The AI Traps Wasting Your Time In An Ecommerce Brand - Mehtab Bhogal transcript

After like a thousand episodes of this podcast and my old version of my podcast, I am making a change, which is I am getting rid of a lot of guests and instead replacing them with just a few guests. And one of those guests is my friend Maytab Bogal, who is one of my favorite people to talk to in all of e-commerce. The whole idea here is there's a few people I've talked to over a long period of time of doing this that consistently I learn a ton from.

Maytab is one of those people and so I said to him, "Hey, could you come back more often? Maybe every couple months, maybe even a little more often than that. We'll kind of see." But Maytab, who's been on the show recently, is back on the show to really talk about >> to dress like you, too. That's why we're twinning today. >> Matching matching hats. Um I didn't even notice the shirts. Yeah. >> Maytab Maytab is here and he >> And he's in style. >> No way, your beard is way better than mine.

But I think that's because you have an ethnic advantage there. I'm only I'm only 37 and a half percent Middle Eastern and you're you're actually full-blooded Indian, right? So, yeah, there we go. Um So, okay, but Maytab is back and the thing is I Maytab is the the reason I wanted him to keep coming back is because he is so good at thinking about the details and financial clarity to how to actually operate an e-commerce business because that is what he does across a range of businesses all day, every day.

Invests in and helps operate e-commerce businesses. If you know Maytab, you know he thinks more clearly about this than almost anybody. So, Maytab, thanks for being here and for doing this and for for for being a recurring guest on the Enterprise podcast. >> Thank you for having me, doctor. >> How did that happen? How did you start calling me doctor? I don't I don't A few people have done this in my life and I like I Do I give off doctor vibes?

I don't understand. >> Yeah, you're the doctor of uh media buying. >> Okay, all right. Well, >> Doctor of marketeering, PhD. >> [laughter] >> Um Okay, let's Maytab, tell people again, just so people know right away, how many brands are you working on at a time right now? >> Um we have three brands that we're very very involved in day-to-day and then probably another kind of six that we're more extremely already equity or minority equity where we're not where we're probably meeting them more every kind of two to six weeks. >> What's um what's what's we? >> Tell them what we is. >> Uh my co-founder and I, sorry. >> And that is called Cart Ventures. >> Yep. >> And uh so you guys you guys are looking always for more people to to uh invest in, right?

More brands to invest in. >> Yeah, ideally brands with kind of 10 to 50 million in revenue if they're profitable. Um if they're distressed, we can go up to 100 million. Um our real sweet spot is special situations. So if anything like a partner a co-founder buyout's happening or something else where it's hard to raise traditional equity, then I think we're we're a great fit. >> Mhm. Okay. Uh so tell people where to find you just before we go any further in the conversation so people know. >> Yeah, we're just >> Cart Ventures.com k a r t a ventures.com and our contact us is there. >> And we actually do check our contact us and we do get back to people in a day, so we're uh we're pretty good about that. >> Good.

The link is in the show notes. So if you hear this conversation and you think, man, I want to have Metpro involved in my business the same way Andrew wants Metpro involved in his podcast, that's where to go to do that. Okay, that stuff's out of the way uh and let's talk right away about the thing that people came here for from the title, which is the impact AI is actually making in your brands. And this is a curiosity question for me.

The genesis of this question is uh you know, I'm working in an agency. It's very easy to see uh and we talk about this on the podcast fairly often what kind of impact AI is making in our business, uh how it's driving value for us. It's it's just fairly clear in a service business sort of how AI can fit. I I have been sort of curious from the outside for a little while about about what kind of uh impact AI is actually having on real e-commerce businesses.

And you are so operational that I was I thought like what are you guys doing right now that's actually that's actually creating value because there are I think some mid-wit AI traps in e-commerce business businesses and there are some broader studies that at least last year were showing that AI wasn't making huge financial impacts yet. So so I so I thought like you were the you were the exact person to talk to here. Tell me this where does your mind go when I ask that question?

What are the biggest places right away where you see AI being able to be implemented and let's help people implement. >> Yeah, I think structurally the way we've been thinking about it and the way we've seen it play out across our portfolio brands is someone who is a really good doer they're able to simply do more. So you have less and less need for that management layer. And let's say you're running an organization where you have I don't know 40 employees.

So you have maybe two layers like yourself, your kind of C-level layer, the VP layer or like some sort of manager layer and then the actual doers, individual contributors beneath that. Um that that need for the management layer is kind of gone or it's much much lower just because each individual person can do so much more and you can afford to pay them more too. So you might have had someone where maybe your org was something like three people each being paid I don't know $80,000 a year.

You can now have one person that's being paid $220,000 a year $200,000 a year but they can do the equivalent of 10 of those people's work. And maybe that person was like really good before but now they're really really good. It's like it feels like almost every role has the same leverage that a 10x dev does. Um or or like a 10x sales guy. Like I'm not Have you ever worked with a sales guy who's just outsells the whole department put together? >> I have not but I've heard tell. >> Yeah, and it just like that or like the 10x dev which I'm sure you've heard of too where they're just so much better than everyone else.

Um now I feel like every role kind of or most roles in an e-commerce organization have that opportunity. >> What would be exponentially better? >> can we talk specifically about the kinds of things that they're doing then? Um, when you see that person who's a doer I I like the basic framing here, which is essentially that like you eliminate a manager and and replace them with somebody who is a doer. I mean, the classic distinction of like manager versus an individual contributor.

This there's just more need for individual contributors. One thing I think about that right away by the way is like having talked to Taylor Holiday about this at an agency level, he's been sort of anti-management layer in agencies for a long time. Um, even well before this was an even AI concern. So, there may be a broader instinct here, which I think is is true of e-commerce in general, which is that it is a world where like people who can just get things done have outsized value in e-commerce businesses.

It's very entrepreneurial type type of businesses where it's like if you could just get stuff turned out, it's good. AI amplifies that, right? So, where do you see people actually doing that, making those contributions in your businesses? >> Yeah, let's go through my list here. So, uh the biggest thing I mean, I I feel like there's a lot of obvious things in marketing. You have like creative side, email for example. Um, I'm not sure if you feel this way about email.

I've always found the work is very very front-loaded. Like retention in general for a non-subscription business. Um, let's just like ignore subs, ignore subscription beauty, etc. If you're not running subs, it's very very front-loaded work where you can have an agency come in, do a really good sprint, get a lot done in let's say two to six weeks, and then more or less be churning out campaigns using AI and just modifying templates that you have.

Um, so it's not really like one person can do that role in a 15 minutes a day. That's how I feel about it. >> If you like this episode with me and May Tab, you're going to like my future episodes with May Tab. So, subscribe wherever you're watching or listening. Do me a favor also and hit that like button wherever you're doing that. Leave a comment with any questions, any thoughts. I would love your ideas on how AI is impacting your business in measurable, revenue-driving ways.

Uh, I read and respond to basically every comment on every episode. So uh so do that. >> [music] >> Thanks so much. So in that case just I just want to keep diving into as much clarity as possible. In that case the AI enabled individual contributor that we're describing here, the doer, right? That person is generating campaigns with AI. Right? And they're just they're just sort of using some kind of workflow to generate the the visuals and the copy.

They're QAing it, making sure it all feels about right. Getting it all sent out in like 15 minutes a day, etc. What is that that's basically what they're doing for that all the time, right? >> Yeah, and they have their hand on They have their hand on the data. Just making sure you know, like open rates aren't collapsing or something like that. Click through rates are trending in the right direction, etc. >> Yep. And then what is the agency doing in in those sprints? >> Yeah, so for example um since I guess like when we get involved with a brand um typically they've been writing email in house or they'll be using what I'll call like a less than great agency.

Um We have two really dorky agencies I like a lot. Just run by total nerds. Uh one guy used to run >> You can shout them out if you want. I don't know or if you don't want to that's okay, too. >> No, I I can't actually and I only ever remember his name. I can never remember his actual name. Oh, it's Brian at Crucial Point. So >> Okay. >> He's a total dork. Uh he used to run emails for Who's that giant retailer uh that always used to get Who he always used to fight with the uh investors trying to short their stock?

Overstock. Overstock.com. Yeah. >> Okay. >> Um and he is like the nerdiest guy I've ever met on email. So they're very good technically. So like if you need someone to come in, clean up everything technically, like oh, you should be suppressing, let's say, Gmail or Hotmail along these specific cadences, like he will know exactly how to get that set up. So they're just good for the speed. Like again, you could sit there and do it yourself, but it's just not worth your time.

Like you're going to lose more money than the time it would take to hire them, have them go in, and just set everything up for you in like let's call it 2 weeks. >> Yeah. So, we have Yeah. >> Good. Good. Um so, we'll typically have like an agency like that come in, audit, clean things up, etc. Um And even help with the initial for >> technical expertise around around lists and send strategies and uh like just making sure that you're not um that you're not hitting spam unnecessarily and you know all that kind of stuff, right?

And cuz that is I mean it's there there are some technical things like this where the the you know the moment you say to me like oh just replace your email agency with AI like I just I just can I just feel like this is what's going to happen, right? It's like or like >> know at the higher level there's a lot of dorky stuff you can do. Like this is less common now, but it used to be very common to rotate your IPs for sends between your welcome flows and other ones to keep them hot.

Uh >> Yeah. >> Like let's say you have a specific flow that's um more likely to be target like marked as spam or something and then you have another IP that you're using for your welcome flow and you know that one's always being opened has a really good reputation. You may revolve those IPs like I This is only something that's worth doing at scale, but um you know are you going to sit there and figure it out yourself like probably not.

You either need a really good email guy who probably cannot justify unless you're huge like let's say 150 million plus. Um or you could just use an agency for a sprint. So. >> My longest 10-year sponsor on this show is More Staffing and that's because More Staffing is a fantastic service that connects you to incredible talent in the Philippines and all the places in your e-commerce business that really move the needle for you.

That means not just intro level $5 an hour virtual assistant type talent, but like high quality people with great e-commerce resumes who spent a lot of time in e-commerce businesses over a long time across every part of the business. It's operations, supply chain, marketing, wherever you need that and that's >> [music] >> that's certainly coordinator level talent, manager level talent, even director and executive level level talent in your business.

Everything I touch in some way, every business I'm working on, is impacted by talent from the Philippines that was that would came to my business from More Staffing because they are just great. They will help you come up with a job description. They will help you source resumes. They will pre-interview every candidate that comes to you so you're not sifting through hundreds of resumes, thousands of resumes from overseas, but instead they'll bring to you the top two or three people relative to your job description.

They'll even give you a one-year guarantee on any hire you make. So if that person doesn't work out in your business for a year, they will help you replace that person at no additional head hunting fee. Which is great. Go to morestaffing.co/af. morestaffing.co/af to get connected to incredible talent in the Philippines so you can keep your off axes as a percentage of revenue low while getting high impact talent into your business.

Go check it out right now. If you're not serious about offshoring now, you should be. It's just a really really big opportunity. Go do it. morestaffing.co/af. This is a sort of related point. One of the big traps I think including for supplement brands and others. Tell me if you agree with this. Is hiring full-time retention people internally on the belief that oh, our retention will get better if I like if I like put somebody on sort of like life cycle marketing.

Like I think it's possible that person could exist and be impactful, but that their job is almost never related to email and SMS if they're if they're impactful. It's more about like offers and and like products and upsell and things like that. And that email and SMS is a small vehicle, but the the mistake I see a lot of brands with material LTV. Sometimes people do this who have relatively low LTV and they think they're going to solve that by hiring a person to work on the LTV or like excuse me, relatively low like retention rate.

So sometimes they say I'm going to solve the retention rate there. That's an even bigger trap because your retention rate is not low because of your email and SMS strategy. >> No. >> Um, uh, your retention rate is low because of what the product is, by the way. That's the reason why it is. Uh, but then but then on the other side, when brands do have higher retention, they'll they think that the solve to getting better retention is something related to sort of like the post-purchase tactical, you know, uh, work, when actually it's related to who you acquire, at what offer you acquire them from, what products go into the order, how you generate more products additionally to that, and those kind of things.

That's what actually drives retention in the business. Um, do you agree with that? >> Yeah, 100%. I think, um, the offer, specifically like you mentioned, is is the biggest thing. And then kind of what happens during the lifetime of that sub. And I always think of that more as like a product. I don't want to say it's like a product role, but it really is more of a you're playing with product, you're playing with the offer more than you are the actual You're not sitting there tinkering with SMS campaigns, email campaigns, like that much.

You're really coming up with new LPs, new offers, new messaging, etc. And and like you said, it's very hard to escape your category. Like if you're in apparel, it's very hard to You're not going to double your your repurchase rates. >> Yep. And if that's true, then it reinforces what you're saying about the use of AI, which is like you can now take a thing that already ought to be reasonably priced and make it materially cheaper, um, and faster, and allow somebody who's just sort of a doer to get it done while occasionally auditing the the realities of the technical setup, like you said, like every every 6 months or so you have the agency come and just make sure everything is good. >> Yep. >> Yep.

Cool. Okay, I love that. That's great. What else? What else? How else How else are you using AI actively in in, uh, in your in your actual portfolio of brands? >> Yeah, so I don't I don't want to get stuck too much on the marketing side just because I feel like you've covered it quite a bit before with Patrick. You had a few good episodes. Um obviously the dev side like a few of our sites are very very dev intensive so they run a bunch of meta fields as work around for variant limitations uh and it's multi-item at checkout.

So I don't know if you know this but Shopify has a lot of issues if you go past five to eight items on on checkout. >> I didn't know that. >> Yeah, it'll it'll chug way more and there's site speed impacts it way more as well. Um so if you've ever run a website let's say like average items per checkout are 12 like cut uh you will run into way more dev-related issues on releases and you'll see the impact of site speed way more. >> Hm. >> Um So there's your fun Shopify fact of the day. >> Yeah, yeah. >> uh obviously running QC for that's become way way easier.

Uh basically making a checklist for AI to work through before release goes out so and people are thinking oh what a big like not a huge deal but we've had crazy things happen on releases. I'm sure you have too. Um especially for a website that's very complicated. Um you'll have major issues that will cost you I don't like 20 to 40,000 dollars a day in just problems even if you catch it after two hours. Like our team's pretty good about that.

But you know you do the math and you look at huge drop off in conversions like oh well that botched release literally cost us like 40 grand. So >> Yep. >> Um >> Yep. >> that's happening a lot less. Obviously being able to spawn landing pages way way way quicker. >> Is that because it's somebody's whose job do you give to go solve the dev problem with a with a AI stuff? Like how does AI actually solve that problem? >> So that's our in-house staff um and he's basically writing adversarial reviews on themes before they're released. >> [laughter] >> Um like okay, that makes sense.

That's that's so interesting. And so you have >> you do the adversarial review mode with Claude? >> No. No. >> Um if you just go to UltraCode it'll kind of do it almost intuitively. I forget how he uses it. He does it a different way. The way I do it for anything that I want adversarial review I just to Ultra code. Uh you don't even have to be working on something that's code related. You can drop something in there and just ask it to do an adversarial review and it'll spawn like dozens of agents to basically look over it and critique itself. >> Maybe I don't know what an adversarial review review is.

Can you explain that? >> Um basically if it wants to give you a certain output, it'll tear itself apart with a bunch of agents. So it'll spawn agents to attack itself and then refine the output. So you you end up with way less errors. So that's the way we do stuff like uh anything numbers related, we use that mode for specifically to avoid um cuz the error rate goes way down. So if you're ever doing reporting or something, you should definitely use that mode. >> Say more.

Say more what you do you do with with numbers like just this is a tangent a little bit but but tell me more about what you're doing there. >> Um so me personally or our team? >> Either way. >> Okay. So personally if I'm ever building out a model for deal we're looking at or something like that, I will have more or less um a few different ideas for how things could go or what the structure should look like and what the impact will be.

I'll just plop my ideas into Claude and then I'll switch it to Ultra code mode um and basically have it model out a bunch of build me a model where a bunch of different scenarios live and it will just save me so much more time versus having done that manually. >> Yeah. >> Um and give you an example like the financials we get will be totally different. Let's say so a company's using NetSuite and they're using their accounting module, it's very different having and getting the data versus something like I don't know like QuickBooks or something.

Um so it it's really nice cuz I don't have to do nearly as much work just to build a quick thesis to see if the math maths. Um and we're never we're never making an investment based on being too precise just because I mean if anyone's ever worked in D2C or consumer knows, it's so volatile, right? It's pretty stupid to try and make an investment assuming you'll be within like 10 to 20% of a thesis. Uh so we want to be directionally correct.

Like I I don't actually care if the numbers are that precise. They just need to be directionally correct. >> Mhm. >> So, for me it's perfect for that use case. And then I'll usually go out and I'll just double-check it, make sure it's good. But, it's been really good lately, especially as of like 4.7, 4.8 on Ultra Code. >> Yeah. >> It's been really good, way better than before. >> So, the basic idea here is that you're using that to create models that challenge your core model and challenge your assumptions.

Um and it sounds like the same thing at the code level with with a theme where you're trying to figure out where is this going to break and you're using and you're using agents to go like attack it. >> Yeah. >> Like I said. >> And if you guys have software set up for copywriting, for example, it'll do a much better job and actually adhere to the rules, too. Um by doing If you use Ultra Code for copywriting, uh which is probably not the most efficient thing token-wise, but it certainly works way, way, way better. >> Yeah. >> Um and then we go to other things, too, like I I think the biggest Like for us, one of the biggest wins has been disintermediating the need to ping someone to go and get data for you.

This is not a problem if you're running a small business where like your only channel is Shopify and you're able to pull most critical things out of Shopify reporting. But, once you're at scale and uh you have a lot of different forms of data that you're ingesting, multiple channels, etc., the really bad thing about most ERPs is that you're stuck doing like a manual SQL query to pull the data that you want. So, you'll have a data guy.

Uh and for for a long time, we've always had a guy at the company whose job was to do nothing but run various reports that people want. >> Yep. >> And and the bad thing with that is, you know, if I have a thesis and it's 10:00 p.m. and I'm just bored on a Friday, like I can't go and explore a bunch of ideas that I have. Like I I can cuz I know enough basic like SQL, it's not an issue, but people on our team can't. So, >> I mean, I I I'll say I can't and I I don't know SQL.

And this has always been a limitation for me, which is that I'm a pretty data-oriented guy, but there's this hard limit against my technical skills. And I'm not going to go learn SQL. So, like what you're describing is totally right. Like and I I this is a great use case. Just like where you've got somebody who's curious and interested in something and then they can go chase down with AI that um that theory or that idea pretty fast and they can go, "Oh, that's a dead end.

I don't need to explore that." Or actually there's something there. I think of all a million product searches that I've done over time where it's like, "Oh, I wonder like you know, is this product moving in a way that I didn't expect?" I don't know, you know, whatever. You just you start asking some kind of question and uh and now you can just access that information so much faster and get get them end to end. Yeah, go ahead. >> Yeah, like a great example is um we use fulfill for one of our brands for the um like their manufacturing uh systems like really really good fulfillment etc.

So uh you know, I was just wondering one time, uh "Do our orders that receive product faster have a higher higher likelihood of reordering versus people who receive it slower?" Um so I just asked and was able to go and pull all of that down very quickly. And I remember that used to actually literally take me like 45 minutes to an hour to pull down. It's a And that's like I'm pretty fast and I's just cuz I've done that multiple times, but if someone is not uh doesn't hasn't done it multiple times, it would've taken them like 4 hours.

Uh and where this gets really cool, obviously >> Wait, what was the answer? What was the answer to that question? >> Um so it always varies. Like we do this analysis with every single company we get involved with. For us, it did matter for a very specific cohort buying a specific type of product. So we did launch it. So now we have a facility in Indianapolis for that company, Mexico, and India. Um so it's fun. >> Because it because it because it made it enough of a difference if you got it out to them faster. >> Yeah, yeah, exactly.

Um and it was really more of an AOV lift as a port as opposed to like a pure CRO lift. Yeah, makes sense. Yeah. Um they felt more confident about the second purchase, so uh but I think where it gets really interesting is we have people on the factory floor, um let's call them like uh for example, like our manufacturing, the guy who's heading up manufacturing for one of our businesses, you know, he's walking around, he's on the factory floor 8 hours a day, then he's at his computer 2 hours a day.

It's hot. It's like I'm just trying to think what the temperature is in Celsius and uh, Fahrenheit. It's like, you know, it's in the 80s, maybe the 90s. It's like really hot where he is. And it's like, if you have a thesis about something, are you going to go sit down and like quickly punch it in while you're on the factory floor doing other things? No, but it's cool when you can just talk to Claude um, using our fulfills MCP uh, lifetime manager as an MCP, for example, now, et cetera.

And just have all that data right there, test a thesis really quickly. In his case, he can see like, oh, who's the top who's the top performing person making X widget, right? He can check that really quickly and he can see like, oh, okay, so I should talk to them to figure out what they're doing that other people are not. Um, and that's what what's really cool cuz before he would have to ping our data guy who's in India and then there's a time zone gap and then he'd have to wait and then he'd get everything back.

So, it's completely sped the whole process up massively and a lot of people are really excited about this, too. When we were hiring, um, we were trying to fill a role for one of our pick and pack operations, we showed them the cool stuff they get to play around with. They were so excited. Like the good candidates will get really excited about not having to wait to explore a thesis. >> Yep. Yep. That's that's yeah, that's that's really interesting.

Um, do you can you think of any examples? You you mentioned the um, the speed of delivery example. Can you think of any other examples where that has yielded something really insightful and interesting? Cuz I think one of the Yeah, anyway, I'll just leave the question there. Uh, can can you think of any examples of that that that that are tangible examples of like, yeah, I did this and here's what I learned and it worked great. >> Okay, this is pretty funny.

This is very ops-oriented. Uh, we were trying to figure out why we had an issue with a lot of shipments arriving damaged. Um, so we had Claude just run an analysis to see if there's any correlation between size of the shipment or items per order and damage and there was. And we said, "Okay, why is this going on?" We can't really figure it out. And then it made sense cuz it spit out that the um like edge crush strength of the cardboard wasn't high enough that because the boxes are so big at like FedEx or whatever, they're throwing other boxes on top and it's it's like getting squished like that.

And it pointed that out and I'm like, "Oh, that's just a good point. Never thought of that." Cuz the small boxes never get stuff on top of them. They're the ones that are on top. But if you have a big box and you have like weaker cardboard, they'll they'll just start throwing stuff on top. There's like a microwave on it. I'm like, "Oh, this is why it's damaged." >> That's so good. That's so good. >> that was helpful.

And then figuring out what the optimal like edge crush strength is um and and all that. So, that was pretty fun. >> Yeah. Um Okay, let's keep going. You um you talked about >> is my other favorite one. If you're bad at sourcing, uh for sure just plug it in there. Tell it to go look at something like ImportYeti. ImportYeti shows you where a company is imported from and who their factories are. It's totally free to use.

This actually used to be a service that we did for you. Yeah, importyeti.com is a all-time. >> Uh >> Like Yeti, like the scary monster, like the Sasquatch. >> Yeah. >> Um If you go there, you type in any brand, you can see which factories they've been ordering from and who they import from and all their important ledgers. Um You can hide this. Uh You just have to send us an email to the government and you can mask it if you want, but yeah, with this this is a great place to start sourcing.

Um and you can use something on the AI side to help you just quickly pull down a ton of suppliers that are relevant in your industry. Tell it to go and use a website like ImportYeti. Uh and this will give you a very very good start and depending on how built out say your agent is, whether it's on email address, you can have it start contacting them, filtering people out, etc. But this is how we typically start our sourcing process. >> That is so fascinating.

I wonder >> You can monitor competitor volume, etc. So, we're That's great. >> That is crazy. Um okay, I had no idea this existed. That's really, really cool. Um Okay, so anyway, so so again, say more about how you're actually using AI to help with that. What is What is AI adding to ImportYeti in in terms of what you're >> for example, it'll go through You can actually search uh ImportYeti by HTS code, too. So, if you have a product with a very specific HTS code, you can go and see all the manufacturers in that space that are exporting volume to the US, and you can filter down by the amount of volume they're pushing.

And it will go out and just like for example, we set ours up with an email address that doesn't go to spam. So, they'll just go out and contact them, get the initial conversation going, it'll put the initial RFQ in to get a quote, and see if they're reasonable or if they're stupid. And then if they're being stupid, I'll just tell them to stop being stupid, and they're losing the quote, and here's why, and they need to fix it right away.

Um just cuz some of them will throw like a number at you if they feel like you're some stupid American company, and you'll just overpay for everything. So, um yeah, we we do all that, and then depending on the country that we're sourcing in, I I'd say like obviously China is like the easiest company to source uh country to source in just cuz the infrastructure is all there. It'll hand it off to the relevant like local office. >> There are not many software tools that I'd consider core [snorts] tools for the e-commerce operators' software stack, but IntelliGems is one of them because you should be all the time efforting the places in your business where you can make a really big impact on profit with every click that makes it to your website, and IntelliGems makes it possible for you to do that by being a profit optimization tool that really goes beyond CRO testing.

You know IntelliGems as a beefed-up CRO tool that allows you to test really big needle-moving things like your pricing, and your offers, and your free shipping thresholds, and your shipping charges, and all those things, even shop by theme tests. You know Intellijem's can do all that. You know Intellijem's can do that while also spitting out the answers to those tests at the level of profit per visit by tying into your cog's data in Shopify, giving [music] you that information there.

That's all the stuff that's Intellijem's has been doing for a long time, but the tool's really getting beefed up in a lot of other ways as well. There's now post-purchase and pre-purchase upsells that are part of that. You can now test in cart and in checkout as well. So, just is a is is a really robust tool for making every click to your website more valuable to your business, and that's really the goal of the tool while giving you great insight about why and how it's doing it.

If you want to get a sense of what makes Intellijem's awesome and how it could work in your business, go to Intellijem's.io/audit right now. So, a u d i t. Intellijem's.io/audit. Tell Intellijem's your web address. Tell [music] it what you are trying to do. It just says there's a a little button like increase your subscription rate, increase your number of orders, whatever it is, and then it will spit out for you a suggestion for free of the first few tests that you should run on your website powered by AI, of course.

Really, really cool tool for you to go go explore and see how this is going to work for you. Intellijem's.io/audit, or if you're ready to go sign up right now, go to Intellijem's.io. Use the code ferris20, f a r r i s 20, my last name. Get 20% off your first 3 months. Really awesome tool. Go use it. Intellijem's is great. >> It's really interesting. Okay, that's awesome. That's a really good use of it. That's that's really clever.

Again, so you've done this and built sourcing relationships through that process that have worked. >> Yep. >> Yep. That's great. Um Okay, what about what about at the level of product more more broadly or sourcing or anything like that? Anything else you're doing sort of on that side of the business operationally that is compelling with AI? The thing I think that Yeah, the thing I think that that ties together the examples you have so far is that they're all pretty high impact.

Um you know, like thinking about like customers experience with the product when it gets delivered both timeline and not having the product damage. That's like a really really big deal in in a business, you know, or product sourcing and finding, you know, better manufacturers and getting better pricing and being able to use that as a negotiation tool. It's a really really big deal. What about What about the rest of the operational product side? >> Yeah.

So, and what I share floral brands since it does so much and uh like retail's very different than the B2C side of the business and I'm not sure how much retail have you done? >> Uh well, so I've had plenty of clients who have meaningful and material retail, but I'm not I'm not closely involved with the actual operations of that side of things. Like the the place that our work tends to stop there is like can we draw any correlation between our ad spend and our retail sell-through and that kind of stuff, but nobody's asking me to help their retail operations. >> So, basically depending on the retailer, you have different you have different levels of product channel fit, right?

Which I think we've talked about before, but if I'm selling to a convenience store, their average price point's going to be very different than if I'm selling to Target and if that's going to live on a certain cap, right? So, you basically have to make products for everything and for us specifically, we have we have a BOM. We have a lot of different components and ways to make specific product. So, in our case, we basically have AI and it has access to our ERP, which is Velo, and we can say, "Hey, I need to make a product for X retailer.

It needs to fit this specific shelf dimension and it needs to hit this specific price point in terms of our available COGS." >> Yeah. >> And can you spin up like 10 or 20 different variants that we could do? And before this, it would take like a week just because of a lot of complexity to come up with like 20 good ideas. Now, you obviously get that in like 40 minutes, right? Yeah. And then it's cool because you can take it to the next step, too, and you can start to roughly visualize these things as well just to make sure they won't look horrible once they're built.

But now you can just do that and then hand it off to the team that will build the prototypes. So it's very nice. Yeah. >> That is a really good example. What it reminds me of actually is like in the um is is just the ability to sort of take product level insight and expertise and make it more available to a broader range of people, you know? So what you're describing there is like somebody normally would have to go and think through, okay, what is what are all of our what what is everything goes into our bill of materials, right?

What is everything what are all the little sub pieces of this? How much money does each one of them cost, etc. Like it's just a lot of tedium involved in that kind of process. >> mind like and for that specific company, there's a lot of layers of complexity too. Like certain types of BOM are more resilient to damage than others. So if you're living in retail, you need to make sure you're not using specific BOM that's easily damaged, etc.

So >> And all of that requires this deep knowledge of each of those materials, right? Like each of the actual pieces of it. And so the ability to sort of run through that really fast. And obviously you have to check it and make sure everything comes out well, but like to get some to get real giant progress on that, you know, you probably get 80% plus of the job done doing that, you know? And and get really far. I when I'm you know, I still have people ask me about this and I'll someone will give me an update, but I still have um uh this solid cologne business that I've got like kind of brewing on the side.

Hasn't launched yet. Um and one of the things that we found with that was like we were not getting enough projection on the initial um formulas and that's a big deal for people wearing cologne. So this is like the the the cologne smells good like the the the actual scent we really liked, but the but it just wasn't projecting enough so people couldn't smell it on you. Um and so I like you know, I have the um the formula for the product.

And so I just like handed that at some point to um an LLM and said like I'm not getting enough projection. What do I need to do to get more projection here? Now, I know nothing about fragrance formulation, right? But like I was able to sort of get some answers. And we haven't actually gone down the route yet of sort of checking the next level of formulation and trying to you know change the product to get there. But and so we'll see if that actually works.

But what that allows me to do is have is to have an experience with the product as the operator of the business and to think like there's there's something missing about this. Like how do I go make it better? In a in a realm that I actually have no technical expertise in, you know? And so now it allows me to say like I don't know how to actually make the technical side of this. And and then I can go and of course test it and all that kind of stuff that you normally have to do.

But it just it just allowed me to go access a thought in a way that I otherwise never would have been able to. It's really cool. It's really cool to be able to potentially go do that without having to go through rounds and rounds with some kind of formulator or whatever it is, you know? >> Totally. >> Um at least as a >> And you can even come to them with different ideas, right? Like you can come to them with three or four ideas and be like, "Oh, could we do this?" And then they can tell you no or yes.

I mean, I I do that with lawyers too all the time cuz we do a lot of distressed stuff where it's like, "Okay, I don't technically know the law" supposedly. But I I think I'm actually better than most lawyers. But I'll come to them now with like four or five very very very thought out ideas and they'll say like, "Oh yeah, this will work or this won't work for this technical reason." So but now I probably save 80% of the time.

Like I literally would spend like let's say 20 30 grand exploring a few different ideas. And now that cost is more like four grand. So >> Right. That's a huge Yeah, that's a huge difference. Yeah. I think generally it's it's one good way to think about this, which is like access to technical expertise >> Mhm. >> um in some places where you could do that. Now, um uh and and again like I think like you said you go from 20 to 30 grand to four grand you don't go to zero, you know?

The point is not that you like that you like eliminate all costs completely. The point is that you like and run around a lot of the biggest costs in the stuff that is more tedious and some of that you know just begins to come up. >> technical stuff too. It's like okay, in that case I I wanted to put a temporary restraining order on a creditor that a business that we were looking at was dealing with and I was like okay, if we wanted to file a TRO, how does that work?

Where do we file it? Etc. Um and like we've done it before, but I was I was just curious because of the jurisdictions like how would that work? And then it actually gave me pretty good answers and I thought were actually better than the original law firm I talked to. So >> Yeah. Yeah. And the point here is like it's the the question is almost so broad as to be kind of silly, right? Like how's AI impacting your business?

Like of course it's impacting in some way. But my hope for this episode is that people will just like have more ideas of places where there's potential impact that can be real in some really specific ways that they aren't thinking about yet just because like I at least for me it's still sometimes hard for me to remember like oh wait, I can do this, you know, like I just like you're just so used to operating in a different way of thinking that you know, that's that's that's the idea.

Tell me the traps. What are the places where you see people trying to use AI in e-commerce businesses um that are that are um actually a bad use of the tool and and don't need that. And I'll tell you one thing really fast with this. One thing I'm concerned about is like there are really good quality e-commerce businesses already that I think like I think in some ways ought to have a really skeptical attitude towards trying to do almost anything with AI.

Like like it's almost just like if the business is already working really really well and growing there's a big like just time suck trap in general of like wait, if it's working like don't don't go and like yeah, there's a million examples of this, but but um but you don't have to go use this anyway, you know, if your if your business is really working and working well and growing and and all those things like great, you know, um let it keep doing those things.

But um anyway, specifically I'm curious if you've seen places where in your your portfolio of brands this is um this has been a bad use of time or anything like that. >> I'd say anytime you're trying to automate something where an employee offshore, let's say they cost $600 to $1200, um like a very entry-level person could be doing it. I would not waste too much time with it unless it's a mission-critical process. Um or and I'm kind of along the same lines like I see people trying to >> Such a good example, May tab, because it's so tantalizing to do that.

It's just so tantalizing. Keep going. >> It's like congratulations, you saved basically no money. Um >> Right. >> Or or in the other case, like to get to fully automated costs way more effort and time than it does to have a human in the loop. So, a lot of our systems do run with a human in the loop where there's someone doing something in there. And it was just 100 times faster to build it, and it works way more reliably, too.

And you should really optimize for speed as opposed to being reliable, if that makes sense. Just because the fundamental is going to change so much in the next 6 months, it doesn't matter. It's kind of stupid to optimize for it. Like imagine if you built something using N8N a while ago, and now you're like you probably feel really stupid. Like honestly, a lot of that ends up being a waste of time. Um >> That that is such a good point.

And you're saying just cuz the model's gotten better enough to where like all the little checks you would have put into your N8N flow or whatever like would have or just useless now. >> Yeah, totally. I mean, like the actual base model is so good, you don't need any of that. It's just like okay. Um or specifically, like don't don't be a guy who's trying to vibe code an ERP or whatever. I think that's such a waste of time.

And if if you've never actually done any big project with developers, you know that the building natural product is really not that hard. It's like okay, yeah, you built the product, but now nailing all the little use cases and making sure it's reliable, like something that needs to be mission-critical software, just buy it. Like you're never going to vibe code Klaviyo. I I talked to someone the other day who was like, "I wonder if I just vibe code Clavio." I'm like, "Are you Like, this is really stupid.

Don't Don't do that. Like, I'm going to come over and beat you with my shoe if you do that." Cuz that the back end's what's really hard. It's not like the cool UX that's hard. There's nothing like hard about that. What's really hard is the back end that they've built and no one's been able to replicate for really, really good reason cuz it's it's really, really hard to deal. Um And same with like your ERP. Like, don't don't mess with that.

If there's a mistake, your inventory's going to be all wrong and you're going to cost yourself way more money. Um So, I would just tell we don't bother with any of that. Um especially if the vendor's like reasonably good. Like, do I like paying Clavio a ton of money? No, but do I know that there's nobody else better? Yes. Do I think like I could build a better version? No. Um >> Yeah. >> So, that is stuff like that silly.

Like, just use it for the 80/20 of impact in your business. Specifically, cutting out OpEx or reducing OpEx is where I think it is. Um And high-level, like I I talked to a guy on Friday night, for example. He was running his business at I think 3% of his spend on people. Uh which is great. Like, that's very good. I'd say good is 5%. He was exceptional at 3%. Um and that includes agencies and stuff, too, by the way. So, very, very exceptional business.

It's growing insanely fast. I think he went from like 3 million a month to 8 million a month within a month. And I think I think he's going to be doing like 12 million a month. >> business? >> Yeah. I think they're going to be doing 12 to 15 million a month like within 2 months. It's crazy to see. Great guy, really smart, totally dialed in. Um but he was like, "You know, I feel I don't use AI enough." And I'm I'm like, "Look, dude, your your cost base is so good on the people side, I wouldn't waste too much time there." Um >> the example I'm talking about when I'm saying like it is a shiny object syndrome problem.

Like, "I feel like I don't use AI enough." when I'm getting to 12 million a month potentially really, really fast in an e-commerce >> It's crazy. >> It absolutely does not matter. Who cares about using AI more? Just do whatever you are doing and keep doing it. Like, how much Yeah, just like Yeah, it is Uh Yeah, at 3% of spent revenue being people like is just like insane. Now, one thing we should say there, by the way, is that part of the reason 3% of revenue can be people is because the business has reached some scale. >> Yeah.

Yeah, for sure. >> So, like because because this is a classic thing with e-commerce all backs, which is that it scales on it scales really really well. But, then keep going. >> Yeah. And I think AI is like a great tool for amplifying the output of a process. So, for example, if I want more output out of creative, yes, you can use AI in the process to improve that. Um but don't like don't try to insert AI where it doesn't feel like it's a natural fit or it doesn't make sense.

And if you don't have great ideas, just go hire an agency that's good at it. Um or you know has good output for the dollar, if that makes sense. Like >> Yep. >> For example, if I know AGF is able to produce creatives at X dollars, I should hire them just like coach me through how to do that and how to build the machine to do that at scale. Um Mhm. Yeah. >> Yep. Um okay. Anything we missed here? >> No. Um Nothing really.

I I don't think >> of other examples I've seen from from clients of mine. Because the the genesis of this whole conversation for me is just this idea that like Yeah, that I'm I'm curious where the impact is actually happening for e-commerce businesses and um Yeah, there just there does seem like a lot of traps for them to me that it's actually if you just focus on the most important things um that that uh that that there's some ways in which it can be a trap.

I I like your basic way of framing this, which is get more people in your business who are naturally individual contributor types and who are really good at that and then empower them to make more of an impact across a broader range of things. So, this is almost like a a broader strategic way of thinking about who to hire and how the process of the organization works. And then that plays out in all these individual areas.

If you're a supply chain person, it plays out the ways you set in supply chain. If you're and you know, a marketing person then okay, landing page and and email and and all of that can be faster and maybe there's some element of that. >> say that's interesting I thought. Say this to you, if you have a team where you feel like you have to keep training them in AI and they're not naturally curious, I think it's a bit like it's a bit like if you've ever hired someone and they're not good with computers or you've worked with someone who's just not good with computers.

It's like All right, it's 2026. If I spend a bunch of money on training you, it's not going to change anything. Like you you don't have to you need to be trained trained. You just don't care. >> Um >> Yeah. >> And I think they'll probably have to train for culture reasons, unfortunately, if they don't get on the boat. >> Yep. And uh you know, in general I found it's extremely hard to stop people from doing the thing that they want to do.

Uh and so if the thing they want to do is to not use it then yeah, then and I'm not showing any move towards that. I think that's it's probably right. Um all right, I think that's enough. We we have some other topics we'll hit next time, but um >> But >> But I think that's good. Thanks May Tab. Uh once again, Um uh work with May Tab. You can also follow May Tab as always a great follow on X. What's your actual handle? >> Um May Tab Karta.

M E H T A B >> and then K A R T A. >> Yep. Um Cool. All right, well we'll see you again soon. Thanks for doing that. >> Cool. Thank you. >> Thanks so much to May Tab for coming on and doing that. I think that was exactly the kind of conversation we expect to keep having. Real conversations about real impact operationally that is happening in May Tab's businesses. As you can tell, he just thinks really deeply and carefully about how to drive profit and growth [music] in e-commerce businesses.

There's just almost nobody I know of who's better at thinking about that. [music] So, go reach out to them. Seriously, if you're the kind of business that would like to have him on your team and his team on your team, go do that. Maytab is awesome, great dude, and is really, really helpful. Just one of my favorite people in e-commerce. cardadventures.com is the place to do that, like I said, and the link is in the show notes.

There, go give Maytab a follow as well. If you want AJF Growth working on your business, reach out to me by going to ajfgrowth.com, filling out the intake form on that website, and we'll get back to you about whether or not we're a fit for you. Our growth is pretty much capped for the rest of this year so far as we can tell. Um but but we're always uh in conversation with people who either want to jump on our wait list or have a call and just see if maybe in the future when something opens up we're the right agency fit for you.

So, go to ajfgrowth.com, do that. Don't forget to like [music] this video or this podcast, wherever you're watching or listening, subscribe, wherever you're watching or listening, so you don't miss any episodes in the future. Email me, podcast@ajfgrowth.com. That's everything. Thanks so much. Oh, oh, sponsors. Intelligence, More Staffing, they're both great. Links for them in the show notes are in the show notes. Go check them out.

Go do it right now. Thanks so much.