI enabled this today. It's like there's a unique use case for this that actually makes a lot of sense and given the fact that we as human beings, this is something I was just talking about with joy. We as human beings actually can't really run e-commerce businesses.

[00:15:57] I think we've sort of discovered like we need some, we need this AI tooling to make it work the way that it needs to work. That's the thing that we've put together and that's the thing that's available to you. So any, anything else that you wanna hit on this, Luke?

[00:16:14] Luke: No, I don't think so. The, yes, actually one more thing. The, it, it requires all of this requires at the end of the day still someone to make a judgment call and decision, right? Until, until it's fully put into, into the hands of until it'll be fully put into the hands of, of the machine for whatever sort of subset of task which again requires a judgment call and a point as well.

[00:16:38] The, the database, the methodology and context that's all informed by ideology of what we've seen across the, across the data set. And then us, like intentionally creating, sort of like, this is how to think about this thing, right? And it's actually better, it's actually more effective to think about this thing in this way than it is in these other seven ways, right?

[00:16:56] Like it requires sort of creating, creating that the stance and that's informed by the dataset. And then ultimately. The, the increased capacity that that evolves from the system is important. But someone's also make a judgment call that is related to their specific experience and them being on the, on the hook for the outcome, which is which is why.

[00:17:17] The profit engine is operated by the profit engineer and there's a person that sits at the intersection of it that allows them to make the most effective decisions on behalf of customers. That's informed by the data set and the methodology and context, but also by their unique point of view and their deep experience in this world as being highly high capacity, highly driven individuals that have a lot of experience in, in this space directly as well, across a bunch of different brands and making and making the call on your behalf and being responsible for the outcome.

[00:17:47] Richard: No, I think that's a great point that a, a tool is only as good as the person who's using it, and this particular tool requires somebody who has a ton of experience and a ton of understanding of what exactly the system is trying to do in order to get there. I think once we get to the point where. AI is able to actually make the judgment calls themselves, then that's a completely different problem.

[00:18:05] It's a different bridge we'll have to cross at that point. But alright folks, well I think that's gonna wrap it for us right now, but we've been talking about the profit system or the profit engineering system for a little while here, and if this is something that you're interested in. Come hit us up common thread code.com.

[00:18:20] If you are a seven or eight figure business and maybe even if you're a six figure business, we might have something for you there too. Hit us up, hit that higher hire us button, common throw code.com. We'd love to chat to you, you, and alright, I think we're gonna wrap it there. Take care, Luke. See you, everybody.

[00:18:35] We'll talk next time.