Highlight Bullets

> Here’s a glimpse of what you would learn….

In this episode, host Josh Hadley interviews Ellis Whitehead, co-founder of Data Brill, about how seven-figure e-commerce sellers can leverage AI and data to scale effectively. Ellis outlines a four-step “data chain” for success: centralizing data, capturing historical records, connecting disparate data sources, and building guardrails for AI. They discuss common pitfalls, the importance of solid data infrastructure, and actionable hiring advice for building in-house data teams. The episode emphasizes that AI is only as powerful as the data foundation supporting it, offering practical strategies for sustainable e-commerce growth.

Here are the 3 action items that Josh identified from this episode:

Timestamps:

00:00:00 Podcast Introduction
Leveraging AI and data for scaling e-commerce businesses.

00:00:58 Guest Introduction
Ellis Whitehead’s background and expertise in data, PPC, and Amazon seller growth are introduced.

00:02:00 AI Hype & Seller Challenges
Discussion about the overwhelming AI chatter among e-commerce sellers and the feeling of being left behind.

00:02:37 The Importance of Fundamentals
Ellis emphasizes sticking to business fundamentals despite rapid technological changes.

00:03:11 Common Data Mistakes in E-commerce
Ellis introduces the “data chain” concept and outlines common mistakes sellers make with data and AI.

00:05:07 Overview of the Four Data Chain Links
Ellis lists the four essential links: centralized data, capturing history, connecting data sources, and constructing guardrails.

00:07:29 Step 1: Centralizing Data
Detailed explanation of why a structured database (like Postgres) is crucial versus using spreadsheets or shared drives.

00:09:21 Technical Setup for Centralized Data
Differences between databases and shared drives, and why structure, speed, and reliability matter.

00:11:38 Non-Technical Founders & Getting Help
Advice for non-technical founders: learning, hiring, or consulting for proper data setup.

00:15:14 Ongoing Maintenance Caveat
Ellis explains that data systems require ongoing maintenance due to changing APIs and data sources.

00:16:45 Ways to Ingest Data
Different methods for getting data into databases: APIs, manual downloads, and handling multiple currencies.

00:19:15 Navigating Amazon API Access
Challenges and solutions for brands seeking Amazon API access, including using third-party services.

00:21:45 Step 2: Capturing History
Why historical data is vital for trend analysis, forecasting, and making informed decisions.

00:24:27 Use Cases for Historical Data
Examples of how historical data helps with leading indicators, seasonality, and strategic decision-making.

00:26:30 Pitfalls of Ignoring Trends
Dangers of relying on static data blocks and the importance of trend analysis for inventory and forecasting.

00:29:10 AI Automation Cautionary Tale
Risks of automating decisions without proper context and historical data.

00:31:01 Tracking Keyword Popularity Over Time
How tracking keyword trends can explain sales drops and inform campaign adjustments.

00:33:24 Step 3: Connecting the Dots
Combining disparate data sources to calculate advanced metrics and gain actionable insights.

00:35:53 Practical Tactics for Data Integration
How to use database views, scheduled calculations, and file storage for efficient data analysis.

00:37:05 Step 4: Constructing Guardrails
Building guidance and guardrails so AI can answer business questions reliably and avoid costly mistakes.

00:39:06 Guardrails in Action: Use Cases
Examples of how proper guardrails enable AI to deliver actionable, accurate reports and campaign strategies.

00:43:12 Building In-House Data Teams
Advice on hiring the right mix of technical talent or using consultants.

00:44:30 Three Actionable Takeaways
Summary of key actions: hire for data roles, let data drive strategy, and only use AI after building a solid data foundation.

00:47:38 Final Recommendations & Closing
Ellis’s final advice: start centralizing data in Postgres and set up guardrails for AI.

00:48:07 Book Recommendations
Ellis shares influential books: “Warren Buffett Accounting” and “1984.”

00:49:30 Favorite AI Tools & Workflow
Ellis describes his preferred AI tools and workflow: Claude, VS Code, TypeScript, Deno, Postgres, and git.

What is Git? (00:50:19)
Explanation of git as foundational versioning software for code and text files.

00:51:22 E-commerce Influencer Recommendation
Ellis recommends following George Meressa for advertising and e-commerce insights.

00:51:51 Where to Find Ellis Whitehead
Information on how to connect with Ellis and Data Brill for further help.

00:52:20 Podcast Outro
Closing remarks and call to subscribe and review the podcast.

• Josh Hadley on LinkedIn
• eComm Breakthrough Consulting
• eComm Breakthrough Podcast
• Email Josh Hadley: Josh@eCommBreakthrough.com

• “Data Brill“: “00:01:51”
• “ChatGPT“: “00:04:50”
• “Claude“: “00:04:50”
• “PostgreSQL“: “00:07:53”
• “Superbase“: “00:08:56”
• “Neon“: “00:08:56”
• “AWS (Amazon Web Services)”: “00:08:56”
• “Google Cloud“: “00:08:56”
• “Amazon Seller Central“: “00:32:08”
• “Helium 10“: “00:34:12”
• “Jungle Scout“: “00:34:12”
• “Amazon S3“: “00:37:00”
• “Visual Studio Code (VS Code)”: “00:49:30”
• “Typescript“: “00:49:30”
• “Deno“: “00:49:30”
• “Git“: “00:50:19”

Websites
“Databricks“: “00:51:51”

Books

• “The E-Myth by Michael E. Gerber“: “00:00:58”
• “Warren Buffett Accounting by Stig Brodersen and Preston Pysh“: “00:48:08”
• “1984 by George Orwell“: “00:48:36”

• “Search Query Performance Reports”: “00:24:03”

• “George Meressa“: “00:51:22”

• “Hire for Data Management Roles”: “00:44:30”
• “Base Business Strategies on Data”: “00:45:25”

This episode is brought to you by eComm Breakthrough Consulting where I help seven-figure e-commerce owners grow to eight figures.

I started my business in 2015 and grew it to an eight-figure brand in seven years.

I made mistakes along the way that made the path to eight figures longer. At times I doubted whether our business could even survive and become a real brand. I wish I would have had a guide to help me grow faster and avoid the stumbling blocks.

If you’ve hit a plateau and want to know the next steps to take your business to the next level, then email me at josh@ecommbreakthrough.com and in your subject line say “strategy audit” for the chance to win a $10,000 comprehensive business strategy audit at no cost!

Transcript Area:Josh Hadley 00:00:00  Honestly, I’m not going to nerd out on all the data I can’t like. This sounds like going to the dentist for me, however. Oh no. I know the importance behind it because this, if set up correctly, is what my team will leverage to move ten times faster.

MC 00:00:14  Welcome to the Ecomm Breakthrough podcast. Are you ready to unlock the full potential and growth in your business? You’ve already crossed seven figures in sales, but the challenge is knowing how to take your business to the next level.

Josh Hadley 00:00:28  If you think AI is your competitive edge in e-commerce, it actually might be your biggest weakness because if you feed it messy, disconnected data and it’s going to start scaling your mistakes faster than you can spot them. Today, we’re going to be diving into the proper way to leverage AI and ensure it’s leading you in the right direction in your e-commerce journey. Welcome to the Ecomm Breakthrough Podcast. I’m your host, Josh Hadley. I scaled my own brand from 0 to 8 figures in sales, and now my mission is to take it to over nine figures on my journey to nine figures.

Josh Hadley 00:00:58  I bring you the unfiltered conversations with some of the smartest minds in e-commerce. Past guests include Ezra Firestone, Kevin King, and Michael E Gerber, author of the E! Myth. I’m giving you the real strategies and systems that help set you up to scale. Today on the Ecomm Breakthrough Podcast, we are joined by a true expert at the intersection of technology, data, and e-commerce growth. Ellis Whitehead is the co-founder of Data Brill and a leading mind in the PPC management, data science and business intelligence space with a PhD in automation and years of experience architecting smart technology for Amazon sellers. Ellis has helped brands leverage data driven strategies to scale profitably and stay ahead of the competition. He’s here to share how sellers can use advanced analytics and AI to break through the seven figure ceiling and unlock the path to eight figures and beyond. With that introduction, and welcome to the show, Ellis.

Ellis Whitehead 00:01:51  Thank you very much, Josh. Happy to be here. Thanks for having me.

Josh Hadley 00:01:54  Ellis, I’m super excited to dive into this right now, because here’s what I see a big problem in right now.

Josh Hadley 00:02:00  I’m in a number of mastermind groups. Yep. And the WhatsApp messages, the Facebook group messages, and also the, you know, the Facebook groups, they’re all just blowing up with people talking about AI. And it honestly feels like I can never keep up with enough. And I feel like I’m behind. It doesn’t matter how how much I’m leveraging AI, I just still feel like I’m behind so.

Ellis Whitehead 00:02:24  Everybody feels that way.

Josh Hadley 00:02:25  Yeah. Is that how you feel and what do you think? You’re speaking to a bunch of seven figure ecommerce sellers here. What do you think they need to be hearing today? In the age of AI? That seems to be changing on a daily basis at this point.

Ellis Whitehead 00:02:37  You know, you’re going to know this from all of your work with Amazon. There are constantly changes that are coming out, but one of the things that always stays the same are the fundamentals. You know, your what what makes something profitable is the same now as it was when accounting was invented.

Ellis Whitehead 00:02:56  You know, so trying to keep this balance of maybe 80% fundamentals and 20%, you know, new exploration that can when things are moving really fast. You know, if you know, well, I’ve got 80% of it down anyway. That can really help.

Josh Hadley 00:03:11  Yeah. No, I love that. Well, you have deep experience there. So as you’ve been working and coaching other, you know, e-commerce clients and helping them integrate their data into AI, what do you see some of the biggest mistakes that people are making right now in the e-commerce space, trying to leverage AI?

Ellis Whitehead 00:03:29  Well, I think what I would like to do today is try to break down what I call the data chain. So I look at this as like four interlinked links of a chain that you need to have in place before analytics, and especially before AI agents can consistently help you make better decisions. So when teams struggle with data, it’s rarely because they’re missing one dashboard. It’s because the system is missing a step. And the four links of the data chain, their sequential dependencies, and if one link is weak, everything downstream is going to be a struggle.

Ellis Whitehead 00:04:02  And so yeah. Let’s walk through some of these chains from the ground up. I guess that’s what I would set as the the basis for the discussion today.

Josh Hadley 00:04:10  Yeah. Ellis I think that’s super important. I mean, when, when LM tools first came out, we all talked about how, hey, it’s only as good as the prompt that you give it, right? And now it’s kind of improved to where like, all right, prompting yes is still important. It’s, you know, garbage in, garbage out. However, I think like you’ve hit on a key aspect here, I think there’s a lot of e-commerce sellers that are trying to incorporate AI, but they’re just trying to tie it into either ChatGPT or clause. Exactly and things like that, but like it needs to sit on top of like your knowledge base, not just like the world’s knowledge base of e-commerce and the world’s knowledge base of data. So I’m really excited to hear how you’re.

Ellis Whitehead 00:04:49  Working.

Josh Hadley 00:04:50  On detecting the data that sits behind it.

Josh Hadley 00:04:53  So the AI just sits on top of your data. The right pieces of data so that it does point you in the right direction. So with that, Ellis, why don’t you kick us off? What are the four essential links that we need to have for our data before we leverage AI to its full capabilities?

Ellis Whitehead 00:05:07  So I would call the first link centralized data. Okay. So this is where you’re going to be putting data into a structured database. And then what you need to do for the second link is capture history. And so in general you want to have at least one year of history that you can work with, preferably daily history. And we really like to have three years or so for our clients. But you need to go back at least say one year, 15 months. The third link we call connecting the dots. And this is about bringing these disparate sources of data together because so much of good insights require a calculation that is not in a single report? We’ve got just the most basic one, for example being tacos here.

Ellis Whitehead 00:05:58  Total advertising cost of sales. There’s no report that says tacos out of Amazon. But there are a lot, you know, there are a lot more. And then for the fourth one this is constructing guardrails. And this is it’s more than constructing guardrails. It’s also constructing the guidance that lets the AI know what it should be doing. And there are elements to this that I rarely hear other people talk about. And so we’ll we’ll get into that. And so those are the four.

Josh Hadley 00:06:30  Yeah I love that. So I think what you’re saying here there’s, there’s these four fundamental links. And what we’re going to dive into is like I would love for you to describe to our listeners, this is how you could actually set this up. And I think one thing that you’re cluing us in on, on Amazon, everybody knows this. There is 1,000,001 different reports. Yeah, right. So what Reports are actually meaningful. Which ones do we need to actually build into this like data warehouse or storage space. Right.

Josh Hadley 00:06:56  And then how are we going to like, connect those dots? Because you’re right. Like you can’t just download the PPC report. That’s not going to have everything you need. You need to include, like your other sales reports from Amazon that they provide, you need to marry those together. So, Ellis, I want you to walk me through like I’m a third grader. I have a third grader. So walk me through like you’re talking to my son at a very elementary level so that we can help these sellers, like, know how realistic it is to actually begin building this in your own dashboards and your own data warehouse. So let’s start with step one.

Ellis Whitehead 00:07:29  Okay. So step one with centralizing data, the first step is actually getting a database. So normally when people start trying to do this what they do is they want to they end up downloading reports and maybe putting it into a shared drive folder or something. That’s not going to work. It is not sufficient. It’s not scalable. The analysis is a pain.

Ellis Whitehead 00:07:53  You really need a proper database. And when it comes to the different database types. Postgres is the name of a database system. It is so far superior for this use case. There’s no reason to even consider anything else. And fortunately, there are a lot of good services that are based on Postgres because it’s so good. This includes super bass, which, you know, most people who are working with AI agents and trying to build things they’re very likely to have heard about Super bass. I’ve been using super bass for a lot, many years, well before there was any hype on it. It’s a good service. There’s also another really nice one called neon that has different advantages. it’s a lot more expensive though if for this use case of like Amazon data constantly going into it. So I wouldn’t consider neon for this. I would either go super bass or if you are more technical and you already are set up on AWS or Google Cloud or something like that, you can just start Postgres on there.

Ellis Whitehead 00:08:56  All of them support it well, and if you’re much more technical and you have your in-house systems, then it is so easy. You just start up. It’s easy to start up a Postgres database in your own, in your own architecture. We probably have 1212 or so, maybe more Postgres databases running on on our system here. Okay. Yeah. So that’s the first thing. Just commit to getting it into a real database.

Josh Hadley 00:09:21  And what’s the difference? Why why is this so important compared to you had mentioned earlier like just a shared Google Drive or something like that. What’s the difference with Postgres or super. Super base.

Ellis Whitehead 00:09:32  Super base. Yes. So the difference there are three main differences. One of the big differences is that databases force you to structure the data. And so this is of course a constraint. It’s a it’s a pain if you have to you know, it’s an upfront pain. You have to structure what’s going to go in there or say what the structure is. But the advantage is that since it’s structured, software can access it in a completely predictable manner.

Ellis Whitehead 00:10:01  And that makes that makes all the difference when it comes to reliability. But then also this is then the second thing to speed. It is going to be 100, 1000. It’s probably 1000 times faster. And so if you’re trying to query things that are a little bit complicated, that involve going out to this table or that table and bringing them together, or you’re trying to get stuff for you, like you want to look at a the trends over a whole year of sales. If you have all your data in spreadsheets, you’re going to have to be opening one spreadsheet after another. This is time consuming for you personally, but also even for software, it’s a slow process to go out and open a bunch of spreadsheets. It’s reliable and it’s fast.

Josh Hadley 00:10:46  Yeah, I think you just hit on something that I had like a brief mind unlock regarding how many times I’ve just, like, tried to use, like ChatGPT or Claude and I just, like, upload a PDF document or upload a whole bunch of like, data.

Josh Hadley 00:11:00  It’s like, hey, I got this report, here’s my Amazon PPC data, tell me what I should do. And it’s like, okay, and guess what? AI is going to go give you an answer.

Ellis Whitehead 00:11:09  But I think.

Josh Hadley 00:11:10  What you just touched on is you are because it’s going into a database, you’re giving it structure and you’re saying this means this. And so when the software and the machine learning is reading that data, it actually knows what it’s talking about. Whereas if I just upload the the report I just downloaded from Amazon, it’s like I think this is this. So I’m going to tell you this, but I might be wrong.

Ellis Whitehead 00:11:32  And so and it’s definitely regularly wrong about how it interprets the columns. That’s that’s true.

Josh Hadley 00:11:38  I love it. So I guess my question for you, Ellis, already as we start dive into this, you’re talking to a non-technical person. I’m very much so like the visionary. I am the leader of the business.

Ellis Whitehead 00:11:50  Yep. It’s great.

Josh Hadley 00:11:51  Is this something I can do myself? Like, how complicated is a lot of this stuff, especially as you talk to a broad audience of e-commerce? Yeah.

Josh Hadley 00:11:58  It’s like, is this stuff that you would say you could do this on your own? Or is this like, hey, go find somebody off of Fiverr to help set this up correctly for you?

Ellis Whitehead 00:12:06  So it’s a great question. I wouldn’t find anybody off of Fiverr. I don’t think they wouldn’t do it right. At least it’s not. Not in my experience. And so, so much. There’s definitely complexity to it. But you know also how complex Amazon is, right? I mean, whether it feels complex or not has so much to do with familiarity. And so if you have some help or you’re just, you know, sufficiently motivated and you start, you know, a step at a time. And so you just say you get your super base instance, you say, I’m going to put one report into there, maybe two reports, and I’m going to start some script that gives me useful information that either saves me time or gives me insights that I don’t have now. And if you can get that so that it’s easy, that already is proof for you to motivate you to, you know, overcome maybe the next complex step.

Ellis Whitehead 00:13:00  And by the time you do that, then this last one won’t seem complex anymore. And so this is yeah, it’s it’s stuff to learn. One can definitely it can definitely make sense to get help along the way. But it’s it’s doable. And you know, my business partner Danny McMillan, who you know, he was not technical at all. He started out as a DJ for years. And the stuff that he has been building over the last 12 months, he is he is Veritably become an engineer without actually becoming an engineer. It’s, you know, just because he’s become he’s been motivated And he’s become sufficiently familiar with with all of the tools.

Josh Hadley 00:13:41  Yeah. No, I do agree that, you know, with a lot of this stuff, it all comes down to like, how much do you want to learn this yourself? But I think the important thing here in any entrepreneur that is maybe in my shoes, where it’s like, I’m not the tech geek and honestly, I’m not going to nerd out on all the data I can’t like.

Josh Hadley 00:13:58  This sounds like going to the dentist for me, however. Oh no, I know the importance behind it because this, if set up correctly, is what my team will leverage to move ten times faster. And instead of just using custom GPT and uploading it, some random reports that sit in the background. If you set this up properly, this is this genuinely is the tool that can help you connect your business. The one caveat I would say is, all right, if you’re not the tech geek that wants to dive into all of this, what I would say is you go find really good consultants like yourself, Ellis. And that’s been one of the best hacks that I’ve had in my own business is rather than me going to go learn something, I’m going to go pay an expert to go teach my team how to go set this up properly so that I have the right people on my team, and you buy a consultant that already has been there, done that, can show you all the watch outs and things like that for your business.

Josh Hadley 00:14:53  So that’s my hack that I would give. And my secret tip to those listening, if you’re like, there’s no way I’m doing this myself. Good. There’s people like Liz that are happy to instruct you, teach you, teach your team how to go set this up properly. And the good news is it’s kind of a one time setup. Like you need to choose your database one time, then go set it up and set up the data structure that sits behind it.

Ellis Whitehead 00:15:14  Right. I will caveat. So first of all, before I add the caveat is, you know, that’s really smart. Everybody should be be doing that. I do that, you know my wife and her business. She does that. It does not make sense to spend months and months trying to learn stuff on your own when you could spend, you know, the equivalent of maybe an employee’s month. Pay to teach your whole team. And. Yeah. And the caveat is it’s not actually set up one time. So anything involving software has a what’s the word like maintenance ongoing maintenance requirements.

Ellis Whitehead 00:15:50  Because you know Amazon they do occasionally change the columns in their reports. And so that ends up breaking your process for ingesting it into the database. Or one of the things that happens around prime days, for example, is their APIs all slow down and their advertising data is maybe like 24 hours behind, or they’re advertising reports or 24 hours behind. And if you know, you set up your system in a way that did not anticipate these problems, you know, they’re going to be certain kind of like emergencies that’ll come up like that. So I wish it were set once and do it forever, but it’s set once and, you know, take care of the problems as they arise.

Josh Hadley 00:16:31  Yeah. No, that that does make sense. It’s just like everything in the universe there is entropy. And the moment you build something, it starts to fall apart.

Ellis Whitehead 00:16:39  So that’s so true.

Josh Hadley 00:16:40  I love that. Well, we’ll dive in further into like, who the right people are to kind of manage this architecture.

Josh Hadley 00:16:45  But let’s move into step number two.

Ellis Whitehead 00:16:47  Actually, can I say something I think would be good to say a couple more points about step one here. This is with, you know, getting setting up your database and centralizing the data. There are different ways to get the data into the database. The most efficient way is if you do it via APIs. But of course getting it in through APIs. That definitely requires either a software service or a good amount of expertise on your team or for yourself. And the Amazon APIs do tend to be pretty painful to use. So if you can get them set up or use some sort of a service for it, that’s a good, good way to go. The other way, though, is you might have something set up to either you manually download reports or you have it somehow quasi automated. You know, even if it’s like clawed using puppeteer to open seller central and download the reports, if you download them to a you know the right or a predetermined directory, you can then have ingestion scripts as they’re called to take those reports it sees are their new reports, their uploads them to into your database, and then moves the files to another location so that the next time the script runs, it doesn’t, you know, it knows it doesn’t see those files and try to ingest them again.

Ellis Whitehead 00:18:04  And so depending on how you go about it, you’ll want to do one of these two processes. And and it could be that you do a mixture of them some some things you do through APIs. Some things you download the reports manually. Of course, there’s a lot of data that’s only available through APIs. So if you want that you need to do it. And then if if you need things like that’s not on Amazon, do you sell in Canada? Mexico? Anything that doesn’t use US dollars.

Josh Hadley 00:18:35  We do with a naff program. Yeah.

Ellis Whitehead 00:18:37  So then you’ll be getting reports in, you know, with Mexican and Canadian currency as well. And it could well be that you want to know the conversion rates. You want that in your database. And so this would be then of course something that’s not on Amazon. It’s not hard to get, but it’s something you know to keep in mind. I would say most sellers that I work with do have to deal with multiple currencies. And if they have the currency conversion rates in their database, you know, that can help out a lot.

Ellis Whitehead 00:19:07  If they’re using this for some form of financial reporting, or just to see everything in the denomination that they they’re most interested in.

Josh Hadley 00:19:15  How difficult is it? So I know we’ve tried one time to go try to get like API access to start pulling reports. And you know, it’s it’s easier said than done. Do you have any like, like tips or secrets for, you know, maybe a brand owner like myself. We don’t have a software company. We are a brand, a brand first. But like when you try to request the API, it’s like, share your data protection policies with me, show me your like articles of organization, blah blah blah. Like it goes down the list. And because yes, they’re not just like giving you full carte blanche access to the API without having some type of structure. So how do you like how do you navigate that if you’re if these brand owners want API access?

Ellis Whitehead 00:19:56  Oh, there are there are two possibilities. Well, let’s say three possibilities. One of them is just take a look at that and see.

Ellis Whitehead 00:20:03  That’s too much. It really is a pain. And the pain doesn’t just end once you’ve got access. There will be change there changes to access requirements every few years. And the APIs are constantly changing. If you’re going to go the API route, you really want to be sure that you have expertise among your employees in your team, that you can, you know, actually delegate their time to this. You know, you might have the expertise, but it’s not where you want them spending their time. So you have to have that available. And for most of the clients and companies we talk with, that is not actually a great option. Even if it’s a large company and they have their software development teams, their software development teams are involved in other software, and they can’t just jump over to Amazon whenever something needs to be taken care of for Amazon. And so if you’re specifically an Amazon company, maybe you do like one of our one of our clients, he has a an offshore software development team that he’s been using for years, and he’s reasonably happy with that.

Ellis Whitehead 00:21:12  And of course, there are others who have their in-house teams. So that would be the that would be a possibility. And the other thing is, there surely are services out there that can help you with this. I know we do this with our clients feed, you know, stream the data into their databases for them from Amazon. And and since we’re so much focused on Amazon, you know, we when there is something that needs changing, we can we can get on to changing it quickly.

Josh Hadley 00:21:39  Love it. Awesome. So again working with experts that know the space.

Ellis Whitehead 00:21:43  Yeah.

Josh Hadley 00:21:43  All right. Let’s hear step number two.

Ellis Whitehead 00:21:45  Okay. So this is where we want to capture the history. There are two two aspects to history. One of them is sometimes you can just request a report that goes back a year or two years, whatever. You can get order data going back for. I guess it depends which reports you’re using. You can get them going back at least 18 months, depending on the report, sometimes more, and add data you can get going back.

Ellis Whitehead 00:22:10  Used to most of it. Used to only be two months. But that changed in November last year to my great joy. And you can get a lot of data now going back just over 12 months. And this is it’s enough to start with. Even though we like to see three years of data, it’s enough to start with. So that’s the first thing. Sometimes you can download the history for a long period of a longer period of time. There are a lot of things where you can’t do that, where you only get the current snapshot. And so part of capturing history is just making sure that you’ve started the process and you’re maybe capturing your BSR numbers, you know, a few times a day, and there’s no way to go back and get old BSR numbers from Amazon. And there’s a lot of data like that. There’s also the streaming data where you can get the certain ad information on an hourly basis for, for what’s publicly available. If you’re not grabbing this right now, you’re going to miss it.

Ellis Whitehead 00:23:07  So it’s a you know, it’s streaming data. You can’t request the history of it. They do have a beta program for getting history as well, but it’s hard to know what’ll happen with that. And the reason for history. Sorry, I should have started with this. So being able to visualize and comprehend your data over time. That’s what I mean by capturing history. So without history, you can’t identify trends, you can’t project forward. You can’t even say whether what you’re seeing now is an improvement over before, unless you just happen to remember it. And so, for example, the search query performance reports that you get from Amazon, if you’re just looking at the reports, they just show a single period of time. You’re either looking at the aggregated data for one week or one month or one quarter, but there’s no way to see the time series of how these things been changing on Amazon. If you get it into your own database, then you can visualize it. You can. And you know, picture is worth a thousand words when it comes to trends.

Ellis Whitehead 00:24:03  Being able to see the line graphs of what’s happening with the search volume or with the overall like total marketplace orders on keywords is super powerful. And so that’s that’s another major motivation for getting the data in there. What for history. But then on top of that, building your graphing facilities capabilities so that you can see visually see the the trends.

Josh Hadley 00:24:27  I love that, and I think when you first started the episode, you said, you know, one year at a minimum, but ideally, if you can get to three years of data, it really helps smooth things out and really helps you like begin to recognize patterns. You know, I think this is a really important list. I, one of my most recent podcasts that I put together was Measuring Your Leading Indicators, and it’s the CEO dashboard. And one of the things that you need to be looking at as like a leading indicator before it becomes a lagging metric, like your net profit in the business, is what’s happening to your conversion rate year over year.

Josh Hadley 00:24:59  Are you increasing in your conversion rate for the same product this same time last year? Right. And again, it’s arbitrary for me to say, like, I have a better conversion rate this month than I did last month. And it’s like, well, we’re not comparing apples and apples here. We’re comparing apples and oranges. Like, I may be at the peak of Q4 when I’m looking at my conversion rate, smashing it right. And then I look in the following month and it’s like, oh no. My conversion rate is terrible. And it’s like, well, hold on, look at that same conversion rate that same time period last year. And those are like the leading indicators that if you’re focused on that, they’re going to flash warning signals to be like, hey, we’re starting to trend downward here. Maybe you want to take a look at your conversion rate and vice versa. Maybe you don’t need to. Right. But the important aspect of this data is it leads to understanding what actions you need to take in your business.

Josh Hadley 00:25:51  We’re not just talking about data just to understand like how much money you’re making or things like that. It’s so that you, as the CEO, can know where to invest your time and your team’s time, because there are 1,000,001 different things that your team can be focused on. Your important focus as the CEO is to set the strategy, and business strategy is nothing more than taking unlimited opportunities with limited Resources and finite resources in being able to prioritize. This is what we’re going to attack instead of the other 999 other things that we could do.

Ellis Whitehead 00:26:26  Great point.

Josh Hadley 00:26:27  Alice, is that step two, are we ready for step three?

Ellis Whitehead 00:26:30  You know, actually there’s some more stuff on step two that maybe, maybe would be good to point out. I mean, about what it is that where history becomes really useful. So like when you come into a new season for a one of your products. So let’s say we’re talking about spring, the spring season starting. And so you have the first warm days of weather, and then maybe you have sports or outdoor equipment or certain clothing.

Ellis Whitehead 00:26:56  And so sales start to climb. And so you might be looking at your sales numbers going up and think great, we’re growing. But at the same time, it might actually be that your BSR and your subcategory is actually getting worse. And so what this means is that you’re losing market share. Your competitors in the subcategory are growing even faster and without having this like BSR history and being able to see what’s happening with it continuously over time. You may be celebrating rising sales without realizing you’re actually falling behind. And there are there are a lot of things you can misread market share. You can misread demand. Competition inventory, inventory time, inventory time over inventory turnover. So this, like you were saying, with might be the end of quarter two if you’re just looking at a period and the sales that were happening in that period and you’re not looking at the actual trends that were have been happening and are leading up to today, then, you know, if if your spreadsheet says that you should be ordering a tremendous sending in a tremendous amount of inventory in January for Christmas sales products, just because your December sales were so high, that, of course, would be a terrible mistake.

Ellis Whitehead 00:28:07  Nobody does that because it’s so egregious, but it’s done in so many cases. I would say most most sellers I talk with. They have their their Excel spreadsheet set up in a way that they just trust that whatever it says is accurate. But it’s it’s always based on a block of time within which the actual trends may have changed. And and so this is a, you know, a way that if you’re using if you have the trend data in your database, you don’t have to just rely on these big some numbers that are in the spreadsheet. And if once we get around to plugging AI into here, of course it can do can help with programming all kinds of calculations, it can be much more sophisticated and realistic for projecting things.

Josh Hadley 00:28:50  I love that I think those are some fantastic use cases I have heard in, and I’ve even seen people talking about this, raving about it on, you know, in some of these mastermind Facebook groups, etc.. Here’s what they’re doing. They’re like, oh, I’ve got open class set up or I’ve now got, you know, Claude doing things for me.

Josh Hadley 00:29:10  I’ve got all these agents. And now my process is I don’t even have to look at anything. I’ve got an agent that goes and looks at the data, and then it sees that I’m running low on inventory. And so it then goes to this other agent that goes in places the Po with my manufacturer and then make, and then that goes over to my accounting team to make sure that we pay them on time. And guys like that’s that’s the future. And I call BS to all of us.

Ellis Whitehead 00:29:35  Oh, that is so beautiful.

Josh Hadley 00:29:37  Is that something that becomes a thing? It only becomes a thing if you have engineered it to work to precision. Because what’s going to happen is you just hit the nail on the head. Oh, wow. We sold 10,000 units last month. We only have 1000 units left. Well, this was a seasonal decoration for Christmas. Your LM, clawed or open claw is going to be like. Oh, well, you told me we’re running low on inventory. This doesn’t match the last 30 or 90 day trend.

Josh Hadley 00:30:05  So now I’m going to go order another 10,000 units for you. And in fact I’m going to expedite it. And in fact, I’m going to er ship it to you because like that’s how smart I am. I see that you’re running low. So here it comes. And now you just have this compounding problem. So yeah, that’s why I think this conversation is so, so important. Everybody’s so excited about AI and I myself am excited. But I’m cautiously optimistic to say it’s only helpful if you structure it and give it the proper context. And then guess what? You’re still going to have to, like, go through and be like, this isn’t still isn’t giving me the right output, and you’re going to have to troubleshoot it and train it just like you would a full team member. And it takes about a year for a full team member to like, start really kind of getting it. Quote unquote. Well guess what. It’s probably the exact same thing with AI. So I just love that use case that you said with with inventory.

Josh Hadley 00:30:54  That’s the fastest way to bankrupt yourself as a e-commerce brand owner is put all your capital into inventory and go out of business.

Ellis Whitehead 00:31:01  Another example for where history is great is tracking keyword popularity. So this is something that I rarely see done. There’s the keyword popularity report that you can get from the on the brand analytics page in Seller Central, or as we do download it through the API. And so we get this. We always download the daily ones and the weekly ones. They have different advantages. The weekly reports are much more complete, but being able to look at keyword popularity on the daily level is very powerful. And so Amazon, what happens with keyword popularity is it’s very much influenced by Amazon search suggestion list. And so Amazon refreshes this list different keywords at different times. You know every four weeks. Three it varies but it refreshes it at a regular interval. And when that happens, keywords that were high demand for a long time can suddenly just drop out and new ones appear in their place. And if you’re not tracking this over time, what you might see for, let’s say your product was selling a lot on one of these keywords that just dropped out.

Ellis Whitehead 00:32:08  You’ll see. You’ll see a big sales drop and you don’t know why unless it’s picked up equally by the new ones. And we’ve seen this a number of times. Sellers been doing great on a particular keyword. That keyword drops out of Amazon’s search suggestions and their sales fall off. And if you’re tracking keyword popularity, you can quickly see why. And you can look at what new keywords came in to replace it, and then assess whether those will convert well for your product, too. And then you make the shift. You update your campaigns, adjust your listings if needed. And and there is one extreme case that comes to mind where the keyword that was replaced, it was replaced by a competitor’s branded keyword. And so there was no way to rank well on that anymore. And the product just stopped selling as much. But the valuable thing was we could see why immediately. And when you know, a decline is outside your control, you stop throwing money at it and focus your efforts elsewhere. And in that case, the situation did eventually turn around about eight months later, the keyword list changed again and the product picked right back up.

Ellis Whitehead 00:33:09  And so, you know, just another case of there are all kinds of different reports and metrics that are very useful to track the history of. And so these were a couple of examples.

Josh Hadley 00:33:20  Love that. What a great example. All right so let’s go to step three.

Ellis Whitehead 00:33:24  With step three we want to connect the dots. And this means bringing together disparate data sources to calculate additional insights and other important metrics as gives us the higher level insights. So currently you might see order data on for example, your Amazon seller app and your advertising data in the advertising console. And Cogs are in a spreadsheet and the search query performance reports you have open in some browser tab and connecting them to answer real questions like is my top of search spend on this keyword actually profitable? Well, that’s too tedious to even attempt to do regularly. And so sellers end up looking at each data source in isolation when the real answer only emerges at the intersections of them. Many metrics that actually drive decisions. They require these multiple sources being combined.

Ellis Whitehead 00:34:12  Otherwise, you simply can’t calculate them. There’s, of course, tacos. There’s your true profit margin there. You have to combine your fee reports with Cogs and sales data, and there’s estimating competitor order volumes. So this of course is something you can go out and use software like helium ten and Jungle Scout and others. And that’s what you know, most sellers will rely on. But there are various ways you can do it yourself without their limitations. The easiest way requires combining catalog report listings, where you get information about BSR on an ongoing way, but also parent child relationships. And you can take this for your products and competitor products. And then you take your own order reports and you’re able and it’s possible to join these in a way that you can then do what’s called regression, you know, certain kinds of a way to be able to interpolate and extrapolate what the sales are for your competitors as well. And and so it’s a lot. If you have this set up like we do for our clients, if you have this set up, then you have so much more flexibility and agility.

Ellis Whitehead 00:35:18  Rather than, say, starting tracking this thing on a competitor in Jungle Scout, but you only start getting it. Maybe you know from the date that you, you know, start the data and you can just, you know, depending on how much you want to spend on the API calls, you can track as many competitors as you want. Yeah. Some some examples of connecting the dots.

Josh Hadley 00:35:37  Love it. No, I think there’s a lot of business cases that go into that. And as you mentioned, it’s all just about like how much do you want to pay for the API calls? But there’s an infinite number of things that you can be looking at there. So great. Is there anything else to add on step number three? Are we ready for step four here, Ellis.

Ellis Whitehead 00:35:53  Let’s see if I were to give something of a tactical playbook for this. I’d say if sellers are just getting started with combining data, the There may be three main ways to approach it practically. One of the main tools will be creating new objects in your Postgres database called views.

Ellis Whitehead 00:36:09  These these are basically like tables, but they let you join tables together without taking up any additional data storage space. The next tactic would be scheduling calculations. Sometimes you can do this right in the the Postgres database itself, but it doesn’t have to be. It can be anywhere. You schedule calculations to create custom reports, and these would be ones that take too long to just pull up spontaneously. You know, maybe it takes a minute to calculate. You don’t want to do that when you open up the dashboard in the morning. And so you have it done at four in the morning instead. And then the calculations right there. And the third one is you might want to start using S3 for file storage. S3 is a very widespread file storage service. It was started by Amazon, but it’s like provided all over the place. And if you’re using Super Base, they have it conveniently built in for this as well.

Josh Hadley 00:37:00  Fantastic, I love it. Those are great use cases. All right, Ellis, let’s move to step number four.

Ellis Whitehead 00:37:05  Step four constructing the guardrails. So this fourth link is about being able to answer your own questions on demand without waiting for some dashboard software to build the right view for you. Okay, so when you need an answer, the ideal aim is you should be able to go get it. This will never always be the case, but let’s say you have stuff set up so that you can get an awful lot of them. And this is where AI becomes genuinely powerful, but only if you’ve built in the right guardrails. So without them, you know we’ve already mentioned this so many times it can go wrong, it can hallucinate, it can misinterpret things. It might not be following best practices and use the wrong way of analyzing. And it can just make this will happen so much. It’ll just make things up about your products. If it’s product specific questions that you have, you know these mistakes, they’re subtle, but they’re also expensive because the AI failures, they look plausible, right? You read what it says.

Ellis Whitehead 00:37:59  It all sounds like it was well written.

Josh Hadley 00:38:01  It’s all A’s. Very smart. And it’s very confident.

Ellis Whitehead 00:38:04  Very confident. Yes. Very convinced. The whole promise of AI for your business is speed and scale, but without guardrails, that speed just means you make bad decisions faster as you, you know, to repeat what you had said. An example of where it didn’t, where AI was not able to work correctly, and then we were able to get it to work correctly. By putting in these guidelines and guardrails. One of our clients, they they have a moderately successful listing, and they brought in a new color variant, and they wanted to test shifting advertising to it because they felt their like intuition was this is going to do better than the others. And so we we had this set up for advertising and we prompted the AI to track the before and after comparison, total sales, conversion rates, click through rates and which aces were actually being purchased. After the whichever one was clicked, we shifted most of the advertising to the new Ason, but not all of it to still have comparisons during the same time period.

Ellis Whitehead 00:39:06  And so over a couple of weeks, the data clearly demonstrated that performance on every metric was favorable higher sales, higher conversion rates, higher click through rates. And the AI was able to put this in such a clear report. I was so impressed. I would have been impressed if a human had put this together. I was very impressed that the AI was able to put it together, and it’s not something that I’ve ever seen on any dashboard. There’s no dashboard in the world that is for that specific scenario and really demonstrably showing why we can be confident in moving forward with this new variant color being the primary one to be advertised. And so that’s what becomes possible when the guardrails are in place. You can ask completely custom questions and trust the answer.

Josh Hadley 00:39:47  Love it. Great example.

Ellis Whitehead 00:39:49  And then there was this other client that we came to. And this is one where the guardrails were not there to begin with or the proper guidance wasn’t there. It turned out to be an extremely interesting project. So he has hundreds of parent listings across 1012 different stores, many languages, thousands of SKUs.

Ellis Whitehead 00:40:07  And they’re pushing really hard to build sponsored brand campaigns across their entire assortment. So the problem is, how do you choose targets for that many campaigns, especially if you’re trying all the new different kinds of sponsored brand campaigns that you can do nowadays. And so they started with Amazon. Suggested keywords was what they were initially doing from inside the advertising console. This ended up with thousands of targets and thousands of campaigns, and this was causing massive slowdowns of the API to upload and download the data. It was just overwhelming. Amazon actually. And Amazon advertising actually works better with fewer, more meaningful targets that can accumulate ad history anyway. And so this was just not the right approach that they had been taking. And so we initially tried to use AI to set up the targets, and the results were frankly, crappy. Okay, part of it, of course, is the many different languages, but even just focusing on English, it it still was too confused. And so what we built and I can strongly recommend this to anybody who’s gotten a little bit further along in this process of centralizing their data, is we built the ontology on top of that.

Ellis Whitehead 00:41:18  So ontology is a product classification that’s formal so that software can understand it. And so it’s just picking out what are the important properties and the allowed values for those properties across their entire catalog. And then having this for each product group and each ace and variant, just the additional variant properties. And with those in place we could take the clients search query performance reports, their advertising history, and the search popularity reports. So all of these three together now and gather all possible relevant keywords and pick the most promising ones with the help of the AI in a reliable way. Because we did it on for each product with the known ontology, and this was simply not possible before we had all those reports together with the ontology, and we also built some keyword relevance tables to help make it more efficient. On top of that, this task was just too massive for people to do, and when the AI tried to do it without that structure, it was making too many mistakes to allow it to work. And so humans couldn’t do it.

Ellis Whitehead 00:42:26  AI couldn’t do it. But by putting in the guidance and the guardrails, it was literally done in 15 minutes once it was all in there. And you know, this is just another amazing case of, you know, where you’re just wow.

Josh Hadley 00:42:39  Yeah. No, I absolutely love that. And I love the different use cases you’ve been sharing here where, yes, AI is a great tool. We should be using it, but only when you have it set up properly. That’s the key theme that I’m getting here now. Ellis, we’re running up on time here. There’s one thing that I definitely want to ask you, though. If I want to build this in-house, what are the right titles of people that I need to hire for? Am I looking for a data scientist? Am I looking for a software engineer, or am I looking like, what are the roles you think I should be looking for if I need to build this in-house?

Ellis Whitehead 00:43:12  So it’s a it’s a great question. You know, you can sometimes just get lucky and have a find a motivated 19 year old who has no qualifications except that he’s he’s a passionate software geek programmer.

Ellis Whitehead 00:43:24  Sometimes that that’ll work out great. I don’t think there’s a particular title that I can recommend here, because it’s really across a large swath of, of skills that are necessary databases, a database administrator. Okay. You could you could get a database administrator, pay them a large salary to just do the database administration, and they would be able to do some of the other stuff the software programmers would be better at monitoring. Is the software or the software? Automate automations that Claude is creating? Are they really reliable? And a data analyst can tell you whether or not the ways that Claude is suggesting analyzing the data. Is that really what we want to do, or is there something better? What you should probably be doing is just find someone who who likes it, has demonstrated some some reasonable ability. And then for the areas that they don’t, you know, cover sufficiently. Bring in occasional consulting for it or training.

Josh Hadley 00:44:23  Love that great recommendation. Well, as we wrap things up, I love to leave the audience with three actionable takeaways from every episode.

Josh Hadley 00:44:30  Here are the three actionable takeaways that I noted. You let me know if I’m missing something here. Ellis. So, number one, I would say the most important thing you can do is go and hire for this role. Now, this can either be, hey, hiring a consultant or hiring a a firm like yours. They can go build this for you, but this is definitely an area where you as the CEO, I’m going to say it is an important area of the business, but like you can’t devote all of your time and energy to this. And as you’ve heard through the numerous use cases that we talked about today, this is a full time job. This is a full time job. Keeping up with the data warehouse, keeping up with just making sure that the output that you’re getting is accurate. So there’s a lot of things that go into this. So my first recommendation is going higher for this. So either internally or externally use an agency action item number two. I would say you need this becomes more and more paramount as the world continues to move faster and faster.

Josh Hadley 00:45:25  With the pace of AI, which is your business strategies should be 100% based off of what the data is telling you. And I think way too often there’s too many business owners that are just looking at their profit and loss statements and then making all of their decisions based off of what they saw looking in the rear view mirror. What I think what you’ve architected here is being able to have a visual database that is able to tell you and flash some warning signals, this is maybe trending in the wrong direction. Your use case with like the you know, hey, it’s spring season. Hey, my t shirts are starting to sell more and my shorts are starting to sell more. That’s great. But are they selling as well as they were the same time last year compared to the competition? Because you may be losing market share even though your sales are increasing, you’re actually losing market share. That’s where the data is going to point you in the right direction. Again, you as a business owner have unlimited possibilities of things you can do and invest your time and your team’s time in, but you have finite resources, so get really good at analyzing the data and letting the data lead you with the leading indicators to what you actually need to focus on when you resolve that constraint.

Josh Hadley 00:46:34  That’s what then opens up those ten x opportunities. And then my third and final action item here is you should only be leveraging AI after you’ve put in a proper foundation, after you’ve put together the right architecture. Here’s the biggest concern that I have. And this would be my biggest action item. Just because you heard some brand owners say, oh yeah, I just I hooked up Claude to this and it’s producing all of these reports for me, and it’s like, I don’t trust it. I genuinely don’t trust it. Because if you heard the conversation we just had with Liz, how many use cases are there where the LM is like confidently telling you something, but in reality it’s not even reading the data properly. And so I would caution everybody to say, yes, AI is powerful. It is the future. But if you do not set up the proper foundation, your AI will lead you off a cliff faster than you just managing your business. Even without AI, I think it’s better to use it without AI if you’re not going to set up the proper foundation.

Josh Hadley 00:47:34  So those are my three action items. Ellis, is there anything you feel like I missed?

Ellis Whitehead 00:47:38  Start putting your data into a Postgres database, like through the Super Bass service and and start getting your guidance and guardrails set up so that AI is working with your real live data. There’s there’s so many decisions that need to be made quickly, and you’re not going to be able to use AI for those kinds of decisions unless it’s in a live database.

Josh Hadley 00:47:59  Fantastic recommendation, LZ. My final three questions are for you here. Number one, what’s been the most influential book that you’ve read and why?

Ellis Whitehead 00:48:07  Well, I.

Josh Hadley 00:48:07  Actually.

Ellis Whitehead 00:48:08  Have these two these these are ones that I’m actually rereading right now. I wouldn’t necessarily call them the most influential, hard to decide, but the fact that I’m rereading them says something. So this one, this is Warren Buffett accounting, and it’s by Stig Brodersen and Preston Pysh. Preston Pysh. I’m reading this with my daughter and it makes it makes accounting and reading business statements really intuitive.

Ellis Whitehead 00:48:36  I love this book. I read it sometime around. I mean, probably a few years before I started my entrepreneurial journey as I was starting into investing in stocks. It’s a it’s a beautiful book for a motivated layman. And then this one is 1984 and almost finished this one up with my son. It’s a brilliant and timeless work of dystopian fiction that explores how authoritarian regimes use fear, misinformation, and the destruction of truth to control entire populations. It’ll probably never go out of style, so to speak.

Josh Hadley 00:49:11  Love it. Great recommendations. Ellis. Question number two here. What is your favorite AI tool and how have you been using it? And especially since you’re an AI data nerd. I’m expecting that you’ve already seen so many things. So I’m very curious to hear, like, what’s your favorite tool that you would think everybody else should be using?

Ellis Whitehead 00:49:30  Claude. Code from the terminal inside VS code using TypeScript and Deno. Not not a Python with as much data as possible in a Postgres database like we’ve been talking about, and all project, all projects versioned through git.

Ellis Whitehead 00:49:45  And this set up lets me use AI smoothly across business operations, bookkeeping, communication, system administration, software development, and everyday life. Just a couple of days ago, I did some bookkeeping for my wife’s business. I was so enthusiastic because the month before I had set, you know, I’d set some processes up to be able to read the PDFs and create these complex bookings. And so for the first time in my life, I just ran a script and it was all done. And that was, you know, because of this setup with Cloud Code, Visual Studio Code, git and so on.

Josh Hadley 00:50:17  What is git? Explain that a little bit more.

Ellis Whitehead 00:50:19  So git is it. It’s a versioning software. It is so powerful. It is the foundation I would go so far to say is the foundation of software development on planet Earth. It is perhaps the most meaningful piece of software that we as humanity have, because it makes it possible to share code changes, text changes, document changes in a structured way with history in a distributed manner.

Ellis Whitehead 00:50:47  This also means that AI can make changes to your repository, and you can check them out before you before you approve them. And it’s mostly with plain text files, so you wouldn’t be doing word documents with this. You know, markdown can do most stuff that word documents can do. It makes a huge difference. You can make changes without worrying about overwriting stuff, because if you overwrite something, you can go back in the history and restore it.

Josh Hadley 00:51:12  Love it. Fantastic. All right. My third and final question for you here is who is somebody you admire or respect the most in the e-commerce space that other people should be following and why?

Ellis Whitehead 00:51:22  First person that comes to mind and so many people come to mind. I would say the first person that comes to mind with regard to specifically following George Marissa on LinkedIn. So I really appreciate things that he highlights as being relevant in advertising and e-commerce. He’s he’s smart, principled guy.

Josh Hadley 00:51:39  Yeah, George is great. We’ve had him on the podcast George from Conrad’s.

Josh Hadley 00:51:43  So go check him out. Well, Ellis, this has been an absolute pleasure. Thanks for spending your time with us today. If people want to learn more about you, where can they find more info?

Ellis Whitehead 00:51:51  You can go to databricks.com and there are a couple banners up there. You can click on the one for the data chain review. If any of this resonate with you, and you want to figure out which links are broken in your business, you can have a call with me. And the other link is to our data ownership platform. And you can go log in there or sign up for the next beta entry.

Josh Hadley 00:52:14  Fantastic. Well, Ellis, it’s been an absolute pleasure. Thanks again for your time today.

Ellis Whitehead 00:52:18  Same here. Thank you so much, Josh.

MC 00:52:20  Thank you for listening. Visit ecommbreakthrough.com for more information. If you’ve enjoyed today’s episode, the best way you can show your appreciation is by clicking the subscribe button and quickly leaving a review. See you again next time!