Meeting Title: LMNT x Brainforge weekly Date: Dec 11 Meeting participants: Shivani, Awaish Kumar, Jason
Transcript:
Them: Yeah. Yeah. It’s just kind of like I’m like, what is the ideal? Temperature zone. And then the heat goes on, and I’m like, maybe I need to take a nap. I turn off the cold.
Me: Yeah. Here. It’s also, like, 40s.
Them: Yeah.
Me: And my dog is really used to me keeping the door open. And he’s trapped. And I try to leave the door open, and I’m like, dude, I starting to shiver. And then he looks at me. I’m like, I’m so sorry. You have to stay in the house. Or I close the door on him, but then he gets worried.
Them: Yeah.
Me: So I don’t know. And I’ll try to keep it, like, cracked.
Them: Yeah.
Me: Still makes him orange.
Them: Yeah. Hey, jason.
Me: Hey, jason.
Them: Hello.
Me: Robert. I think he had a blood vessel issue in his eye today. So I think he’s going to the doctors, like, right now.
Them: Okay?
Me: Which I haven’t seen him in glasses in a while. Until this morning, I was like, too nice, smart. And he’s like, no. I think I.
Them: Okay?
Me: Think he. May be.
Them: She looks smart. Or is it just a blood vessel?
Me: I. Don’t know. He crossed that he didn’t look like he was in the brain was working. I think he’s okay.
Them: I. I only. Sorry. I was. I only ask because, oddly enough, I have a lot of, like, issues with my eyes. But it’s like. It’s one of those. Like, it doesn’t impact my vision at all. I just have really weak blood vessels, especially in the wintertime. So, like, if I rub my eye in a wrong way, all of a sudden, my eyelid turn just, like, blood red, and everyone’s like, you should go see someone. I’m like, I can see totally fine. It’s. It’s been checked out by many doctors. It’s a. It’s called a subconjunctival hemorrhage. Which basically, it’s just when a blood vessel pops, but it looks nastier than it is. But hopefully, hopefully it’s not a problem for. For your friend there.
Me: Yeah. We? I just have, like, contact. Like, both of my eyes are a little bit different vision. And then. I think, because my whole job is sitting. I mean, sitting in front, like, two feet from a laptop for, like, 10 years. It’s probably not. It’s, like, not the best thing for eye health.
Them: Oh. Yeah.
Me: And I’m just like, you need to, like, blink more. You need to, like, keep. So I have eyedrops, like, everywhere in the house. Like, blown in my car. Because that would direct. Her recommendation was like, keep just using eye drops all the time.
Them: Y. Eah. I just keep telling be told. Hey, you should really use the glasses that we gave you. I’m like, yeah, okay.
Me: Awesome. Cool. So we have a deck. We can just go through it. We have a lot to cover. So I think we just get started.
Them: Let’s do it.
Me: Okay?
Them: Sounds great.
Me: Great. So. Sort of the structure of these weekly calls is a lot of just out of the rico. Where? We blocked what’s coming up next. And then setting up, you know, setting up or adjusting a plan for the next week. And then as we get into holidays, I think we’ll start and we. We start to form more understanding of element. I think January and February will be. A little bit more clear in terms of pacing.
Them: Yep.
Me: So in terms of, like, Wins this week. We, you know, had great discovery calls. We started establishing a documentation around the element. Data platform, which we’ll review. Today. You know? We worked on the ETL assessment memo. We also have some stuff to cover around Emerson and what we found in the data.
Them: N.
Me: Today. And then I think we have a pretty good sense of, like, what the sources are for, you know? Phase one, especially as it relates to. Et. L. Each of these is. Links to, you know. The relevant. You know, asset. So I’ll share out these slides. But I think product a week. I mean, not a ton of code ring. So the engineering side, I feel like, you know, we’re just getting started. But. Yeah. Conversations with everybody on the team has been very. You know, and it’s been helpful. One part of the end of this presentation is we’ll just talk. I’ll share a little bit about Element versus a lot of the other clients, your sides that we’re seeing. But. We haven’t seen anything. That is out of the ordinary. Or. Like two out of bounds. I think, as you saw today, for wholesale, a lot of the tagging and those things that we do with a lot of folks who are used to, you know, things. Being run in Shopify, Amazon having things come in through other sources. So so far. I would say. It’s fairly. Standard. From what we see. Maybe jump in to take a look at, like, Kind of a Gantt Chart. And where we are. So. Let me just kind of switch to that. And. You know, so far, I think we are, you know, in discovery process for these three. To talk about, like, how we think about the discovery process. We are continuously, you know, taking notes on our side and, and also going back into those notes to sort of relay, you know, answers we’ve heard. In addition to just, like, getting sucked back into those docs. We’re also starting to build out our platform documentation. So these are what are the sources, what are the tools? Who are the owners of those tools? And. Starting to build. What the, you know, end to end pipeline. Potentially may look like. I think it’s great that, you know, in about two weeks we’ve met the owners. On the technology side, we’ve met, you know, several of the business stakeholders. And then. Shivani, you still continue to give us like. Okay? What is the, you know, North Star here? You know, so far I’m not too worried about, you know, our pacing. I think for me, the real, you know, action starts as. We start to like land data and model things. Which is really. The bulk of the lines and. And things here. So that’s like, that’s where we are. On our side, you know, as you guys know, you have me. Robert is also involved awaitious here. As we start to understand, you know, the different types of functional skills we need. Will there been more people? You know, I wanted to have Robert there today. Because he’s just like a lot of experience in, you know, demand planning and forecasting. Awaish and I both have a lot of experience in setting up the infrastructure. And then we have, you know, one or two people that we can bring in, for example, if we need to go more towards analysis work. You know, for anything in the middle here. And as we start to do a lot of data warehouse set up, data ingestion setup, data mark setup, we will loop in, most likely an analytics engineer from our team that’ll, that’ll support us. So any. Questions here. We’ll go through the GitHub, we’ll go through the spreadsheet. But. Just.
Them: Ice.
Me: Anyway. Yeah.
Them: Saying? Yeah. No questions right now.
Me: Okay?
Them: So. Yeah, actually, I think we can keep going. I have some questions, but I think they might end up getting answered, so it’s great.
Me: We talked about sort of the discovery calls. We do have the partnerships discovery call. Next week as well. And. Then. I think the way we’re setting up the groups, Shivani is fine. I don’t know if on your side, you can turn those into channels. Because randomly just look like group dms just in case like. We can name them to. Brainforge element wholesale. Or, you know, that’s probably. Only. Ergonomic thing that could be. Nicer. Do you know if that’s possible?
Them: I’m check. Like, if it’s just a group thread.
Me: I think you can try to convert it.
Them: Can.
Me: I looked and. It didn’t. Let me.
Them: Change to a private channel. Not all external organizations allow conversion. So it’s, like, locked for me to do that.
Me: Okay? Send me the same. It said the same moves for me.
Them: So I guess we’re, like, both walking each other out of the thing.
Me: Ok? Ay. I will go. I’ll. Go. Let me ask my team if this is looks like a setting on our side.
Them: Okay, Because I’m like, open conversation, details, settings. Okay. I can change the one with Laura. Because I own it. So. Okay, so change conversation to a Slack Connect channel. And then what do you want me to call it?
Me: You can just. Call it Brainforge element. Group so wholesale. And that way. It’s easier for me to add folks on our side, so it’s organized.
Them: I see. Okay, okay. I’m not. I haven’t used slack in a while before this company. So it’s been a minute. So I’m like, I don’t know, the slack thing.
Me: Okay. We’re just like. We have so much going on in stock that tend to just be like, Particular.
Them: Okay. And then. And then I guess the goal is to do the same thing for.
Me: Something. Tack and then. For. Ecommerce, carlos.
Them: Okay, I can’t do the tech one. I don’t know. I can’t do it. That one. Let me see. The other one. Sorry, that’s my. I had. My attention was a little bit elsewhere. What are we trying to do right now?
Me: We’re just trying.
Them: Just change things into better name channels. Like after the.
Me: To. Write a group, chat the group DM to a channel.
Them: Just so that it’s better named and more organized for them. They can add people, but I. I can’t do the other two. It’s. It’s not allowing me. To. So if that ends up being an active chat, just, like, we can, like, just let me know and we can make it later. Okay, perfect.
Me: So that’s probably the only thing there. Maybe one thing. I’m going to jump around, but let me just show. You kind of explain what I was talking a little bit about? Brainforge. And aluminum. Let me just find the right repo. In GitHub. Let me just give it a little bit of. What we’re doing there and sort of address Jason some of your points about making sure all the business context get lifted. And kind of like after doing these. Quite often. I think there’s a little bit of a method here. So I’ll share. Just open the repository. And Shivani you should have access to. So maybe if in peril, you just want to confirm. That you can also see stuff. Here. Would be good. So this is our. Yeah, so this is our GitHub repo.
Them: Okay?
Me: For us on the data side. This is our home. So from the start. Of engagement all the way to when we’re writing. Python, SQL, dbt code and sort of building stuff. It will all live here. Part of the reason that we started actually doing documentation in here is that as we’re leveraging. AI driven development tools like Cursor. It’s actually very helpful for us to also have business context right where we’re working. So what you’ll see here is a little bit of organization. And so, like, just have the repository structured. But you’ll also see that we are keeping, you know, notes. You know in here as well. And so, for example. This was just notes. Like that we. We took from. You know our call. I believe with Jason. And so the reason why we have it here is one. You know, as we’re starting to make decisions based on this, It’s a nice home for us to have that. But also, as we’re leveraging AI to help us speed up, like the development process, Having this available is context is super, super crucial. The last piece also is Shivani when we end up like layering on an AI chat with data solution. This context is like the gold. And so the more we write this down and have it in an environment that can be access programmatically, You know? What is not programmatic is like GitHub or more documents. It’s like going to make that system a super, super effective. This is something that in. In the old world, AKA like two or three years ago. We. Really wouldn’t. I don’t know. I would never go to the degree unless we had a super, super technical company. To be writing stuff in GitHub like this. But because of how helpful it is to have alongside AI. That’s the only reason we’re writing here. That doesn’t mean we’re not also just moving this stuff to the drive. So all. All of these that are relevant are getting moved to the Google Drive for sharing. And. And we’re owning, you know, taking notes and bringing stuff back. So that’s sort of like. The reasoning for writing things here. I don’t think there’s going to be. A lot. I don’t think there’s going to be most situations. We’re not going to be pointing business takeovers to this repo. At all. It’s going to be much too technical. Instead, we will just actually just move those documents to Google Docs. That. Point. Does that kind of make sense? This is something that I pushed our company to do because of how helpful it is for AI driven development to have this context. But it is. Like. It is like a new thing.
Them: Yeah, I mean, it makes sense to me. I the comment about the kind of the docs is, you know, depending on who are sharing things with, if it’s, you know, to an end, like a, like a Carlos or a Laura, they would not know how to get into this. So if there’s something externally to them.
Me: Yeah.
Them: We just need to make sure we’re aware of that.
Me: Total. Ly on. Everything that gets Jared externally to the business will be through Drive or SoC, basically. I don’t. Expect. Anyone but the data team and tech team to sort of be in here. So really, that’s like the story around GitHub. We’re not doing a lot in here. It’s a lot of writing. But as we implement code, it will all live in here. And when we get to that point, we’ll talk a little bit about how we do technical development for data work. The other benefit here is for anybody in the future that.
Them: Okay?
Me: Comes and has tried to do data work.
Them: Yeah.
Me: They have story of the entire. Thing when we started last week. Every single, basically everything from consumption is now documented.
Them: Great.
Me: So that is like a dream. That’s another thing that really wasn’t possible. It was very rare to be in a situation where you had this level of documentation. So I’m really glad that we’re doing that. So our goal is always, like, if Brainforge were to leave and you guys were to get new data people, they would come in and be like, oh, my God, this is the best setup ever. That’s kind of like what we try to develop towards.
Them: Yeah. We love it. Big on documentation. So good to know that this is formal.
Me: Yeah. I think people will be happy with the amount. That we’re doing. So let me get back to the slides here. Okay? The other piece of documentation we have is this data platform. Document. And so I want to just highlight that here. This is something that again, because there is going to be some this is going to be business facing. We do in a spreadsheet. But it’s also, like, much better. Conducive towards. Stretchy versus a Google Doc Whoop. This is going to have is just the core pieces of business context. The data sources we’re working with. As we fill out the tools we’re using. The contract information for those tools. Who owns the relationship? What is a pricing structure? So just. Like this is just like business documentation about the data platform.
Them: Yeah.
Me: This. Is also very rare that a data team has this. Most of the time when we walk in. It’s like I don’t know who. Bought that tool. Or I don’t know. Why? What is our naming convention? Or so anything that ends up looking like a list or spreadsheet that we collaborate with a business team on will end up here.
Them: Yep.
Me: I’m not a fan of one off spreadsheets that just die and everything should end up here if we end up. But if this ends up too crazy. There are ways to fix that. So we’re going to continue to just build these that we have templates for when. We get to dashboards when we get to metrics. And so the next thing we’re going to be working on next week is a sort of metric dictionary that. Awaits. Just sort of started. So we’ll be going through that and then we’re also going to start to have things around taxonomy. So tagging today was a great example of that is like a data taxonomy that. I know the wholesale team. Understands the importance of. But definitely it’s hard. I’m sure it’s hard for them to think through, like, what are the ways to create a taxonomy? So for us, taxonomy goes all the way to how we mean files, folders, how we name dashboards. How we naming Pimas databases? And then, of course, the tagging. So the data team is going to be the downstream customer of tag. We shopify the way things are named, the way IDS are coming in, and so we have a great opportunity. To have an opinionated structure there. That allows you to scale. For example, if you know that wholesale is going to scale to these multiple brands and you want to build out that tagging scheme, that’s what we would try to do here. That’s just kind of like what’s going to happen in this sort of data platform documentation. This is where we’re working through. Each of the sources. As we start to understand all of the sources, we’re marking priorities. We’re starting looking step by step on getting access to the platform, getting access to NEXT or API keys, which can be sometimes different than just getting a login. And then ultimately, when we hook this up into an ETL tool, it’s just making sure that that data is flowing. So at any moment, anybody asks, a question. What sources are available to model? What have we landed in the warehouse? What’s getting into a bi tool? This is the source of truth. Unfortunately, these are things that. We will have to. Maintain. I am also not a big fan of creating documentation that’s sort of just dies. That’s the moment you write it. But these are things that, as we go, this is what this team will be tasked to maintain. If things become cumbersome, there are opportunities to automate. But you’re not turning off and turning on sources, like, every day. Some of these will just become, like, a weekly or monthly ritual to keep in track. For this next phase, especially this month and next month. We’ll be doing a lot on the metrics dictionary and definition side. So labeling metric names, definitions, the sources they’re pulling from, who owns the metric. What are values? What is the source of truth? Now, that’s the next sort of phase. Typically what we call a KPI. Standardization is the exercise. So that’s what we’ll be going into.
Them: Just one style of big thing just while we’re on that piece is I don’t. I don’t love when things have, like, too much this is me thing, but I don’t love when there’s, like, too much conditional formatting and then I get distracted by the colors. So, like, where. Where. It’s really important to highlight something red. Great. Like. Like red should be like. I should really pay attention to it. If it’s inactive, maybe gray it out or something like that. Like some. I’m just gonna. I’m not gonna do too prescriptive, but.
Me: I’m also the same way in that.
Them: But I just like when everything’s, like, red and green, I’m like, where do I look? And so that’s why you can play with the formatting. That would make me feel better.
Me: Okay? Great. No, that’s also.
Them: Okay?
Me: I feel like I’m the same way that I have a lot of. Conditional forerunning and formatting. These docs need to be great.
Them: Yeah.
Me: So we’ll totally do that. Our company colors are green, so a lot of stuff tends to be green.
Them: Okay? Yeah.
Me: That we produce. But we don’t have to do that.
Them: It’s just like, if you actually want our attention on something, then, like, great, let’s, like, put it in a color, and if you. If it’s like, it’s fine, but I don’t want to color. That’s my individual, like, kind of formatting thing. I have a question. So, like, you know how you’re gonna get into, like. Okay. You’ll get into definitions of metrics. But I guess, like, if I were to say, Like, you know the example that Phil said that’s like, oh, one, one tool has week ending, or one platform has weekend on Sunday, and one tool, one platform has weekending on Friday, let’s say. Right? Like, like, where will that show up? As you’re like, as you’re like, describing to me the data that we have and the, like, discrepancies between the data sets. Like, where will I find that? That’s like, hey, by the way, Shopify gives you a worldview of this. The geographies. You have zip codes, but you don’t have state. I’m making shit up. Like, you have. You have zip codes, but you don’t have states. Like here, you have states. We don’t have zip codes. I’m just like, where will I see, like, the. The split of what’s? Available by platform.
Me: Yeah. Yeah. So right now we’re working on memos for every sort of core data source.
Them: Yeah.
Me: Which is like, what’s in here. What is the data looks like? For example, let me walk through what it is for Emerson, right? So we went there. We’re like, here’s all the stuff we’re sort of seeing for Walmart. You know? We’re trying to understand. Okay. What is the shape of this data? How much is in there? What’s valuable. And then we look through. Okay, here, roughly some of the things that we’re finding. In your example of like, okay, when does the business week start for element? That is something that will end up in that metrics. And dictionary. But also on a source by source basis. You’re right in that there isn’t a clear place for us to do the comparison. All of that conversation around standardizing will bring up the fact that if we’re like, If it’s difficult to standardize that, that means there is a discrepancy. So those will end up as notes in that document. Where basically we’ll say weak start. Here’s what is in Shopify. Here’s what it is in Amazon. Here’s like a recommendation.
Them: Love it. Okay, that’s what I’m talking about. Yeah? Yeah. Like, even, like, Laura talking about, like.
Me: And.
Them: Sometimes we’ve changed tags to have, like, them be go from trusted health to specialty retail. And like that. That I’m like, okay, is it timestamped like that it was that tag, or is it now that we just look back and we’re like, we assume that there were always that category.
Me: Yes.
Them: Which I imagine if she’s reporting monthly numbers and then suddenly you’ve changed the tag. A look back isn’t going to tie out. Right. So I’m like, are those things timestamped or not? Kind of in Shopify. And maybe they’re not. Maybe we already know that they’re not. But I think. I think, like, just making the, like. It’s like, in a world where we want to have, like, more omnichannel understanding, like, what are the. What is going to be the tricky part of, like, accessing historical, like, just naming those challenges really clearly. And I think you’re doing it the right way, which is like, here’s the discrepancy, here’s the recommendation. Let’s discuss.
Me: Y. Eah.
Them: Right.
Me: Like it may not have ever been surfaced. To, like, someone in your position or to the tech team or someone like Phil. This tagging thing was happening. Right. People just figured it out, like, okay, we got to tag things. We went from no wholesale so much. And, like, we got. So that’s where, like, we will have to come to a decision. On like, okay, is this the right way? But also, you’re right in that commonly the problem, the team is reporting on stuff. Right. And it is possible. Yeah, maybe. It’s taking hours to do. But as the team mentioned, like, two years ago, they changed something and then they’re like, well, we don’t have any data from four years before. Okay, so then we have to model out those, like, phase changes, and then you’re right. If there are tagging changes, And reporting at a Shopify. Which it does not. It doesn’t really account for that. You will be. Modifying historical reporting. And so that is for us to basically go. And kind of see what the impact is.
Them: Yeah.
Me: And see. And so that. That’ll be the thing where it’s like, okay, we need to keep track of tag changes. I think what we use is we use this concept called, like, slowly changing dimensions, where we’ll keep track of the fact that during, you know, 2024, the tag with this for this customer. In 2025. The tide changed. And when you’re doing historical reporting. Apply 20, 24. During that year, probably 20, 25. Doing that here. The tools themselves do not do that. You know, and so that’ll be something we have to enable in. A data modeling side. But those are all things that we will hash out. As we’ve standardized, you know. Each. Each. KPI. Like. Cac across each channel. For example. Like, we’ll back into how Carlos is. Is drilling that from the source data and then surface like, okay. These are the different tiers. The reason this is not apples to apples. For this reason, here are some. Here are some choices we have to make.
Them: Yeah. Perfect.
Me: Cool. I guess maybe just to highlight, you know, we. We talked about ETL this week. We arranged a call with polyatomic. Next week. Yeah, I know. We kind of came out of the call and, you know, sort of like, Halfway in between two. You know, options. I actually was on the phone with Polystomic yesterday for. For another client. And it’s funny because I think I. Hearing them. I undersell how much they’re like very much like. We will build what you need, no problem. And so I think it’ll be nice for you guys to. To meet them. And explain your requirements, especially Jason, I think. You know? It’ll be great for you to. Meet their team next week. And then really, the way this goes in terms of if we were to, you know, start with them, it’s actually not that you have to just, like, sign a contract. They have to connect our sources and see what the volumes are. Additionally. They don’t know exactly which tables from every source we need. So once things are landed, we will go through and say, oh, you, you landed. Like these hourly reporting tables that we would never need. That are like hundreds of millions of rows.
Them: Yeah.
Me: We. Could take that out of estimation. Right. So the common issue that, that people do during the States is they’re like, yeah, stick everything from Amazon. And you don’t realize that Amazon is a table called Search Reports Daily. I was just looking at this for someone else. And it’s sinking, like 30 million records a day. Which is every single search report by every single search query by day that your product got, like, got surfacing. Nobody needs that. That’s not, like, relevant. And so one of the mistakes that folks make is that they sync everything. They say we need everything. Don’t worry about them. They sign a huge agreement. And these guys charred by the row. Like that’s just what this industry does. And so once we land the data, we will go through and comb and make sure that we’re not sinking anything that’s. That’s irrelevant. And then they’ll give us a quote.
Them: Okay?
Me: I guess. Did we want to go into sort of what we found in Emerson? Awaish.
Them: Yeah, I can go over it. Like, in Amazon, we have, like, data for Walmart and then Target. So I have just explored some of the data coming from Walmart. And this basically just shows the, the volume of the data, which is there. Overall. So it’s like, more than, like, you can see 32, like, 40 million rows. Which are coming in from different tables. And then there is an explanation for each table, what it contains. Like Walmart item attributes. Like, it’s more like just a master catalog of all the items. Listed.
Me: Ape.
Them: For.
Me: L awaits this source rate.
Them: Store trade. It’s like it’s storing the data for stores, like the geographical trades or any operational details and anything changes. Any.
Me: Okay?
Them: Course that are going there. So that’s why it’s, like, a lot bigger. And then there is, like, omni sales. The most important ones are store sales and omni soils. Which basically store all the sales information, like either the revenue and everything. Item being sold and, like, what I figured out. Yeah, from there, like. Then we have this DC calendar, which basically is just a kind of way of like. Like we were discussing how we are going to see if our calendar doesn’t match the exact annual calendar we have, if we might have to. Get some fiscal calendar or something. We can define some standard tables like that. To match it with our company’s calendar. Yeah. If you move on to next slide that basically says that what this information, which I figured out what it really helped me. It was like I used it in basically the estimation for our ETL tool assessment, then for like, estimating the price of our data warehouses. And, like, given the amount of data, how, how frequently and how much we are going to process like that. We’ll basically adapt our cost and then. Yeah, it’s more like now the data insights. This is more like getting into the data and what I find on a high level. Still like there’s a lot of scope on of the on Digging deeper into each store. It’s line item and the performance and looking at the trains. But this is just more high level information. So like I got like there are like for Walmart, more than 600, 6,000 stores listed. But like the, the sales were elements like products are being sold is more like around 40, 500. So you can see like still there are 1600 plus stores, which doesn’t. Doesn’t sail Element products or. Or there’s no sales or something like that. Similarly, in terms of items, also, like I saw in the catalog, like, there are 26 total items listed. For element. But in the sales table, I. I can. I mean the distributions like the. In the. Like the data for like the distribution centers. It seems like we are having inventory for all these items, but still like I see the sales for only 17, so more like more than 6, 7 products, which doesn’t we don’t see any sales for. I don’t know what those. They are inactive and no longer active products, but something like that. But we need to make. Dig more deeper into it. Like what. What’s happening there? Is that something that you already like that list of? I guess it would be nine items. Is that something that you can share with us, like, after this call? For me and reference. Yeah, because I, I, I might not even know the answer, but somebody on the, like, we don’t have to work through that today at all. But I’m more just, like, curious. So if, yeah, if you share that list, that would be awesome.
Me: Awaish. Let’s put that in the memo as a follow on. Yeah.
Them: Okay? Yeah.
Me: For us to see, like what the distribution of sales is across Walmart stores. And like, maybe whoever owns the Walmart relationship. Shivani. Can be like, oh, yeah. That’s when we just started in, like, so many stores.
Them: Exactly. And so I haven’t. I haven’t introduced you to Russell yet because I. I’m thinking that unless we want to hustle on this, I’ll see his availability. But like I was thinking, we’re kind of starting with E. Comm. And. And wholesale.
Me: Yeah. Just. Hoaging at it as we’re like, okay.
Them: I think it’s good to poke at it. So it’s like, if you’re feeling like, hey, it would be helpful, but a face to the retail side of things. We can try to see if Russell’s available before the sprint ends. If you’re like, let’s just pump that for January. And we’ll tee up some questions, like, let me know when you want that Discovery call to be I. And then I can see if he’s available for this Sprint. But. But it’s like, if you’re starting. I kind of like that you’re starting to dig into some retail stuff. But it’s just.
Me: Option two. But if it helps to just, like, say, like, we found this, and if it’s like, cool this, send that to him. Like via Google Doc. If he just wants to read it, that’s fine, but. Yeah. I think it’s just good to see. Like, this is actually great data, by the way. Like, this is all. You know, we have this stuff that’s more of, like, what I. What my goal was here is, like, okay, cool. This is, like, really, really rich. First, it’s. If we didn’t get, like, if we just stop stuff that’s kind of crap, then I could be like, okay.
Them: Yeah.
Me: I don’t know what’s the worst, you know, so especially if you guys are having conversations about using Emerson or like, for example, if we, when we talk, if Emerson only allows us to go to Snowflake and this Walmart data is like, super, super important for Russell that that’s why it’s like, so helpful.
Them: And that question he might not know the answer to. Like, Russell might not know the data side of, like, the, you know, the data side of what’s permissible. Because that, like, I. If for some reason, Emerson’s like, we, this is where the data is and that’s it.
Me: Ca. T. Y. Eah.
Them: Like. And you can’t pipe it to anything else than like. That would just make the data warehouse decision snowflake. But I don’t know why that would be the case. I don’t know anything. Jason, do you have a sense.
Me: I think you’d be surprised. Some of these vendors. Are like. They just do that. They’re like, nah, it’s just what we do.
Them: Emerson folks, from what I’ve heard, are, like, really, really nice.
Me: Okay?
Them: Like stand up people what I’ve heard. But Jason, do you have any insight here or how we could like introd like I know we’ve asked them to give Utam and Awaish access or. Or sorry, maybe Jason, you just set up the access. But should we, like, alert them that we’re doing this, like, initiative and kind of start asking them about the warehouse piece? I think probably at some point, I don’t know. When that. When that time is. Yeah. Like, at some point, like, if we decide to go with Snowflake, for example, like, we’ll have to ask them for kind of like the private to share. So.
Me: With my ass. Jason, in order to make the warehouses, I just want to know if. This is the only way they can deliver it to us. Like, that’s my. My question to them is, like, what other delivery methods do you guys support for this data?
Them: That’s a good question. Yeah, I don’t know. Yeah, because I think. I think the way that they’ve done it for us right now is just. Was probably the quickest way. Given that where the data lived, and they just said, oh, let’s just give you access to it. And then we can offer this, but. Yeah, I don’t know if that’s the only way or not.
Me: Okay? Yeah, that would be my only question for them. At this point.
Them: Okay?
Me: Okay? Great. So I think we’ll. We’ll centralize Awaish I think what centralized the question like the Emerson data related ingestion related questions. As well as, like, maybe we just have some follow up questions in that doc, and then. Surviving up to you. I would rather push, like, because I just don’t want to get people’s hopes up also that we’re, like, working on this stuff and then maybe call them back for, like, two months, also because we lived on a retail side, so I would prefer to talk to him. Sometime next month. But if, if, if he does find it interesting to, to see some of this. This data is sitting there and. And it’s not that hard for us to run these queries. So it’s not like it. It’s not like, a huge distraction for us. If he’s like, there’s some lowing for. We can get wins for him, but it will take us a while to, like, model.
Them: I. Think Phil was really. Phil was really interested in like, what is this data? Like, you know, like, now that the. Now that the target. Did we just talk Walmart so far?
Me: Y. Eah. Yeah. We just have to learn. So far.
Them: Like, now that the target data is there too, like, I think it’s. Then you can say, even just like, how is retail? You know, if you wanted to just do a little analysis, like, how’s retail doing state by state? Just. Just because you can, like. You know what I mean, like.
Me: Happ. Y. That’s what. I’ll let you make the call.
Them: I. I think that would be great if you were like, this is, by the way, your retail breakdown by state. I’d be like, cool. Like, I just be like, that’s a fun gift to give to someone for the holidays.
Me: That’s how I feel.
Them: Like. Here’s a little bit of. Here’s a little something.
Me: So that’s what I’m saying. So maybe for. For. Probably. I don’t know. Wait, should you see the target data? Maybe that’s playing on like getting. You’ll have these slides. Let’s get DB like, one or two more for the target side of the house. And then. You can. Make it charitable donation. Shivani with those to whatever you need.
Them: I’m just like. I would want to know. Maybe it’s just a gift for me.
Me: To.
Them: But I’m like, if you’re in there, you know, and then I think bills. Phil’s thought was like, once somebody actually datamined it is in there. Like, what are you gleaning from? What is it called?
Me: I also. We know. I see this. I have, like, a million questions, and so that’s why. I have a ton of questions that I want to ask.
Them: Yeah, like. Like Target is the only place so far that we’re selling sparkling.
Me: I think so, because I think I saw it in. In my Austin Target in the front.
Them: Yes. I think. I think that’s right, but I’m. I’m not. Because it’s like we’re gonna go into Walmart. Like, we have a roadmap. But I. I think that might be right. But it’s like, how is the velocity of that? I think I showed you the spins. Data version of it. Right. Which is the point of sales data. But, like, how? I’m just curious, like, when I think about product velocity, how do I. I have spins data, and then I have this, like, this other data set that’s like, what those. What the store is ordering from element, and I’m just. Curious. Like how we have that conversation.
Me: Y. Es.
Them: And like, even if, like, like, if you think about, like, future of supply, demand, like, You know, Dan is like, well, sometimes these retailers are ordering from us in a really lumpy way. It’s not smooth, it’s not consistent. So, like, even, like, visualizing that for somebody to understand, like, like, how do we, like, is there a world in which we could get Walmart to be more smooth? Because that has a lot of implications for how we forecast and, like,
Me: I. Have a great. Other chart for you that we just did for another client. This exact problem, but with Amazon on PO order velocity.
Them: Yeah.
Me: Because Amazon has super spiky POS for this other folks. And. They will be like ordering a couple. And then one day the dead we like 100 pallets. And it’s sort of a scramble, so we help them sort of think about how to accept. Like those POs to come, especially because these guys will order. In sort of seasonality. And. You don’t know what they’re. You often don’t know what their. Their order. Algorith. M is. But. If you guys have some data, it is easy for us to look back and see like what is the order velocity? And for your demand planning team to say, like, okay. We. We know that in. In terms of this holiday or this season, they always do this. You know? We. Just did that for somebody so total.
Them: Yeah.
Me: Ly. We’ll do another, like, little poke around and get, like, a couple slides. And then. Yeah.
Them: Sounds good. Okay?
Me: Okay, we went through. GitHub. And then maybe I’ll talk a little bit about upcoming week. So the goals for this upcoming week are to have a very similar conversation about data warehouses. This is going to be a little bit. This is going to be. A larger conversation. I don’t think it’s a larger because there are, like, so many more options. But there’s just a lot in the data warehouse. Data movement is complicated, but storing data. And running. Jobs on top of it and giving access and security. So I want to just make sure to go through each of the core. Pieces of that. So that’s what we’re talking about. On the etl side. I think what we’re going to talk to probably talking next week, and I think we can hopefully arrive at, you know, some type of go forward pass. And then we’re also going to, we’re starting to keep track of like our discovery sort of summaries. And then in addition, in the in the GitHub, there is a running. Sort of like wiki, that we’re kind of keeping track. Of everything. And. We’re just all like, this is sort of just like. I think of it like Wikipedia. And so we’re just starting to write here. Basically like everyone remaining, what we’re finding out channel by channel. Again. Like, this is not business facing and it’s not going to serve as like disarm slides or. These aren’t, you know, it’s not a memo. But. For me. When I bring on anyone from our team, I’m going to say, Go do the wiki and ask any questions. So. I. Don’t know. Why, sure. Anyone? If you guys. What you think about this, but I find that this is kind of nice. We have some of this stuff internally where it’s just easy to. Like if you were to look at one place. For everything. It’s here, but. For, of course, like, wholesale may not care so much about the other side, so. We will, you know.
Them: Yeah.
Me: Parcel out.
Them: Yeah, it will be useful to onboard any new. Person from paying for it to come on and get up to speed.
Me: Okay? Great. Let me go back to this. And then. We have the partnerships review call, so Robert and I will be there next week. And then this month. You know, we. We want to sort of have a decision or go forward path on UTL and data warehouse nicely as well. In the data warehouse side, I don’t believe none of the recommendation we will make are like year long. Sign up before you get anything. Type contracts. Again. Like, when we get to the DI side, it will start to look like that a little bit. But those are two decisions that I want to make sure TAC and everybody is comfortable with. At that point, we want to start to initiate, you know, landing data into data warehouse for our core data sources. We want this month we’ll be a lot of establishing the metric dictionary. I think a good win meeting OAS before the end of the. Before we get to Christmas is to, like, maybe just do a review of, like, all the metrics we’ve seen everywhere. And sort of show the breadth. And then. We can kind of then prioritize, like, what we want to focus on. Or. I don’t know. I think that that would be a good thing to drive towards going into Christmas so we can kind of see we sort of ideally are. The culmination of a lot of discovery is on that standardization.
Them: Yeah.
Me: And so if you think about it, Nt n We’re coming at it from both sides. Which was like, sort of how I wanted to work. This project is we’re going to start to land data. None of which really requires. Understanding of, like, how is CACT defined? Right. We’ll get everything in there. And then from this side, we’re getting business buy in on those definitions. That when we go into the middle, which is where all the SQL and modeling is written, we can start to establish those. This is like a great plan. For this. I think. The foreshadow where there may be some difficulty. Shivani is like when we make a decision on a metric. If. It is different for what we’ve done. Or. Like, sort of like. I don’t know. I think the challenge we’ll be getting signed off from everybody and getting. Like, okay. Company. This is how we’re thinking about it. You know, it’s always. Like, you know, a conversation that we’ll have. But nothing that. We haven’t done before. And so we will, you know, we’ll be there sort of along the way. But for me. If we can have a look at some type of metric dictionary by the end of. Before we get to Christmas. That would be, like, great.
Them: Okay? Let’s shoot for it.
Me: And then, yeah, I guess, like, briefly, I kind of highlighted this. But I think, like, this is actually, like, you guys have a pretty good setup. I mean, you guys are really, really huge, and you have a great operators internally. We work with a lot of companies. Your size and larger that are a lot more of a. Mess. And so, which is great. Like, I think it makes our job really easy. Like, the feedback has been really, really crisp. And so it allows us to sort of run oftentimes when we come into organizations where they’re not so opinionated or. Each individual channel is not being run with, like, a lot of care, and so. It’s. It’s just. It’s just almost so overwhelming, like, we have to really pick our battles. I think we have a clear direction on where to go. But similarly like to our Emerson thing. We just can’t. Like, we are a data people, so we can’t help ourselves, but do this type of stuff. Like, we will always be looking at data and looking at each channel and finding insights. And as we’re doing more and more, you know, fun work for companies like Poor Order estimation, really great forecasting, things like that. We’ll start to hopefully set the roadmap for what the rest of 2026 looks like, which is really getting into, like, you know, Analytics. That’s kind of like, generally like what I’m seeing. I think it’ll be, as you guys see, our pacing. You can let us know, like how you think about, you know, the speed and everything. The way we work, as I mentioned, is I’ll sort of loop in people have necessary for giving projects. Like if we’re doing a forecasting project. We have some great people that can come do that.
Them: Yeah.
Me: Keeping along the whole way. During this phase. Is. Like they’re just gonna be, you know, they’re just gonna be dead weight right now.
Them: I have a question for you on the forecasting piece. So, like. We’re like, explore. We’re in exploratory conversations with a company that focus on, like, supply, demand forecasting. The way that they do it is they pipe the data, you know, like. Ad jason. Is that okay for me to share? The.
Me: If it’s another ncs sign. I like to sing with other tools.
Them: I can’t remember the name of the company. To be frank, it’s called Atomic.
Me: Okay?
Them: And it’s, like, started.
Me: It’s an agency.
Them: I actually don’t know. I just joined Nicole with them today, and it was more for our team to give a demo about how we do supply demand. And I’ll just say the person who owns supply, demand. Right now is Dan. And you met Dan in, like, an initial call, and it’s, like, super manual. But it’s like he’s doing it. He’s chugging away at it. Does he think it’s the. You know, the best? And he has a lot of ideas of how it could be better and everything. And so, like, I’m like, okay, a quick win. Could eventually be instead of him having to, like, manually add data. The way that we did supply demand at my previous company was we had a model in Google sheets that was backed by, like, coefficient reports. And then the coefficient reports would update regularly at a set cadence. And then that would, like, pump data back into the tool. And then you would have kind of like these trailing numbers and kind of reforecasts, like things that would look at that weekly.
Me: Yes.
Them: And I thought that was pretty good. I’m sure there are more sophisticated ways to do it now. And, like, especially when you think about, like, SKU level stuff. Sparkling versus whatever. Grapefruit versus raspberry. Like, like, you can get into, like, how many tabs of this Excel spreadsheet would you want? Right. And so I think it’s like, yeah, we’re like chatting with Atomic. They’re kind of like, we could do this by piping to Shopify directly. Piping to whatever directly. And I’m like, we’re doing this data project, so eventually we want things we don’t want to build. So many different connectors. Basically, they’re like, we can do the connectors. We don’t want to do that. So, so the hope is that either, it’s like, I guess when the time comes, it’s either like, okay, we’ve got the data and the data warehouse. What, what are brainforges? Forecasting capabilities. And then we’re like, checking on the vendor of Atomic versus this and figuring out which one we want to go with.
Me: Y.
Them: Or it’s like we’re just telling Atomic hold off until we have the data in a warehouse if we really like what they’re doing or something.
Me: Eah, for sort of like atomic look. If. They have a really great UI ux. Like, I just pulled them out. It seems like they’re great software. And I would suggest that they sit on top of the data that’s in the warehouse.
Them: Yeah.
Me: Directly to the source, you get into the same problem that we’re.
Them: We don’t want that. Yeah, we want it cleaned. We want it like.
Me: Like you should say. The data team will hand you the clean orders file or the transaction file with every single transaction by channel. That will be the gold standard for. For that.
Them: Yeah. That’s what I told them today. I was like, I think we’re gonna slow the roll on this until we, like, have data that’s really nice in a warehouse.
Me: We. Do.
Them: But. But also, I’d be curious to learn more about the people who do forecasting on your team and what they’ve done, because it’s like we’re not. We haven’t signed anything with them, so I think it’s just. I’ll just share that. The. That is a big pain point for the stakeholder. Dan, who does this regularly. He’s like it feels. Look, I ran supply demand at Brave, and it’s. It’s a weight. It’s a heavy weight to bear.
Me: A lot of.
Them: It’s like if you fuck it up.
Me: Confident. Oh, my God.
Them: If you mess up. Supply, demand. Everybody’s so pissed and you’re the one messing up.
Me: So this is. Robert’s entire background. Is in supply chain supply domain forecasting. He was that flex fort and then did this basically. Led the forecasting supply chain team at Ruggable. And so his whole background, again, forecasting, a lot of the things that we talk about, you know, one, we did learn a little bit of the model, but what we’re driving towards is the KPI we talk about forecasting is the distance between. What you forecasted in actuals.
Them: Yeah.
Me: That is ultimately like saying the quiet part, but, like, when you forecast, you’re trying to hit that. Right. And so when you talk about building a great forecasting model, we’re talking about minimizing the distance between those things.
Them: Yeah. And there might be, like, iterations of forecasting, right? Because it’s like, on one hand, it might be, like, supply, demand, okay? Atomic and, like. But then you have Laura talking today about how many applicants came in for wholesale.
Me: Yes.
Them: And her okrs around wholesale revenue. And like that could be its own, like, sort of mini version of like, how do we help somebody like Laura, who’s a PNL owner, understand her own trajectory of what wholesale could.
Me: And the more complicated a source. The less you’re going to find a software to do it. You know, like, everybody’s going to be like, shopify, whatever. But what you’ll find is if you have multiple Shopify or tags are messed up. They’re going to be like, oh, this is custom work. You need our enterprise plan. That’s what kind of happened. There are a lot of. I will have, actually. I’ll message Robert. I estimated a month or two ago, like, what’s the gold standard for forecasting software? You know, I think he has some recommendation, but this is also stuff that. We do for a few clients where we basically come in, look at, look at how forecasting is done, and then arrive at like, okay, here’s the money that’s on the table because you’re missing because of a wide gap between estimates. A lot of that should be. You can. You can totally drive from. From. You will be able to drive from the work we’re doing. I think it just depends on the sophistication and. Like the Gap right now.
Them: Yeah.
Me: Quite honestly to make a decision between us doing it or going with an off the shelf tool.
Them: Yeah. I need to learn more about what they offer.
Me: Y. Eah.
Them: What they do, but okay. Cool.
Me: Okay? That’s. All I have.
Them: Thank you. This was very thorough. Helpful. Overview of what’s been happening.
Me: Yeah. So I think what we’ll do is I’ll share the sec out. We’re just going to keep. This will just be a running deck, so we’re just. We’ll just make the next one right above it. And. Then. Yeah. So we’ll set up. I’ll send some summaries from stuff calls this week. We have the UTL calls next week. Can we go ahead? And did we end up booking. The call for the warehouse, I think on Tuesday, right? Or.
Them: Did I do that? Let me check. I like. I meant to. Let me see. So we did. We have that on Tuesday. Nobody’s accepted that invite, but, you know. Okay? Only calendar we should be all right. I’ll be there alone.
Me: Jason. Thank you.
Them: I just.
Me: Send a message in Slack and you’re like, Hey, can you. You might not give me an emoji or something?
Them: Yeah. Jason. Anything else on your mind? I’ve just been a listening part of this time. No, I think this is fine. I mean, obviously next week we’re gonna have to kind of, like, really kind of like, accelerate after kind of the conversations with polyatomic and the data warehouse. I know. That’s kind of what’s on my mind, is just making sure we can get your decision. Because we’re going to be going into the rest of the session. You know, just make sure that we’ve got, like, availability, everybody. One question from this is, did we say that we needed to get in touch with Emerson to ask the many things about BigQuery? Like, is that a key input for the Tuesday conversation?
Me: Yeah, Jason, if I could even just draft you a question to ask them, that would be really helpful for the two stay call.
Them: Yeah, go ahead and do that. And then I could pass that on. Perfect.
Me: And then I’m going to do my audacion on it. Also, in the Tuesday call, we’re going to have notes about the A piece in both of them. On. Like what we’ve seen so far within, like the data warehouses. I will preface with like. It’s just, like, all brand new, and a lot of it is marketing. But we’ll show you, if possible.
Them: Yeah, okay. Great. That sounds wonderful. Thank you, guys. This was great.
Me: Up.
Them: Okay, thank you. Bye.
Me: Five.