D2C Revenue Rocket — Episode 7
Guest: Aaron Schwartz, Co-Founder & Co-CEO, Orita AI

[00:07]  ZAK
Welcome to the D2C Revenue Rocket Podcast. Today we're joined by Aaron Schwartz, co-founder and co-CEO of Orita AI, which builds AI customer segments across email, SMS, and direct mail for hundreds of brands like Spanx, Caraway, Faraday, and Harney & Sons, to name a few. Their segments enable brands to email customers when they're actually listening — it improves deliverability, it improves revenue, and it makes the email marketer's job a lot easier when they're choosing what segments to send to.
Aaron has spent years building and scaling companies across e-commerce. He founded and sold a brand, co-founded Passport Shipping, and was President at Loop Returns. Through the Orita podcast, which Aaron runs, he regularly sits down with top operators, founders, marketers, and investors to unpack what's actually working in e-commerce today — from AI and personalization to retention strategy and profitability. Aaron, welcome to the show.

[01:04]  AARON
Thanks, man. That was very stressful listening to you read what I've done. But thank you for saying nice things.

[01:11]  ZAK
You have a pretty awesome background — you've been in this space for a while. Before we jump into our questions, can we dig into that a little? Can you walk me through how you got started in e-commerce, and why e-commerce?

[01:25]  AARON
Sure. I went to business school in '08 and had been a consultant before that. When I went to B-school, I knew I wanted to be an entrepreneur, but I always thought of entrepreneurs as — get a Subway franchise, get another one, do it the traditional way. There was no high-tech entrepreneurship in how I thought about the world. Going to get my MBA at Berkeley, I obviously saw a lot of it, but my first business was a sustainability business and we did everything wrong — no engineers, we had no idea what we were doing.
Then my second year, I took a class from Eric Ries, who wrote Lean Startup, and Steve Blank, who's basically the godfather of customer discovery. And I was like, holy shit, we did everything wrong with the last startup. Long story short, my buddy Gary was like, hey, do you want to start a watch company for the summer and just see how it goes? I said yes. I'd never really worn watches, he wasn't an expert, and it was basically about bringing lean startup methodology to a physical product company. It started scaling weirdly quickly — we sold like $100,000 worth of watches to Facebook and Google. I told my parents I wasn't going back to consulting. And then seven and a half years later, I was still running a watch company that was no longer scaling very quickly. That was the start of my commerce journey.

[02:59]  ZAK
Wait — did you start that while you were still in B-school?

[03:02]  AARON
I have a photo from April, right before graduation, with a whiteboard — Gary and me with a list of things we needed to learn. One was how to source watches, how to ship. We literally had no idea what we were doing. Our first sale was July 9th, 2010 — to a buddy, Mike Roskam. I think he was the first purchaser. Cleveland connection.

[03:27]  ZAK
That's awesome. I love that. I went to B-school knowing exactly what I wanted to do — and graduated with absolutely no idea. So you grew the watch company for about seven and a half years. What was next?

[03:52]  AARON
Late 2016, early 2017, it was clear we needed to pivot — we tried turning it into an influencer marketing platform. I could give you a thousand things we did wrong, but there were some really cool experiences. Afterward, I didn't know what to do. A friend recommended, 'Nobody's going to know what to do with you, so just go work with a bunch of companies for free.' So I hung around with maybe half a dozen startup founders as a free resource — one was running a jet leasing company, one was the founder of Boba Guys.
I ended up spending time with this guy Alex, who had gone through the same accelerator I had. He had some ideas around a cross-border shipping problem. I had that problem myself — 20% of my traffic was international, but 1% of my revenue, because it was such a pain to ship internationally. We ended up co-founding Passport Shipping in late 2017 and it went off to the races. That business is still going strong — he's crushing it as CEO. That was my switch from the brand side to the commerce enablement side. From there I've been an investor and advisor in 60, 70 companies — too many at this point.
I joined Loop Returns for a stint — came in as a consultant to run go-to-market, then switched over to product, and was President there for a while. Then I co-founded Orita. I was very happy on the sidelines just consulting and advising, and then I met my co-founders DB and Zach in the summer of 2023. I wasn't advising them for a month before they asked, 'Do you just want to be a co-founder?' My answer was everything is great in my life — but sure, let's see what happens.

[05:54]  ZAK
So you guys officially got started in 2023 — when did Orita launch?

[05:59]  AARON
The two of them are brilliant ML engineers who had been working together for a long time. Simplifying the story, it was basically a consulting business in search of a product to productize. They did consulting on demand planning and had previously worked at a same-day delivery company doing routing. If there's a hairy data science problem in e-commerce or retail, they'd solved it for somebody.
The question they were searching for was: what business can we build around this? When they figured it out — the early idea of Orita was essentially, 'Hey, you're way over-emailing your list. We can use data to tell you who to message and who not to, and you'll make the same money with less.' That's when I was brought in as an advisor to help with go-to-market and fundraising. We've been chipping away at this one problem for three years: who wants to hear from you, and who doesn't? There are years of work behind that — which is why we're building something differentiated that Claude can't just replicate.

[07:07]  ZAK
I think that's just awesome — you guys really figured it out. We have a lot of overlapping networks and everyone I talk to says the same things about you guys: one, super nice guys, and two, absolutely brilliant.

[07:31]  AARON
They're talking about DB and Zach when they mention brilliance. I'm the A-minus. Very consistently liberal arts B-plus, A-minus — and I'm fine with it.

[07:40]  ZAK
You guys really did figure out a major issue, especially with email deliverability and all the changes over the last couple of years. The timing for your product is perfect. You guys are exploding.

[07:55]  AARON
It's been fun. We also started moving upmarket. We're a big data company, and the more data a brand has, the more impactful we are — which is the opposite of most tools. We finally started leaning into that. It is funny though, because AI is taking over everything, but there's still a lot of resistance — 'No, no, we should still do 90-day engaged segments' or 'We should segment based on my personal experience.' And we're like, guys, for some of our brands there are billions of data points you're not leveraging. I feel like we're at the last vestige of people realizing that math does win. Hopefully that flips in the coming months and we can get back to hyperscale.

[08:53]  ZAK
So break down how Orita works. The way you've explained it to me in the past, it just makes so much sense — you can't help but think, wow, I need this product.

[09:00]  AARON
There's a lot of pressure. I've been trying to whittle this down more and more. The way to think about us is: we're focused on solving one problem — in a given channel, whether email or SMS, how do I stack rank my list on a daily basis for who wants to hear from me and who doesn't?
If you think about marketing, there are really two problems: capturing attention and converting attention. Everybody is fixated on converting attention. It's why there's so much focus on personalization — what discount should Zach get, 10% or 12%? How do we improve our welcome flow or abandoned cart flow? That makes a ton of sense — if you can eke out 10 basis points of improvement, go get the dollars.
But with owned channels — email, SMS, even direct mail — you actually haven't captured the person's attention yet. On Instagram, the attention is captured — she's scrolling, go hit them with a great ad. On TV, the attention is captured — somebody's watching, hit them with an ad. But with email, the real question is: how do I get Zach's attention so I can convert him? That's our whole focus.
We get all of a brand's data and give them a list on a daily basis — email this person, don't email this person. When you break it down that way, you start having much tighter segments, capturing all the clicks and purchases. Over time, you build a healthier program and make more money because you're doing what the customer wants, not what you want.

[10:56]  AARON
One more thing — the framing we talk to most customers about is: when I ran a brand, I sent every email to every customer. When you're a marketer, you're thinking, I've got something good to say, we worked hard on this, I'm going to send it out. But take off your marketer hat and put on your consumer hat. You know you don't want to hear from every brand every day. And because someone clicked once 85 days ago, they shouldn't get every single email going forward. We want to help marketers actually figure out when their customers are listening.

[11:38]  ZAK
One of the things you said before that makes a lot of sense — we audit tons of Klaviyo accounts, and the segmentation is almost always where the biggest improvement opportunity is. The way you described it once was: if you're emailing your 30-day, 60-day, or 90-day engaged segment, someone who clicked once in 90 days and someone who clicked 101 times in 90 days are getting the same message — but they're drastically different customers.

[12:25]  AARON
Exactly. And the flip side is: if somebody clicked 91 days ago, they're out of that segment. Maybe they're not out for the week, but they're out for that day. But you don't have an average customer — some people might take 70 days from your first email to actually make a purchase because they need to explore your catalog. The way we're wrestling with it is exactly what you said: there has to be a better way. And the better way is there's an insane amount of data you can process. That's literally what our team has been working on for years — how do we use all the data to stack rank?

[13:03]  ZAK
Back when you launched your watch company in 2010, you could email your entire list with no problem whatsoever. Talk a little bit about how things have changed from 2010 to 2015 to 2020 and now.

[13:28]  AARON
I think there's one big fundamental change: today, 85–90% of all email in the world is spam. Even if that was true back then, the tools weren't available to protect against it. If you just think about Gmail, they have hundreds of machine learning engineers working on all our behalves as consumers to protect our inboxes.
Day one of any sender, the assumption is you're a spammer — because nine out of ten senders are. People think about IP warming and earning trust, and that's fine. But the difference is Gmail has gotten better and better. Now your promotions inbox is about relevancy, not just recency. They've changed the rules and made it very clear that opens don't matter — yet every marketer we talk to still pays attention to opens. It's just an old habit.
Actually ranking and getting into the inbox requires you to be great at the core metrics Google cares about: clicks, time on site, purchases. Same as SEO — if you get a click and someone bounces immediately, it signals to Google that the content was clickbait. The exact same thing is happening with inboxes. The way to fight machine learning is with machine learning. Go use the data you have.

[16:52]  ZAK
And you guys make it easy to use that data.

[16:53]  AARON
That's the idea. We're great at ranking your list from one to a million on a daily basis. If you'd talked to our customers two years ago, 50% of them would have said 'I don't really know what they do' — that was a real problem. We had to translate really complex stuff into dead simple language, which we were only mediocre at at first. We're pretty good at it now.
We do statistical testing, but nine out of ten retention marketers don't know how to use it — performance marketers love it, retention marketers don't. So we had to change how we do our reporting. Now the job is to expose more and more of our algorithm so you can actually go deeper. You understand what we do, you see the value — now it's about teaching you how to use it to solve different objectives.

[15:59]  ZAK
That's a good breakdown of how it works. AI and the advances over the last couple of years have been undeniably amazing — but there's also a lot of hype. What do you think people are getting wrong?

[16:05]  AARON
I don't think anybody is getting anything wrong with the hype — it is incredible. But there are two mistakes I'm trying to help our team avoid.
Number one: AI for AI's sake is not useful. AI is useful in service of a job to be done. What are we trying to deliver to our customer? What should this report say? What should the dashboard look like? If you use AI as an accelerant, awesome — but don't forget what the actual goal is, and make sure you're tracking whether the output from AI is actually moving the metrics or the experience forward. I think people get lost on that first and foremost.

[17:19]  ZAK
Yeah — humans are still controlling the overarching strategy for what we're leveraging AI for.

[17:29]  AARON
I think so. Even if AI helps with strategy, don't forget to ultimately track whether you made things better or worse. It's analogous to a debate we had internally about our signup flow 18 months ago. Our flow was really ugly, Zach — but everybody who went through it would click, and we'd run a free audit for every brand. We're a big data team: if you don't have the right data, we're not going to try to sell you. We'd give you a free audit, some insights, and then you'd sign up or not.
So we had this flow that worked — ugly, but it worked. Then we redesigned it. I was told it was better. I asked why. 'Well, look how good it looks.' I said: the job to be done is not to look good — it's to get 100% of people who hit this page to convert to an audit. And who's even tracking whether it's better? The answer was: we weren't. Whatever the tool is, it's about the end result. Why are you doing the work?

[20:37]  ZAK
So did you change back to the old one?

[20:40]  AARON
There are certain things I've decided not to talk about further. But I think we did okay — eventually things got better. It was an ongoing debate. The definition of 'better' is a metric, not a taste. We're an internal tool. If you're building a consumer brand, yes — taste matters 100%. But for us?
The other thing I've been noticing — this goes back to a business school class on design thinking — is that AI, even when you say 'give me a draft' or 'help me as a thought partner,' it still creates constraints. It's like when someone says don't think about a pink elephant. You instantly think about it. Once you get that first draft, even if it's awesome, even if it took AI 10 seconds instead of 10 hours — it narrows your field of vision. It takes a lot of work to say, okay, let me pretend I didn't see this. How do I explore the edges?
I think a lot of people are missing that. You get something that looks so good and so convincing. Even if you tell yourself it's just a draft, it's already narrowing your aperture. It's a skill to step back and say, how do I diverge again — not just converge on the path the AI put me on.

[22:06]  ZAK
I want to double tap on that because it's so accurate. I'll find myself using AI to accomplish some task, and it'll take me down a path I wasn't expecting — and there'll be good stuff there. But I'll find myself stuck in that channel, wanting to bring in other ideas, and sometimes it's not easy to do that. You need to know when to cut your losses and start over completely.

[22:40]  AARON
It's hard. But it's a skill.

[22:41]  ZAK
You have to think broadly. You might have all this great stuff from the path you went down, and you want to pull the best parts out but leave the rest behind. You need to know how to extract what worked and bring it into a new direction.

[23:08]  AARON
Yeah. In design thinking or brainstorming, you diverge — no bad ideas, more ideas — until you're so annoyed by the exercise. That's when the next breakthrough comes, because you've exhausted everything. Then you converge with a group, cluster your post-its into themes, write everything up — and then you diverge again. What are we missing? What new connections can we make? Then converge again.
I think a lot of people just converge and then make it really good — but it ends up being a local maximum. If you use AI to build your marketing strategy, it'll be good. It'll read every blog post ever written about e-commerce marketing strategy for brands between $10M and $50M serving female demographics. Great. But it'll still be missing everything that's unique to your brand, and all the proprietary knowledge that a great agency has.

[22:22]  ZAK
[AD BREAK] Want to see where your e-commerce brand is leaving money on the table? Book a free revenue audit with the team at ECD Digital Strategy and get a clear plan to increase revenue. Go to Klaviyo.com/ECD to schedule yours today.

[22:35]  ZAK
You see a lot of e-commerce brands — from their Klaviyo data, their email data, and from talking with brands on your podcast. What do you think separates good operators from great operators?

[24:45]  AARON
I think the best operators know their customer inside and out — whether because they are that customer, or because they're extremely diligent about interviewing those customers, or they have an awesome CS org doing great customer discovery. One of the best brands and best e-commerce exits in a long time is Chubbies. I'm friends with Kyle, the founder, who then co-founded Loop and is now building Good Day Software — a modern ERP that replaces NetSuite. They knew their customers inside and out, and you could see it in every brand touchpoint.
So first and foremost: the best operators are close to their customer. The second thing — and Kyle is still probably the best example — is facility with numbers. You can get into the weeds, you're thinking about margin from day one, you understand what happens if you open retail stores. When you get that blend, you get companies like Hulken or Alex and Lee — I think they're over $40–50M with seven or eight full-time people. Crazy efficiency. Assumed awesome product, deep understanding of your customers — not just what they like, but the experience they want — and actually understanding the numbers. That's the deadly combination.

[24:23]  ZAK
I was going to ask about what makes great operators on the e-commerce software side, but it sounds like it's relatively the same.

[24:35]  AARON
It's a good question. Customer empathy matters, obviously. But my opinion has changed recently — maybe I'm talking my own book here. Because it's becoming easier and easier to build products or self-serve, the best companies going forward are going to be the ones making craft products: really, really well-done things you can't just buy code for, built on data sources that aren't public. What in ML terms is called an expert system.
Building A-plus products is how you win — because if something can be whipped up in a couple of weeks, so can the brand itself. The best high-growth companies are going to be the ones solving the hardest problems and solving them in extreme detail, because they care about the craft of it.

[25:56]  ZAK
That's good advice for software companies too. With vibe coding now, it's wild what you can create — but it doesn't go super deep. What you're saying is: if you already have a software company, creating that moat means doubling down and going deeper and deeper on your specific product.

[26:09]  AARON
Exactly.

[26:12]  ZAK
You have a weekly podcast — or is it more than once a week? It seems like there's a new episode every day. I don't know how you do it.

[26:21]  AARON
I had an episode this morning with Lou from Grid and Pixel. I think I've got two more this week. We took a week off last week and I felt empty and alone. The Orita podcast — which is almost about to be renamed, though I won't spoil the reveal nobody actually cares about — exists because I like talking to people in this ecosystem. I ask to spend time with people I want to learn from. I make it a point to go back with two full pages of notes because I'm learning from the conversation. It's very selfish.
But at this stage of my career, if I'm learning something, you're probably teasing out something useful for a brand owner, a commerce tech operator, or an agency owner. I do it because it's fun, I love this ecosystem, and it's nice to interact with really sharp people who know their craft better than I do. It keeps me leveling up.

[27:26]  ZAK
I love that. You gave us some great advice when Clovey and I were getting ready to launch our podcast — you lent us a bunch of time and gave us real insights on how to run a successful show. I appreciate that.

[27:43]  AARON
Obviously get super cool guests, like you're doing on this one. That was the number one piece of advice.

[27:52]  ZAK
Exactly what I'm doing right here. How long have you been doing the podcast?

[28:03]  AARON
About a year and a half, maybe. The first time I ever did anything like this was at the beginning of COVID — late March or early April 2020 — where I basically invited 20 guests to talk about what the hell was going on in commerce. I had four people in four different 45-minute slots — a VC, a brand owner, an agency — and the people who were supposed to leave after 45 minutes just stayed. By the end I had at least a dozen people in a grid all answering questions.
I've always enjoyed moderating panels where I get to learn. We started with two or three people on every episode. I've been doing more one-on-ones lately, but I'm going to get back to two guests because the interplay — even between people who don't know each other — is where you really tease out the insights that you'd never get from a written answer.

[29:24]  ZAK
We did one panel discussion at ShopTalk with Klaviyo, Postpilot, Shopify, and Clear.io. I loved it — more people, more ideas, one idea leading to another. The give and take, and how the conversation grew naturally, was really good.

[29:55]  AARON
I agree. The risk is always people talking their book. You have to be very upfront: look, you can tell me about your technology, but I need you to abstract from it — assume they're working with a competitor. What should a brand be doing? As long as you get guests like that, it can be magical.

[30:16]  ZAK
Sam Moorhead from Klaviyo was on our podcast and was awesome. He's going to start co-hosting some podcasts with brands.

[30:24]  AARON
You suckered me in — I was like, all right, Sam, let's do it. He's coming on, which I'm excited about.

[30:30]  ZAK
He was great. Okay, so over the last year and a half of having guests on the podcast — how have the themes changed? Or have they not?

[30:46]  AARON
It depends. Tariffs were the only thing on everybody's mind for a month or two — and then six months later, they were still the number one thing for the best operators. It wasn't a blip. It became: I've got to build resiliency into my business.
AI has been interesting. I'm wrestling with how to ask better questions to get into actual AI workflows — what are the jobs to be done, how are people enabling their teams, how do you set a higher bar for your team than before? I think about AI more from a company-building perspective: what is the customer experience you want to deliver in a year and a half, and what team behaviors do you need to develop now to get there?
From a brand perspective, the channels discussion is always fascinating. Direct mail is way more on people's minds than even a year ago — people are genuinely getting it now. We're starting to talk about connected TV with more people. And I'm always curious how people think consumer behavior and consumer discovery are changing. For a while it was all AEO — but it's more nuanced than 'people are going to use ChatGPT.' Why does somebody buy your product? How do they feel most connected to your business? Those are pretty consistent themes.

[32:09]  ZAK
I'm seeing the same things with direct mail and connected TV. Two years ago, almost nobody was talking about connected TV — now it's coming up constantly. Same with Postpilot. Whenever we do e-commerce audits, we're always looking for levers that drive the most additional revenue the fastest — and Postpilot has always been one of them. What's cool now is that I'm starting to see Postpilot already connected on a lot of the accounts we're auditing. People are getting it.

[33:09]  AARON
We built a product with Postpilot that we've been running for a while, and we just rebuilt it from scratch — version two is pure ML. It figures out, from a retention standpoint, how to build the highest incremental lift — really understanding something bespoke for each brand, specifically where in the audience journey a postcard will incrementally move them forward. We're quietly rolling that out right now.
But yeah, it all goes back to what we're building: does Zach want to hear from me in a given channel on a given day? Not just, does Zach want to hear from me. Plenty of tools ask the second question. But when you start asking and answering the first one — channel by channel — you realize some people really love direct mail. Great, send it. Some people love text messages. Some love email. Most people don't.

[34:15]  ZAK
So the product you're rolling out right now is essentially channel affinity at a broader scale — email, SMS, Postpilot. Is that right?

[34:41]  AARON
We're adding connected TV. If you think about where we're going, it's channel by channel. I'd frame it less as channel affinity and more as this: if you have a budget — not a dollar budget, but if you're sending an email today and Zach is at the top of that list, send him the email. If he's also at the top of the SMS list, don't overthink it — email and text Zach.
The better question we're working on is what's the delta? Zach might be in the 70th percentile on email, and 50th percentile on SMS. But this might be the highest he's going to be on SMS for a while — so yes, you should text him today. On email, you could wait, and he'll be at an even higher engagement level later. That's the predictive nature of what we're doing.
And this is why it's taken us so long — it's not just about Zach's behavior. You have to look at the brand's behavior too. Is today a high-risk or low-risk day for this brand in this channel? When you go that deep, it becomes much more than 'email or don't email.'

[38:44]  ZAK
I see that as being extremely powerful — especially the more data a brand has, the more you have to work with.

[38:55]  AARON
Exactly. One of the reasons we love the Klaviyo ecosystem is that the data flows very cleanly. You don't just see clicks and opens and time on site. You can see reviews, returns, loyalty data, server-side tracking. Did this customer come from a shady de-anonymization site, or did they genuinely sign up to a good pop-up?
When you look at all the data, you realize a return 30 days ago is worth way less than a click 30 days ago — but it's not worth nothing. In the same way, in a 90-day engagement window, a click at day 89 isn't everything, and a click at day 91 isn't nothing. When you have all this data across all time for a brand, the machine starts finding connections that humans would never pick up on.

[39:52]  ZAK
Building these kinds of segments manually would be impossible.

[39:58]  AARON
You can't. You can 80-20 it — maybe even 90-10 it — but it would take a ton of time. And guess what? A week from now, we're a week closer to BFCM, a week further from a product launch in early May. The right time to message someone shifts. Maybe our model says 82 days. Maybe now it's 87. But you don't update it. That's a slippery slope. And by the way, 82 days captures 90% of your audience — but there's still a 10% you're missing. In spirit, you could do it manually — but it would consume all of your time and you'd still fall short.

[40:43]  ZAK
You brought up AI workflows earlier as one of the podcast themes. What's the AI workflow you've created recently that you love the most?

[41:07]  AARON
Honestly, I don't have a great answer to this — which is embarrassing, and I should be fired. I'll tell you what I'm focused on instead.
There are two problems I'm most interested in solving with AI right now. One: we don't have a COO. Ostensibly, I'm co-CEO overseeing go-to-market — but I'm nobody's dream COO. I'm trying to build not an 'AI COO,' but get all the outputs I need: faster visibility when we shift from expectations. We expected to grow 15% this month; we grew 12%, or 18%. What's the earliest signal I can get? Same for pipeline — what are the inputs that drive toward outputs?
The second thing, which I think is more interesting, is we're rebuilding our entire FAQ. We're using AI to tease out what questions customers are actually asking. We have a huge knowledge base from all our calls and internal documentation. What we're trying to do is rebuild everything so we can build a self-serve model — not because we don't want to give customer support, but because so much of the time people just want an answer. Right now we unintentionally gatekeep because people don't know where to find things, or because there's a nuance specific to Zach's brand versus Aaron's brand. We want to expose as much information as possible publicly so it's dead simple to understand how to get the most out of the product.

[40:12]  ZAK
Will you have a customer-facing chat agent where people can ask questions against that knowledge base?

[40:20]  AARON
The base level of that is yes. My reticence about launching it yet is that I want the answers to be pristine. And then there will be something custom for each brand — for Zach versus me. Imagine someone asks, 'Can I get rid of everyone at 50th percentile and under?' The answer is: yes, as in don't email them today — but never delete them, because people cycle in and out. There's nuance.
Our surface area is narrow — we're your segmentation partner, not a full platform like Klaviyo. We should be able to answer the 200 core questions. The next level is when Zach's data through the Klaviyo MCP is X, Y, and Z, and Aaron's is A, B, and C — our answers need to be completely different. That part needs to be deeply protected. And you and I, on any given day, should be getting completely different recommendations from Orita. We already do that — we just need to build the chat experience to surface it in a way that's self-evident.

[41:57]  ZAK
Awesome.

[42:00]  AARON
Our chatbot has a name, by the way: Atiro — which is Orita backwards, because we're lazy and creative. Megan, who used to run partnerships for us, was the first one to suggest it. We said great, done.

[42:13]  ZAK
I love that. Anything come up in your podcast interviews on this topic that's been particularly cool?

[42:20]  AARON
There's an overwhelming push right now to use Claude's co-work feature. But the more interesting conversations are with people building shared projects — where it's not just single-player, but multiplayer. It's one thing to be more efficient at your part of a project. It's another for everyone to be inputting into the same place.
It's like going from Word docs to Google Docs — you now have real collaboration. I think AI for collaboration might be the most interesting and highest-leverage trend right now, at least in the near term.

[43:17]  ZAK
We're seeing some awesome stuff internally at our agency with exactly that. I hadn't thought about it as 'AI for collaboration,' but being able to build training on specific projects and share it with everyone working on that client — yeah, that's it.

[43:30]  AARON
It sounds like what Deloitte tried to do in 2004 or 2005 with a project called KX — knowledge exchange. You build a presentation for a client, you put it in the system so the next person can learn from it. I think one of the things that makes you guys genuinely unique is that you really know your customer — you build a great experience and then you do the actual work of capturing and sharing those lessons brand to brand.

[43:47]  ZAK
That's one of the things that's helped me build this agency — I had an e-comm brand before, so I knew what a good agency looked like from the client side. And I actually just launched a new e-commerce brand about a month ago, which is exciting.

[44:40]  AARON
I'm not going to ask for a discount, but I do want the website — because friends don't ask friends for discounts.

[44:50]  ZAK
Lazy Leaf. It's a horticulture marketplace — indoor plants, outdoor plants, gardening materials. As you can probably tell from my background, I like plants.

[44:59]  AARON
I do too — I'm just very bad at gardening. But this sounds like a great place for me as a lazy leafer.

[45:07]  ZAK
Exactly — you can be lazy and we'll give you all the great products you need. How are you enabling your team to adopt and lean into AI more?

[45:15]  AARON
There's a spectrum. Our engineering and ML teams are obviously all in. Some go-to-market folks have made it a core part of everything they do. There have been two big shifts in truly the last month.
First: we were playing around with Gemini and other tools, but we just got Claude Code for everybody. We did training right away — code, co-work, let's go. Second: we started an 'investment day.' Once a month, no meetings — you tell people what you're going to get better at, you can work with others or alone, and the goal is to figure out how to improve. Get better with AI, build something, explore something new. Then we all share out afterward. Last time, Dan — who just joined our team — ran a session on how to use Claude. That was from the engineering side. Steven ran a session on the day in the life of a customer — because he came from a great agency, and our engineers can be on customer calls, but that's very different from really getting the ground truth of that daily experience.
I'd say we're middle of the pack. I live in the Bay Area and see what everyone else is doing, and I'm like — we've got to level up faster. But we're doing the work.

[48:14]  ZAK
Being in the Bay Area means the companies around you are above what most people are doing from a technology standpoint. What advice would you give founders building in e-commerce today?

[48:31]  AARON
My general advice to any entrepreneur: focus wins. Customer intimacy matters, having a team that cares matters — everything is about customer orientation. But then it's focus. Entrepreneurs and companies with split focus end up screwing themselves. Even if your executive team is directionally aligned and everyone is doing good work, you start growing in slightly different directions. That's fine if you're exploring a broader surface — but if you're wrong, you need to figure that out quickly. Being wrong in the same direction means you course-correct fast. Being wrong in different directions is a slow disaster.
Focus as much as possible.

[48:29]  ZAK
Telling entrepreneurs to focus is a tough one — because we're wired to do the opposite.

[49:42]  AARON
I was a very, very good startup advisor. It is unclear that I'm a very good startup runner. So yes.

[49:49]  ZAK
I think you've got a pretty good track record — and Orita is looking awesome and exciting. What's next for Orita? You've teased a few things, but anything else to share?

[49:58]  AARON
Number one is more channels. We are better than anybody at stack-ranking an email list. Klaviyo named us a premier partner. There isn't another segmentation partner that does what we do that's even a preferred partner. We're great at this one thing — and it's taken us a long time, but we're now launching SMS. Not because we were lazy, but because we don't want to launch something unless we actually solve the problem. If you can vibe-code it, we're not doing it.
We're also launching a Flows product. We've always focused on campaigns — with Flows, we figured out where we can be useful on frequency and timing, which is a big data problem, not a rule-of-thumb problem or a one-time A/B test. Because if you're not updating continuously, your data is dated a week later. And then we're doing connected TV.
At the core of everything: maintain our lead as the best stack-ranking tool in the world. Our biggest team is ML and data science. The way to think about what we're doing is: how do we bring the magic of Meta's or Uber's ranking algorithm into e-commerce for email — and then bring it into every other channel.

[51:09]  ZAK
Nice. Anything else you'd like to share with everyone listening?

[51:14]  AARON
Not much — this was a pleasure, it's always fun chatting with you, and I'm excited to see you in person later this week. For any agencies listening who want to learn what Zach does: we partner with a ton of agencies. Our job is the ranking — we don't do creative and strategy. We want to be a great partner to everybody.
For any brand out there: I'm Aaron, find me on LinkedIn or at aaron@orita.ai. We do a free audit for everybody — because honestly, sometimes the data shows Orita won't be useful for a brand, and we'd rather tell you that upfront. We'll do the audit, share some insights, and hopefully help you figure out if we can actually drive value.

[51:14]  ZAK
I want to back that up — when I was first learning about Orita, I asked a couple of operators how much time it added to their workflow. They looked at me like I had six heads and said, 'No, no — you don't understand. It makes our job significantly easier.' That says it all.
Aaron, it was awesome hanging out. Thanks for coming on the podcast. And if you're serious about growing your e-commerce brand, grab a free revenue audit with the team at ECD Digital Strategy. We'll analyze your store, your marketing, and your retention channels to show you exactly where the biggest revenue opportunities are. Completely free — you'll walk away with a clear growth plan. Go to Klaviyo.com/ECD and book your audit today.

[52:19]  AARON
You too, man. Thanks for having me.