What Most eCommerce Brands Get Wrong About Profitable Growth

Liam:

Rob, incredibly excited for today’s episode. I think your journey through ecommerce, starting right from the start and working your way up through the trenches to where you are today, is incredible. I’m excited to dive into that. People commonly hear you speaking on panels and at conferences, and I think the insights you have are consistently great. I hope we can bring some of that today. For those who don’t know you, why don’t you share who you are and what you do?

Rob:

Hi, everyone. First of all, thanks for having me. I’m Rob Varon. I’m the VP of Digital at Jordan Craig, which is a streetwear brand. And I always say, if you expected me to work at streetwear, there’s something wrong with you. Always gets a laugh. But yeah, as the VP of Digital, ultimately, I’m responsible for the P&L.

After helping hundreds of brands, even helping them spending one, two, three hours a week on them, I ultimately decided it was time to go brand side, where I went to a health and wellness company. Actually, not too far from where we’re filming this. And then I was recruited by Jordan Craig to take over the digital side of the business. So that’s a very quick summary of my career.

Liam:

And how did you get into the world of ecommerce in general? Like why did you eventually decide to go brand side?

Rob:

Yeah, ultimately what sparked the interest on brand side was just the ability to really commit my whole time to everything. You gave Meta a million dollars, you’d get 5 million back. You sent an email, your click rate was 10%.

Let me make sure that email is touching paid social, which is touching organic and touching brand. And when I was agency side, it was just too siloed to really make that big of a difference. And while you can add, I’ve had clients add 50 million plus in incremental email revenue. Those are obviously massive enterprise brands.

Liam:

And you love being able to be kind of across everything. What does your day-to-day look like now?

Rob:

Sure. So day to day is obviously managing a team. Well, I wish I could do it all. There’s just too much. So I have an amazing team reporting to me from every facet of digital. And I spend a lot of time now really building AI agents and workflows and making sure that not only I can be my most efficient self, but everyone on the team can be their most efficient self.

Liam:

Something you spend a lot of time thinking about, and I’m excited to come to AI in just a second, but starting kind of at the top, something you spend a lot of time thinking about, talking about, and are very adamant about is profitability.

Rob:

Ultimately, I think you have to focus on profit. It’s so easy to scale revenue. Any business can grow 5 million if you spend 10 million in ads. You could sell literally anything. But the reality is for a company like Jordan Craig, which is fully bootstrapped, profit is what keeps the business running. Positive cash flow. And it’s really a commitment to profit. You can’t just say you’re profit driven. You have to commit to it. It’s holding yourself accountable. It’s holding your vendors accountable. It’s everything. We have one North Star. And yes, everything goes into that: repeat purchase rate, acquisition, COGS, CAC, everything. But as long as you have one North Star, honestly, you can make the same amount on Meta, Google, Snapchat, Pinterest, et cetera. Generally, you could pull back spend. A lot of people are wasting spend.

If I give anyone a dollar, whatever channel it may be, whether that is a paid media channel, whether that’s an agency, whether that’s a data collector, they’ll say they made you $2, $3, $4, $8 back. But the reality is maybe only $1.50 of that was incremental.

Liam:

Incrementality gets talked about a lot, but what do you think it actually means in practice?

Rob:

There’s the brand value, the value of repeat purchase. There’s a lot of people who are going to come and convert. They’re going to see your site, they’re going to convert, be happy. There’s a halo effect. Ultimately, you could project how much you’re going to do without that extra spend. Right. So once you know that number, incrementality is every dollar, every penny even above that.

So let’s say I know I’m going to do 10 million on just the brand value alone through retention and other channels. That means, OK, 10 million and one penny, that’s incremental. And that’s where it’s coming from ads and data and vendors, et cetera. The first thing I would look at is looking at their spend and looking at what they’re getting for their return.

And I would bring in an incrementality provider. If you’re a $500,000, even a million dollar business, it is very tough. At that point, you should just be looking at MER, so how much you spend versus how much you make. But when you get up to $5, $10, $20, $50, $100 million, that’s when you really need to be focused on incrementality.

And I would say, take a step back, bring in one of the providers. We use WorkMagic. We love them, very happy with them, and run the incrementality test. Pinterest is a perfect example. Pinterest actually under-reported. They said they were only bringing in about a one ROAS, but when we did the incrementality testing, we found it was actually a 1.75 on new customer ROAS. So it left a lot more to be scaled and we grew and continued to spend.

It’s dependent on each brand. So a cosmetics company, which is obviously very high margin, they probably only need a 1.2. You talk about an auto parts company, they probably need a four or five. A lot of apparel brands, by the time you include COGS, shipping, and everything else, most generally need between a 1.5 to a 1.75 to break even. And how we look at it is we want, and not only want, we need every first purchase to be minimum break even, ideally at least making one penny of profit, because then when it comes to email, SMS, web push, app push and other retention channels, we know we’re really going to increase that LTV.

Liam:

How do you think about the balance between acquiring new customers versus relying on retention? What are your thoughts on it?

Rob:

Absolutely. I think for most brands, you need both. Ultimately, you need acquisition to be bringing in new customers. You can’t make your second purchase until you’ve made your first purchase.

Liam:

So it’s like fill the top of the funnel, but we need to extract so much more out of the middle.

Rob:

Exactly. Anyone who says they’re amazing at one and poor at the other is leaving a lot of money on the table. Ultimately, I think a lot of people over-fantasize the value of email because it is the most profitable channel by far. And while I say most people over-fantasize it, I’m not downplaying that it is extremely important and it is very profitable. That’s what gets you from a $150 LTV to a $300 LTV. And because you don’t have that second piece of CAC, it becomes significantly more profitable. However, email doesn’t put new people onto your list.

You need these paid media channels to really get people to come to your site. Unless you’re literally buying names, which no one should be advising, especially in ecommerce, you need these channels to get people to your site. Then you need the good pop-up that’s going to get them to join your list.

You need the welcome series. You need the browse abandoned. You need all the flows to give them that personalized experience. And I don’t expect someone to convert right the first time they come to the site. So that’s another layer where email and SMS, web push, app push, they all make the difference. They get you from first base all the way to home plate.

Liam:

Do you think there’s been like a push recently to say, you know, email and SMS marketing, retention marketing, it’s over-attributed?

Rob:

Looking at just email is wrong. And looking at just acquisition is wrong as well. Granted, you can grow focusing just on acquisition. It is much tougher to grow just focusing on retention. But email is highly important. Like I said, it increases your LTV. It gets people across the finish line. So email is very important. Again, you want to look at the incrementality. You want to be doing A/B testing. You want to be having holdouts to see how many people would have converted without your emails and SMS. Yes, all the ESPs are significantly, significantly over-attributing their revenue, but that doesn’t mean there aren’t platforms like WorkMagic that can measure the actual incremental value of these programs: email, SMS, web push, app push. And what you’re going to find is it’s very incremental.

Liam:

How is that working? Is it tracking the IP address of these shoppers?

Rob:

It’s link tracking. It’s a lot of geo tracking and it’s holdout analysis. So we say, okay, we’re going to take out this percentage of our email and see what the value coming back is.

Liam:

And you’ll sometimes see a drop?

Rob:

Yeah, absolutely. If you look at our revenue chart, we always send an SMS at 12 o’clock, except on Sundays, we send it at one o’clock.

Liam:

Every day.

Rob:

Every day. And we see revenue, boom, spikes up right at 12, right at 12:05, five minutes after. And then email, we generally send those at 5 p.m. every day. Boom, email goes up and then it starts to narrow down.

Liam:

Wow.

Rob:

But that boost is there. I always say you have to remind people you exist. And that’s what email and SMS does. You think about brands like Geico. Every American knows who Geico is at this point. Take Disney also. Every American knows who Disney is.

But why are they advertising Disney Plus and Geico? It’s because if you’re not advertising those, your competitors will. And people have short attention spans. Email’s the exact same thing. No matter how loyal they are to you, email makes them more loyal. SMS: oh, I was in the market for something. I wasn’t thinking about Jordan Craig, but now I am.

Liam:

How do you think that AI is going to change the whole landscape of email? Very wide question, but what’s your initial thoughts?

Rob:

Ultimately, I think AI gives the ability to mass produce content and personalize at scale. There’s so many metrics you’re getting, whether you’re getting zero-party data, first-party data, you can take that information and really start to scale out. So we send at least 40 campaigns a month on just email alone.

Liam:

So millions of emails.

Rob:

Millions of emails, yes. And we have some level of personalization. And a lot of times, okay, if we know you’re a big man, which is size 4XL to 6XL, we’ll give you a slightly different creative, but it’s still 90% the same. Soon it’s going to be 10% the same. And AI is going to allow us to have a kid’s email that’s unique, a big men’s email that’s unique, a men’s email that’s unique. So instead of 30, maybe 40 campaigns a month, it’s 120, 150. So really personalizing that experience is something that AI is going to be able to do. Right now, yeah, you could hire 20 email designers, but soon it’s going to be one.

Liam:

What does personalization mean today to you?

Rob:

First name personalization isn’t enough. I first want to say, don’t still do only that. That is one step of a hundred-step process. Look at what categories, what colors. So for example, we have a camo collection. If someone’s always bought camo and nothing else, great, let’s show them camo colorways for all of our products when they check an email. If someone has only bought fleece from us, okay, we could show them our denim. And yes, some of them are going to convert. But if there’s only a 10% crossover between our fleece buyers and our denim buyers, let’s make sure we’re continually showing them fleece in the colors they want, in the sizes they want. And we have the data we can put in: who was the model, what was the style, what was the look. There’s so many small elements within email that can be optimized. Something as little as the CTA can make a huge difference.

Yes, we could do that already if they are an existing buyer, maybe based off if they’ve added something to their cart.

Liam:

You couldn’t send a unique flow to every shopper though.

Rob:

Exactly.

Liam:

They’re all pretty much the same, right?

Rob:

So yes and no. So we have 86 flows. So we very much value flows.

Liam:

So much effort to create all that.

Rob:

It is a lot of effort. I will say it’s easier to be the person telling someone to do it than the person doing it. So we have an amazing, he’s now the director of marketing, but he started as our retention marketing specialist. He’s been promoted, but he’s amazing. His name’s Javier. He just designs, builds, codes, does everything. And it’s a lot easier to do that when you have someone great working for you. But ultimately, every flow has multiple facets. So if we know you’re a big men’s buyer, all your flows are going to be about big men’s. If you’re a men’s buyer, if you’re a kid’s buyer, all those elements are taken in. If you’re a multi-buyer versus a one-time buyer versus a non-buyer, we’re also looking at that. So you may get a different discount. While most of our flows don’t have discounts, we do put discounts in some of our flows. And the number of times you’ve purchased from us is going to have an impact on that.

Liam:

That’s very interesting. And then when you have like a site-wide sale or something, you’d have to go into every single one of these flows individually and update them. And that’s where I think the true value of AI is. A lot of these things are possible today. It’s just a balance of how much time. It’s not possible for a marketer to send a unique email to every shopper. There’s no way you’re going to sit down and write 30,000 flow emails a day to all of your shoppers and uniquely mention in the copy of the email, in the subject line, the item that they viewed, make it personal. But now that’s possible, right? And I think that’s the most interesting part.

Rob:

Absolutely. And yeah, 30,000 emails would be a horrible day if we only sent 30,000 flow emails in a day. But yeah, exactly right. It’s going to be at a much more personalized level. What we do now is when we have site-wide sales, ultimately we pause those flows, we clone them, and then we put in the new design and then we go back and unfreeze or turn them back on.

Liam:

In the near term, what are your thoughts on campaign generation?

Rob:

It’s huge. Look at November. So from November 1st to December 2nd, which was Cyber Monday, we sent 62 emails and 55 SMS. So that’s a lot of messages. On Black Friday and Cyber Monday alone, we were sending eight emails on each day. So soon it’s not eight emails on Black Friday. It’s 15 million unique emails on Black Friday.

Liam:

Is that possible today?

Rob:

I think we’re very close. I think today, no. By the end of 2026, maybe. By the end of 2027, absolutely.

Liam:

It’s something I think about a lot, right? What do you think that most brands get wrong when it comes to personalization?

Rob:

Let’s even start at the very basic. They’re not optimizing for sending cadence. I spoke about 40 campaigns a month. That’s not every single person on our list getting 40 campaigns. Besides the deliverability problems that would come with that, there’s also the fact some people don’t want to hear from you 40 times a month. A lot of people do want to hear from us. We have a few hundred thousand on our email list who want to hear from us every day and multiple times on some sales days. But if we mailed everyone with the same cadence, that wouldn’t work. I don’t even consider that personalization. I consider that a level below, but I think not enough brands are even getting to that level, where a lot of brands will say, oh, I send eight emails a month or 12 emails a month.

Liam:

Blast it out.

Rob:

Yeah, they blast it out. Everyone gets it. Or maybe they even say I do segmentation and I send 12 emails a month. But the reality is there’s probably 20% of their list who wants to hear from them 20 times a month. And had they spent the time to do those eight extra emails, they would have got X dollars more in incremental value and increased the LTV of those customers.

So that’s the very first thing I think companies are doing wrong: really not optimizing their sending cadence. Then I think when it comes to product recommendations, the more you know about a customer, that should be playing in. And a lot of platforms will have dynamic recommendations, and you need to look into that and see how is it doing. You need to run an A/B test. Could I do better? Could I give better recommendations? Is there a better recommendation engine than my current ESP? And with that, we ultimately see that, yes, we can find better tools or we could actually outperform ourselves. And then I think AI is going to take that to the next level. And then third is, especially for buyers, if you know the style they want, show them the style they want. If they’re purchasing a 6XL, they probably don’t want to see a size small. It’s a very distinct experience.

Liam:

Yes.

Rob:

Inversely, sometimes you do just need to send out campaigns and you’re doing a new release. And okay, you are sending a model who’s maybe a size medium. Is it perfect? No, but it’s better than not sending at all.

Liam:

I reckon campaigns are going to be one click pretty soon.

Rob:

Absolutely.

Liam:

That’s something we’re working on a lot: agentic campaigns. And you see the release with Claude today, Claude Design. It just paints the story of like, we should be using this in the world of marketing.

Rob:

If you’re not using AI at this point in the world of marketing, it’s not only the present, it’s already the past. So every price you now see for our product is run against AI, where we’re constantly running models to see what is the optimal price.

What’s the difference between a $40 shirt and a $42 shirt? $2 in incremental profit. And we could sell the exact same number of shirts, because we buy inventory so we know how much we have. It’s a set number. It’s not infinite. We’re not a drop shipper. If we have, let’s say, a thousand shirts, the difference between $40 and $42 is $2,000 in incremental lift. And you go, Rob, that’s only $2,000. Apply that across all of our products. And inversely, there may be a product that we had for $40 and it wasn’t selling. In apparel, dead inventory is the worst thing you could have for a P&L, because you paid to get the product. And if you can’t move it, you’re screwed. It’s just lost money. It’s a pure net negative to your P&L. So we use the AI to say, at what price point do we need to be to move this product? And before, you’d have to guess. Oh, maybe I have to go down to $20, $25, $30.

Liam:

To see when people will start purchasing.

Rob:

Exactly. So now we’re using AI to optimize that number: what is the maximum number we can do to clear out that product?

Liam:

And it’s accurate?

Rob:

Very much so. And not only that, we are doing it where we say, we don’t want to sell out tomorrow, because obviously if that’s how the model is trained, it would say, put it for a dollar and it’ll sell out in the next hour.

Liam:

Yeah.

Rob:

I could promise you we could sell 10,000 shirts if we put it at a dollar in under an hour. So we want to make sure we have inventory so people can come to the site and see things for the next three months. Give us a price where we are projected to sell out between 75 and 90 days. And that’s how much we are looking at it.

Liam:

And where is it getting that information from? Is it looking at competitors? Is it looking at the price point? Is it looking at the margin?

Rob:

Great example. So absolutely, we tell it the margin. It’s looking at all historical orders. We’re literally putting in survey data. And if a vendor is giving us data, we’re putting all this data in, putting it in the knowledge base, putting it in specific prompts and making sure that it’s analyzing across that field.

Liam:

And what tools are you using to build these agents?

Rob:

Sure, we use two. So for more basic tasks, we use a company called Relay.app. And that is something like check this email or do this basic task for me. For the task I just mentioned, we’re using a company called Hazel, which is hazel.ai, which is built on the Anthropic models. And what I like about them is it’s very trained for business. So while you could just use Claude, Claude is trained on everything. This is saying only focus on ecommerce business, which is what I really like about it.

Liam:

How are you pushing AI through the business?

Rob:

I got lucky here. A lot of people in the company want to use AI already. But the real value is them seeing how much time they saved. I’ll go out and build an agent for someone and they’ll be like, oh, this task used to take me 10 hours, now it takes me one hour.

Now with Relay.app, the data is getting pulled directly from Shopify, putting it into a Google Sheet, and then he gets to spend time actually analyzing the data and not just collecting the data. So what’s that for him? That’s a lot of boring tasks that he no longer has to do. And before, where he could do maybe four big analyses a month, now he’s doing 10, 11, 12 big analyses. And all those analyses are helping the brand grow.

Liam:

Is there much on the AI agent front when it comes to retention marketing, or is it still pretty early there?

Rob:

Sure. So even when it comes to campaign creation, it integrates with our platform, so we have all the data. Before, we were manually uploading all historical data and saying, this is the type of campaign we want to run. Rate us. Tell us what we should be doing differently and tell us where we’re leaving money on the table.

Liam:

And it will accurately build the next campaign, or it will just give you the strategy?

Rob:

Right now it’s just the strategy. I think we’re less than six months away from where it will be fully designing the campaign.

Liam:

That’s the most exciting part. As a leader, are you most excited about, or do you focus most on, revenue increase when bringing AI into the business?

Rob:

So ultimately AI for us is profit generation. Even our models are trained that way. If you can make us a million dollars in revenue and 500,000 in profit, or 1.1 million in revenue and 700,000 in profit, it knows to go for the second option, because it’s going to say, this is a profit-driven business. We are prioritizing profit. So that is number one.

Jordan Craig has been around for 37 years. The human element matters a lot to us. We are not interested in reducing headcount from AI. Now, could it be where before we would be hiring two additional people? Yes, that is a potential reduction. But we’re not going to fire people for the sake of the headcount savings.

Liam:

I agree. All of a sudden you’ve got an email marketer who can, you know, supercharge and drive so much more revenue out of campaigns because they’re not focused on needing to do the bare basics.

Looking ahead, how do you think that agentic AI will change ecommerce outside of just the marketing world? A lot of people tell me that the most important thing has actually come back to the basics: the product descriptions and your website being so important, because this is where search engines and all of these tools are collating their information. So what do you think is the most important thing to get right today?

Rob:

Yes, obviously AEO, GEO is super important. Even with Relay.app, we have it once a week crawl our whole website and give specific recommendations on how other AIs could see us better. I think that’s very relevant. That is making sure each product page is optimized, your about me page or about us page. That’s super relevant in today’s day and age. But AI is so much more than just marketing. It’s for product developers. Here, we want to create a shirt that looks like this and we put in a design. Estimate the revenue.

It could say, I project this to be a dud. Or I project this to be a star. Let’s say it projected to be a dud. That just saved us a huge headache. If we bought 50,000 pieces of something that we can only sell 500 of, that’s a killer to the P&L. And again, that’s not marketing related. That’s operations. And I think there’s so many pieces there. Yeah, of course, copywriting for all the products and descriptions. I mentioned pricing. AI can touch every single element, and it allows you to be a super spy. If I want to know what my competitors are doing, and I’m not going to name them on this podcast, but if I want to know everything they’re doing at all times, anytime they do something…

Liam:

It’ll tell you.

Rob:

It’ll tell me. Yep. So we have Relay.app. Anytime any of our competitors do anything major, I get an email saying company X did this, be on the lookout.

Liam:

And is that like pricing changes, product drops? Could it be more advanced than that?

Rob:

It could be anything. It scans the Meta ad library and says they just launched these new ads, here are their discounts. These products went on sale. This is the site-wide sale they’re doing. We even have it seeing if they have any of the same models. That one isn’t perfect yet. We’re still optimizing that.

Liam:

Same models as you.

Rob:

Exactly. Yep.

Liam:

Wow. That’s very interesting. If you were working with a new brand today, what’s your recommendation on where to simply get started?

Rob:

Yeah, I would say ultimately it’s easier now than ever to get started, and to really just focus on as many agentic items as you can. And even before that, speak to Claude, speak to OpenAI or ChatGPT or Gemini, run your ideas past it first. For a lot of people, you may have an idea. It could tell you 10 things that are wrong. So I think that’s always the biggest thing.

Liam:

Ending on this note, what do you think is the most overrated thing in the world of AI right now when it comes to ecommerce?

Rob:

Sure. I think the most overrated is still image generation. I think we’re about six months away, maybe 12 months. For a basic tee, so for the tee I’m wearing, AI could do a pretty good job with that. But a lot of Jordan Craig’s clothes are very unique, with patches and specific techs.

And if I want to do a photo shoot in the Bronx with these very specific things, you could still tell it’s AI. In six months, 12 months, you’re not going to know the difference. And I think that’s the biggest piece. And the reality is once we get there, it’s so much savings.

We spend so much on photo shoots all across the city. And now it’s, OK, let’s say we want a London shoot. We don’t want to fly the whole production team to London. Now it’s done in a second. Again, I think too many companies are overly reliant on it for content today, but in the next six months, my statement would be that not enough brands are relying on it for content.

Liam:

Everyone talks about what’s going to happen in the next three years. 12 months ago, we weren’t even talking about completely automating a campaign.

Rob:

And you have to be able to look into the future. Just one of the last things: I remember going to conferences in 2022, 2023, 2024. 2022: AI is not going to replace your job. No jobs are going to be replaced. And I remember sitting at a table and I said, are you lying to me or are you lying to yourself? 2023: AI is not going to replace your jobs. And I asked the exact same question. Are you lying to me or are you lying to yourself? 2024, exact same thing. Last year was the first year where the conversation became, AI is great. And the group I’m most concerned about is Gen Z and Gen Alpha, because right now for a lot of entry-level jobs, AI can do it faster, quicker, more efficiently and cheaper. So I think the most important thing you can do at this stage is get really good at AI: building agents, prompting and being AI native. Five years ago, you could see the future, or at least I could see the future, but that’s no longer the future. Now you’re in the future.

Liam:

Love it. Well, thanks, Rob, for coming on. I think this was an incredible episode. I think the core message is, if you’re not using AI today, you’re already behind.

Rob:

And use it for good. Use it to be more profitable.

Liam:

Yeah, ChatGPT searches isn’t AI anymore. I mean, that’s just generic, right? You actually need to use it to grow your business.

Rob:

Absolutely. Grow your business and focus on the metrics that matter most. And for us, that’s profit. For a lot of businesses, it’s still revenue, but I’m excited to see more and more brands get more and more advanced.

They cover:

Why profit should be the North Star for every growth decision

Why platform-reported ROAS can hide what is truly incremental

How email, SMS, and push work together across the customer journey

How Jordan Craig uses AI for pricing, campaign planning, and competitor monitoring

Why marketers will become managers of agents instead of being replaced by them

If you lead ecommerce, growth, retention, or digital, this conversation is a practical guide to building a more profitable and AI-native business.

00:00 Preview

00:53 Introduction

01:27 Rob Varon and the move from agency to brand side

02:16 Why ecommerce growth has become more complex

03:36 Building an efficient digital team with AI

04:13 Why profit is the only North Star

05:50 What incrementality actually means

07:39 Why ROAS can be misleading

08:34 Acquisition, retention, and the real customer journey

10:18 The metrics that matter for repeat purchase

11:52 Why email and SMS still drive revenue

14:11 Personalization beyond a first name

17:48 How Jordan Craig plans lifecycle campaigns

19:34 Coordinating the customer journey across channels

21:27 Why brands are already behind on AI

21:51 Where AI agents create real value

25:09 Getting teams to adopt AI

27:02 AI should generate profit, not replace people

27:58 The rise of the agent orchestrator

28:50 Where AI fits beyond marketing

31:30 What AI is overrated at today

33:03 What ecommerce operators should focus on next