Latest / Leaders on Shopify with John Readman / The Shopify secret nobody talks about (and why AI could cost you 4% of every sale)

Transcript

The Shopify secret nobody talks about (and why AI could cost you 4% of every sale) · Leaders on Shopify with John Readman

• John Readman: Welcome back to Leaders on Shopify. So I'm really pleased to have Keith joining us from Milk Bottle Labs today. So I'll let introduce himself, introduce his business, and then we'll get into all things Shopify. Keith Matthews: Thanks for having me John, thanks for having me. So I own Milk Bottle Labs, we are a kind of a mid-sized agency based in Dublin, we're Shopify plus accredited agency and we're around since, officially since 2015, I found Shopify around 2010 and then we started as an agency 2015, officially, we actually got accredited in 2014 as an official Shopify expert. And we've been, I suppose, I'm one of the OGs now, we're around, know, fully operating around nine years. And the only two pieces of software that we would regard ourselves as experts in is Klaviyo and Shopify. And obviously the two of them go hand in hand. Our customer base is primarily Ireland, the UK and the East coast of the States. Although we just launched a store in San Francisco last week. And we, you know, personally, I just love the platform. ⁓ always getting better you know, mean, the long term for us is to stay doing exactly what we're doing. So that's us. That's Milk Bottle John Readman: And when you found Shopify, what was it, because you've been in this world a long time, what was it that sort of thought, hang on, I'm gonna back this horse and I'm gonna go all in on Shopify? Because there's lots of other agencies out there who are still thinking, oh, well, maybe we can get more fees if we stay on big commerce, or actually we're gonna build something really complicated in Magento, although who would do that right now? So what was it for you and why did you go all in on Shopify? Keith Matthews: it's a good question and you fit on a couple of things. For me personally, was, 20 years ago I started buying domain names. ⁓ And for those who remember, to open a bank account, you had to get a, you had to open your bank account with let's say Barclays and then Barclays would have ⁓ a merchant account section inside Barclays. So to actually transact online, and to set up online was the barriers to entry back in early 2000s, late 1990s, 1990s, 2000s were phenomenal. So I was a relatively young kid just at a university who had a bank of domain names. I started buying domain names, 1999, 2000s. And every time I attempted to launch a business, I was just blocked. You needed to be registered for this, you needed to register for that, you needed to have two years bank statements. So I just kind of sat on those domain names and then I just stayed in touch with with tech. did a post grad in web development back in 2002, I think. And of course that was just HTML, you know, so, and I'm not a developer, interestingly now with the advent of Claude code and lovable, can actually, it's interesting how it's such an advantage now that I can actually read code. So then I heard about Shopify and when I heard about Shopify that was back in the day when they actually had the number of their customers on the front of the website. So when I found Shopify they had a ticker and it read 65,000 customers and then it moved to 85,000 and then it moved to 100 and I think when it got to 125,000 they took it off the front of the website. Signed up to a trial. and I couldn't believe that you could stitch together payment gateways, stitch together the domain and within 48 hours, or three days to allow the domains in those days to propagate, you had an online store. So what actually happened then was I built a few online stores and then the partner team reached out to me and said, would you be interested in becoming a Shopify partner? At this stage, it was just a side hustle. My plan was to actually run some ecommerce stores, not become an agency. John Readman: Right, got you. And are any of those still going? What was your favourite story of setting up your own stores? Keith Matthews: Yeah, the time, the first store I set up was back in, ⁓ was probably maybe 10, maybe 2009, 2010, was a drone store. So ⁓ did a deal with the UK ⁓ for DJI. Again, ⁓ was a long time ago, ⁓ was early days, but all of the transactions were fraudulent. So I had to shut that down, ⁓ is funny. I had a toy store. And of course that was going really well. then during COVID, DPD and British Post were throwing the packages over gates of houses and they kept breaking. So I had to shut that down because it's too much damage. I have a store at the moment, a print on demand store, kind of a half fashion store, which is run through Printful. And ⁓ just one point as an agency owner, I think you should always have your own store because... The amount of updates on the Shopify platform that are being rolled out right now, you know, like there's a few updates being pushed through the changelog every day. And there's also, you've also got Shopify additions, which is their kind of quarterly announcement where they announce, you know, future updates, past updates and major platform So I the best way to ⁓ actually the platform is to actually have a store. John Readman: Yeah, that's interesting. Yeah, my daughter's got a Shopify store and she sells scarves and bikinis online. But yeah, just sitting in the background, seeing and having access to a, and it's one thing working on a client's behalf on a Shopify store, but actually being in the weeds of doing it. So yeah, I think that's excellent advice. Excellent advice. ⁓ Keith Matthews: Yeah. We've always done it. Yeah. John Readman: Yeah, so and if you got any what are you sat on some amazing domains that are going to make you a billionaire? one day? Keith Matthews: Back the day, you did was when Ican owned, Network Solutions owned the original domain registry for .com. This of course prior to all of the other extensions, ⁓ the other TLDs being launched. I'll shorten this story, but I worked in construction ⁓ for a long, time, ⁓ time, ⁓ and I ended up working quite close to the Ryan family who are the originators of Ryanair, the airline. And that triggered when I got to know them and when I was in their properties or whatever, working in construction, mainly renovating them for a builder that actually lived locally. ⁓ would you believe it, ⁓ domains that I own the longest are actually connected to the airline industry. So they're Shopify related, but... The advancements in Claude and Lovable are helping me launch them. That's all I'll say. I won't say any more on that. I'll let you know. I'll let you know offline. John Readman: Yeah, let me know, let me know. So I went through a phase of buying domains as well. as you know, I've got this charity's bike ride and we're cycling from London to Sydney. I bought loads of cycling related domains years and years ago. And I had one called wearecycling.com. And I also had .com.co.uk or whatever. And I just thought, that might be, might do with some sort of. Keith Matthews: ⁓ John Readman: thing, community or podcast or something in the future. I got this approach, actually you probably get this, I'll tell you what we'd like to buy your domain or we'd like to look at this. And I was like, "are you serious?" and you just go with a much bigger thing. Basically after a bit of digging I've that international ⁓ government ⁓ had launched their whole brand of everybody cycling around We Are Cycling. They'd already built the brand, they'd already built branding agency but they hadn't checked if domain was about Keith Matthews: Wow. John Readman: Anyway, thanks for a couple of holidays. ⁓ Keith Matthews: the boys yeah yeah yeah but back just to go back to the just to finish what I was making back in the day of course you could register a domain name and ICANN or Network Solutions wouldn't they'd send you a paper invoice so so you could you could register a domain name have a think about it and then buy it so many's a many's a drunk a night I'd go home onto a laptop and start buying domain names you know and but at one point I had 500 but I had to just reduce it because they they're now John Readman: ⁓ yeah, yeah. Keith Matthews: They're now relatively, I mean, if you own 20 or 30 domain names, there's a cost there every year, so. John Readman: I know when I went through only six months ago and go I'm never going to use that. Why am I paying the... I'm just cleaning all of those up. There's no value in any of these. Just get rid of them. So yeah, think a lot of people in ecommerce have been through that sort of little piece. I suppose if you're... ⁓ we're keen on understanding now, if you're ⁓ ⁓ on Shopify who say they've to that reasonable scale, they're at a few million, maybe four or five million now, ⁓ what... Keith Matthews: Yeah. John Readman: Do you see us like the key apps infrastructure and the challenges when people are getting to that sort of scale? What's your sort of advice around what are the key other apps or other bits of infrastructure that they need to be thinking about to maybe go from five to 20 million? Keith Matthews: Yeah. So one of the things I noticed with brands that scale, like we're working with brands at the moment and some of them, we would have started working with them when they were doing a million a year and now they're doing, you know, 30, 40 million. One thing they don't do is invest in people, which is interesting. like ⁓ successful that I know, they make investments in people ⁓ and make investments in tech. But the investments in tech for Shopify are a lot less than any other platform. I mean, since day one, we have always put rewind.io.io into every single build. Because one weakness of Shopify is any stock, now they have updated the permissions on the store, but staff can still go in and delete data. rewind.io is probably the best insurance policy that you can have for your store. across like, but there's very like obviously as an agency we're using GitHub, we're storing everything in a secure repository. That's kind of bog standard right now. But like, I don't have that much to recommend to be quite honest with you John, because looks after pretty much everything. I mean, the SSL layer is on Shopify side. And what I would say is protect your data and protect your store by only authorizing the right people. John Readman: And for those listening who aren't familiar with rewind.io, can you just sort of explain what it is, how it works, how you pay for it, and just how a client would install Keith Matthews: Yeah. Yeah. Yeah, so again, Mike Potter is one of the OGs of the Shopify ecosystem and he has developed an app and starts at about, it starts, I think it could start free and then go up to maybe 10 bucks a month and you install it in the store and then every single day it backs up your inventory numbers, your products and all of basically your data, okay? So if somebody goes into the store and maybe maliciously deletes or delete any line of data, your sales history. Shopify doesn't back that up for you. so ⁓ now if using an agency to manage your store, well then the agency is obviously using GitHub and then the last version, version control is in place. ⁓ Hopefully, hopefully. one thing that you probably need to check before you talk to an agency. ⁓ So what does is you just install it, it's a subscription piece of software same as any piece of software. John Readman: Hopefully. ⁓ Keith Matthews: and it just works in the background and it protects your store. And we did have one incident during COVID. I can't quite remember the details of it, but we shut the store down, Shopify Plus store at the time, it probably doing five, six million a year. In the middle of all of the COVID panic, somebody in the team made a mistake. We just logged straight into rewind.io We asked them what time, they deleted. And let's say it was 12:01. We just rewound the store back to 12 o'clock. and everything was pushed back into the store. And if you talk to the team at rewind.io, or you talk to Mike, or listen to any of them, the amount of horror stories that they prevent is phenomenal. It's phenomenal. John Readman: I think there could just be human error as well, people in a rush, people under pressure, or doing stuff on their phone. It's, yeah, yeah. Keith Matthews: Yeah. Yeah. Now, if I was to give basic advice, which anybody I suppose can give you, I would stitch into every single build ⁓ because it works it's built for Shopify and Shopify is a major investor ⁓ in Klaviyo. And for a reason. They're very much embedded. ⁓ And then ⁓ the other thing suggest is don't launch a store without a reviews app. Now that's very basic advice, but in actual fact, Those three apps will help you go a long, way. John Readman: Yeah, and if, as you get bigger, do you need to refine things? Do you need to change things? What are the sort of scaling pains that people have? So, getting selling X hundred products a day, fine. Now we're selling X thousand products a day. What changes? Keith Matthews: Yeah. So in my experience it's the infrastructure surrounding Shopify that changes. a very very good piece of software. There's ⁓ very famous dot com called micksgarage.com which is an auto site which was probably one of the first ecommerce stores ever set up in Ireland ⁓ and up two brothers, the Crean brothers and they're exporting all over the world. And during COVID, they created a piece of software called Wave OMS. that is a very clever piece of software that makes picking products for your Shopify store in your warehouse 10x easier than just printing off the labels and walking around the warehouse. ⁓ ⁓ you're scaling, you obviously need a physical footprint, but ⁓ the systems around the Shopify store. ⁓ you know, ⁓ think, you know, ⁓ most... the most famous Shopify stores, and one of them is belonging to one of the Kardashians in the States. I think I heard a number that that store is probably doing over 150,000 checkouts per minute on Black Friday. So if you're a Shopify merchant, which is what Shopify refers to me as store owners, you really have nothing to fear in terms of scaling the actual platform. It's your physical resources to physically ship and physically store and the logistics that you have to really plan and work out. Shopify, you don't actually have to do anything. You know? ⁓ John Readman: And I think that's things people sort of think, ⁓ well, I'm just going to set this up and see what happens. But then what happens when it starts becoming successful? You might have to move out of garage. And I think now ⁓ ⁓ infrastructure. And there's a big ⁓ in brands is do you use third party or do you try and do it yourself? And at what point do you do that? And there's cost benefits to doing it all. Keith Matthews: ⁓ if look at, let's say the top two brands that we after, ⁓ one of ⁓ has their staff and warehouse and they like a personal touch. And the minute that delivery notification goes out, the customer, they're waiting. They trust the and they know that it's going to arrive on time. The other brand in a similar vertical, ⁓ has everything outsourced to a warehouse in the UK, a warehouse in Belfast, and a warehouse in Ireland. And they ship across the world. But that brand that has outsourced everything is also outsourcing a lot of the Shopify management to us. And they're outsourcing all of the aspects of the business. The other brand is scaling at the same pace, and they've taken a completely different decision to do everything in-house. So really, the answer to that question is dependent on your own team structure. your overall strategy to scale. I have seen a lot of brands outsource everything ⁓ then they get to a point where they're losing control ⁓ then they say, hold on a second, let's lease a warehouse, do a costing on it. And a lot of brands take it back in-house. John Readman: Yeah, and I think there'll be elements of personal preference, risk, and there probably isn't an exact right or wrong answer. It probably depends how big the product is, how much it's all or all that as well... And as we move forward, everybody's getting obsessed with AI. You can't go to an event now where somebody isn't talking about AI. And if you'd asked me sort of six, eight months ago, is AI good enough to generate decent creative that you're happy to put on your website and your socials and this, I'd have probably turned up my eyebrows a little bit and said, I'm not sure if it was my brand. Keith Matthews: Yes. John Readman: use it just yet. we play around with it a lot, but I think ⁓ before Christmas it got a lot better and got really interesting in my opinion. What experience have either Milk Bottle Labs or your clients have in terms of using AI in this whole process? ah around content production. Keith Matthews: I most of the brands that we would work with would be outsourcing that, but there is absolutely no doubt that agencies they're using have been very upfront with them and they have told them that they're using AI. I've used ⁓ personally and you, ⁓ I mean, AI studio at Nano Banana Pro is, I mean, if you prompt it correctly. there's absolutely no doubt about it that you can get the outcome that you need. of the things that I'm seeing is a lot of people are copying each other. ⁓ so ⁓ easiest thing in the world... ⁓ yeah. Like, mean, ⁓ amount of people that have copied Gymshark is just ⁓ you they really are, they're still in the field, I think, because they're authentic, ⁓ because they're copying anybody. They're creating ⁓ what they, know... John Readman: It's like an arms race, isn't it? Keith Matthews: the young CEO creates what he wants and what he believes in since the day he ever started. I don't think you can get away without using it. I think that it's improved, you know, it's improved an awful lot. And I think the answer to the question that you mentioned probably is a mixture of the two. Okay. Like if you're a small brand that's getting started and all you're doing is paying Facebook to market. Well then, you can probably get away with AI tools. if you're a big brand ⁓ you've been using, you've already been using, you know, physical for example, if you're a fashion store, ⁓ I think you can just suddenly switch it off and move over to AI. But AI can definitely assist. It's not going to wipe, it's not going to wipe that kind of marketing ⁓ side out completely. ⁓ I it's going to compliment it. But the truth is that I don't know the answer to that question right now, John, because that's going to probably change in a month and change in two months. John Readman: ⁓ It's changing all the time. got very interesting client who ⁓ shoes online. We did some testing around conversion of shoes just as a product shot, or shoes ⁓ with people's legs in in a outside. ⁓ And this sounds obvious to you and me, but when you've got a shoe on a leg in a situation, it will convert better than just a shoe on a white background. But their whole cost to do all the photo shoots is just prohibitive when you're selling shoes at an average order value of say 30, 40 euros. So it's an expensive exercise. So we were testing this hypothesis of, well, can we do this? We tried it last summer, not good enough. It just wasn't good enough. We tried it again just before Christmas and actually we put it in front of the merchandisers and said, well, what do you think? And they're like, yeah, that's good enough. And then I suppose the interesting thing is how do you then scale that and build that into your sort of process? you're onboarding a new product onto the website, you're getting the product shots. Well, right, so now that product shot, we're gonna take it from three different angles, then we're gonna put it into, via some sort of MCP, into Nano Banana with our content. platform and then we're just going to programmatically put all the prompts in and then suddenly we've got all this all this content. ⁓ Keith Matthews: Yeah. John, just to jump in on that, I use Lovable. I mean, ⁓ now I the Google AI Studio. Like you ⁓ could quickly create that system for yourself. for the, which is, ⁓ yeah. So like for the print on demand store, for example, I use Google AI Studio, but I also have a, a Lovable system ⁓ built. John Readman: I know exactly what they've done. Keith Matthews: which is using multiple different connections to Perplexity, Claude Chat GPT and Google AI studio. I have it pre-prompted so that if I upload an image, it will create five or six different versions of that image, all created with AI. And then on download, those images then are the quality of the images improved as it's downloaded to the computer. you know, like... I need for that particular store is an image because the printify app then will create that into a t-shirt or a logo or whatever. And this is low level stuff, but it's actually automation, ⁓ scale will like larger brands could also use. And I would suspect that that's exactly ⁓ the agencies are doing. They're building their own micro apps in house and ⁓ Yeah. ⁓ John Readman: Yeah, to speak of processes and systems and I think the big exciting thing, I'm excited about it, so we talk a lot to Google about, and when you talk to Google, the answer is always spend more money on Google, or when you talk to Meta, spend more money on Meta. Some of the large retailers listening to this will resonate with that. And I think they're always pushing YouTube, right? You need to advertise more on YouTube. I'm sort of thinking, well, is 2026 actually gonna be the year that more people can advertise on YouTube? ⁓ I one of the biggest costs for advertising on YouTube, yeah, the media is expensive, but if you can prove the media works and you hit your ROI or your ROAS, or whatever measurement you're doing. it's you can the actual big barrier to entry to video ads is creating the video right and i've been seeing so many examples of high quality video ads and going back to your point before, Keith, about the prompt. If you can prompt, say, VO3 in the Google Studio, Gemini Studio now, properly, even non-creatives, well, I'm a non-creative, I don't want to put you in the same bucket, Keith, but people who are not necessarily naturally creative can produce some half-decent stuff. And I think it's going to be interesting to see over time, is that going to unlock opportunities for brands to get more into that sort of upper funnel brand awareness? Because everybody's been fighting at the bottom PPC for years so I don't know if you've got examples of people using of video generation and watch your view on the growth of YouTube. ⁓ Keith Matthews: Yeah. Well you see, we're not, we're like, look, there is no doubt about it, but YouTube is the channel. It's actually, find it quite funny sometimes. It is the fastest growing channel out there. It is replacing people's televisions in houses. I don't watch terrestrial television anymore. I just subscribe to whoever I want to subscribe to and I watch everything on YouTube. And yet, John, it's the one channel. people don't focus on because what they do is they hold their phone up and take the video and then edit it probably with AI and all they have to do is turn the phone around into landscape mode and then they will have a two or three minute video for YouTube. YouTube is still a massive, massive opportunity ⁓ one of the channels that people, I don't know, is it an age thing that Instagram is just cooler? and the Facebook is out of preference for most people but look, if I was given any advice on this podcast I'd say to everybody produce content for YouTube. John Readman: Yeah, and I think the unlock at the moment, we would talk to another retailer that sells sort of fancy, high-end retailers, they're famous for these handbags. And they were like nervous about AI because it might ruin the product shop. And I get that, especially when you're selling a premium product. And I think a lot of the premium brands will be slower to test and evolve this. But this is quite a premium brand and they... decided to test as an example was right we're still going to do the photo shoot with the model with the product but instead of then next session booking the catwalk or the whatever to do the video shoot ⁓ not going to do that we're going to take more photos and I'm going to feed those photos into AI to create video and ⁓ results are phenomenal and again the cost and effort and so still doing the work and this but they're overall time of post-production and everything else and the cost and or they may have made decisions in the past where we're not going to do videos on those products because we're not sure it's worth the effort whereas now they can do videos on all their products if they've got the photos. Keith Matthews: That's good approach. One thing would say about the AI tools is ⁓ it does ⁓ very confusing to what you should use. ⁓ My advice anybody is just to stick with one. So if you look at Claude, Perplexity Google AI Studio, Lovable, Manus, which has just been bought by Facebook. Yeah, yeah, I use it when I come out first, but. John Readman: Everybody's getting excited about Manas at the moment. Keith Matthews: You know, when you've got five or six different subscriptions to five or six tools that are now all coming up to each other's level, I would say to anybody is just to choose one tool and master of them rather than getting stressed not knowing what all the others are doing or ⁓ prompting for perplexity. The advantage Loveable is ⁓ it's connected the whole lot of them. if... ⁓ John Readman: Yeah, I know. think also understanding how MCPs work and how you can connect it all together, I think is really important. My personal choice at the moment, but it will change. Probably next time I see you, I'll be onto something different. But I think right now for creative, it's Gemini, for everything else, I'm all in on Claude. Especially they've got their co-workspace and they've got all this. ⁓ It's ⁓ and pushing it is everybody, Google launched something new and everybody goes, oh we're going to do this. Keith Matthews: Yeah. John Readman: And this time last year, Google, think were way behind OpenAI and Claude I was actually talking at a conference yesterday on AI and the interesting thing is 75 to 80 % of people are using some form of LLM every day in their work. Most of those aren't on a company paid for subscription and they aren't necessarily, they don't have the guidelines, they haven't checked their privacy settings which is a real challenge and worry. The company doesn't necessarily have an AI policy at all. ⁓ the scary fact, Keith, and this goes back to your point about ⁓ are the companies that are scaling properly doing? Well, it's actually investing in their people. ⁓ were talking about getting good people investing in their people. ⁓ actually not many companies are investing in proper AI training. And think part of the challenge is, and I think you hit the nail on the head there, is ⁓ because don't know what to train them in. ⁓ Do go all in on Claude? Do we go all in on Gemini? Keith Matthews: Yeah. Yeah. ⁓ John Readman: we do let's pick one as a company and then let's try everybody properly but you've Johnny over here saying ⁓ well I use Perplexity for this thing or you've got someone else go well I'm going to try out Grok or DeepSeek or this or that Keith Matthews: So just to jump in terms of scalable brands then, John, ⁓ and there's a lot that, you know, we're talking, let's say, about brands that growing, brands that are ambitious, okay? ⁓ That's one thing I should have said well. Brands that are ambitious are generally the ones that are successful. Not the brands that are greedy, not the brands that are copying. The brands that are patient and ambitious are the ones that we see successful, and they're the same people that invest. ⁓ But there's an interesting mantra going around or a bit of a... commentary about Shopify and AI, and Shopify is all in on AI. mean, Shopify, you can build in certain, their standard themes, their free themes using AI, and we can see it drip feed into the platform. But one thing I would say is, ⁓ again, back to my point about ⁓ systems around Shopify are changing, but if I was doing five or 10 million on Shopify, I wouldn't be handing any of it over to AI. I'd still be investing in the humans. and training the humans to use the tools correctly. John Readman: I think that's it. I think and your thing about I think the other word we need to use for the successful leaders is curious, right? So they need to be curious about what's coming and be excited and understand well, how's this going to work across my business? So one ⁓ thing in AI and thing I think Shopify are on the front foot with ⁓ is this whole UCP relationship of like we're going to enable you to buy if you're on Shopify. Do we know when it's coming to Europe? Yeah, well we can just switch it on. Keith Matthews: No, no, have used it, we have developer access to it and we set up on one of our, actually two of our biggest stores, set up, we did a search and we successfully purchased it through shop. So we have developer access to it. And look, there's no doubt about it. It's very obvious what's going to happen is that you know, the chat GPTs of the world and cloud are gonna switch on advertising. and this platform that Shopify have helped build along with Microsoft and OpenAI, it will become the standard for many people. But the problem is there's a 4 % surcharge in every transaction that goes through AI. John Readman: came up in a conversation I had the other day ⁓ ⁓ was actually with Emma Jones who's advising business on agentic commerce and in the room were like the PayPal's of the world, meta, everybody, right? I don't know why I got invited but it was quite nice to in the room ⁓ and, ⁓ too Keith, but it was knowledge. ⁓ Keith Matthews: Mm. Because you're very good at what you do, John, that's probably why you're invited. ⁓ John Readman: around this and it's all out there you can go read it but even from some of the big brands they're like no well it's just like free i'm like no it's not free all right it's just like a super affiliate program it's super clever and i get it ⁓ for those who aren't familiar with UCP and what is going to happen and how's it going to work can you just give us a bit of an explanation from your point of view when you've tested it what involved from ⁓ putting on the merchant site to then what happened when I was in the LLM ⁓ Keith Matthews: Yeah, so I haven't seen it on the merchant side, I'll explain the brief history. Shopify got together with Microsoft and OpenAI and a couple of other tech companies and they basically created a kind of a world standard called Universal Commerce Protocol. And what it basically means is that if you search inside ChatGPT, primarily it's open to US businesses. search comes up, it's surfaced the very same way as it would be in Google. ⁓ Google will send you to the website or the Shopify store. if you click on, ⁓ you search for Gymshark, reference Gymshark again. So you search for a Gymshark hoodie. ⁓ When hoodie is surfaced in ⁓ ChatGPT OpenAI, instead of you clicking on the image or the link and going to the Shopify store, it bypasses it. and you go straight to a checkout inside Chat GPT you put your details in and you purchase the product and the transaction goes through Shopify's app, which is, I think there's about 160 million people now using that. So the customer never your store. So from a merchant point view, a store owner, that means then there is absolutely no opportunity to maybe upsell them something else. I Didn't see any or didn't see any details on how a user would subscribe or sign up to further updates that kind of thing so Basically, what they're doing is they're removing friction, right? Which is ironic because have actually made the purchase process using the shop Pay app probably the most frictionless purchasing experience for most people because now you can go onto a Shopify store and checkout with two clicks and What OpenAI have done is they've taken that away and they've created it now and they've basically removed the step to visit the merchant store, as I said, which means now your traffic is going to go down, your sales, more than likely when this kicks off, will go up, but the transaction fee is 4 % on top of Shopify fees. On top of more than likely, you'll have to now advertise on LLMs as well. So ⁓ a whole other channel. And to be quite honest with you, it's a tax on the Shopify merchant because it's going to be expensive. John Readman: Yeah, it's going to be interesting to see how it plays out because I think will probably take the view of we've got to be seen to be in there and at least learn and try. But they'll start looking at their customer acquisition costs and go, well, can we justify that? Or actually, can we just limit the inventory that we put in because we've got the margin versus actually we don't want to put that inventory in because we don't have the margin? And there was a part of a discussion I was involved with the other week and I sort of like the idea of this but this is maybe a bit fanciful. But will we end up with retailers in the future that don't have a physical store and don't have a physical website, they just have basically a connection to the LLMs and they just ship product. Keith Matthews: Yeah, that's probably there, it's probably, that's probably a long way away because, ⁓ don't how quickly it's just gonna take over. I mean, there's certain people of a certain demographic or a certain age that would never buy off OpenAI, ⁓ and it to be tried and tested. ⁓ like, none us can predict the future, but whether that's 12 months or five years, I don't know. John Readman: Yeah. Keith Matthews: But ultimately, what will more than likely happen is that you'll just have to have both. But they're probably, John, there probably will be brands. Back in the day, designed stores mobile first. mean, maybe there will be businesses there that will be LLM first. Maybe. ⁓ John Readman: I think it's gonna be interesting. And again, if you think about it, if you're trying to build your own brand, it's probably not for you. But if you're a reseller of other people's brands and someone's just trying to find the cheapest Nike Air Jordans in size nine, whatever version, flavour, in whatever colour, and you've got a warehouse full of them, well, do you need a website and do you need a physical shop? Well, if there's enough traffic and enough demand, and you're making it easier for the customer, who knows? I think it's gonna be an interesting space to watch. And if you're in that enterprise space now, understanding where UCP fits into your model is really key. And I suppose modeling out all of this, and I suppose also then coming on to sort of measuring it, because you'll have no you'll have no click, you'll just have a sale. So that won't appear in your GA4. ⁓ Keith Matthews: 100%. Yeah. Yeah. Yeah. John Readman: and it will appear in your Shopify analytics, but then what we've seen at the sort of enterprise level, most people are extracting stuff into some form of different type of analytics platform or some sort of different app so they can join it all together. going back to your point about advertising, so OpenAI, so we had a client going, wanna advertise on OpenAI, I've seen they're letting people advertise on OpenAI, I wanna do it. So we, the agency guys contacted OpenAI, 250,000 pounds a month minimum spent. Keith Matthews: ⁓ stop, serious? Yeah? Yeah. John Readman: Yeah, so that's going to be interesting. So that's going to be, I don't know, your Unilever's, your Coca-Cola's But I would have potentially looked at it the other way. And I don't know that's just, they haven't got the tech set up, but from what I understand, it's going to go, I think they, so well, the work that the ad platform they're using, maybe they're worried it won't be able to scale in the right way. Keith Matthews: yeah. John Readman: And then there was the other one in the Super Bowl final where Claude took that massive added identity seriously. We're never going to advertise. Well, how sustainable that is. And ⁓ sort of saying Claude are going more after the enterprise work ⁓ and taking on like joining up and being the glue that pulls together ⁓ office all your other stuff and sitting in the middle and just being this thing that makes you be more productive. Keith Matthews: Yeah. ⁓ know, you'd wonder if you look at what Shopify have done over the last 10 years, you know, eating Magento's lunch and eating ⁓ lunch, you know, and it is now becoming, and obviously it you know, commerce at one stage was probably their potential threat up a few years ago, it's not now. You'd wonder really, if you look at all the LLMs, is there, would it come down to one or two or is there room for ⁓ all? All of them seem to be doing the same thing and announcing the same thing and won't after each other. John Readman: Well, interesting one, there's a, ⁓ what was it called? the music app, have you seen that? ⁓ then Google last week launched a music app, right, inside Gemini. So who's gonna pay a separate subscription? All right, so it is gonna be interesting sort of how it ⁓ works. Keith Matthews: Yeah, yeah. Yeah? Yeah? Yeah? Yeah, and that narrative at the moment around this, I don't know if you've been following it, the narrative at the moment in the last four weeks has been the death of SaaS John Readman: Yeah, think, yeah, another thing, I think it's more about what is the function of your SaaS? And is it doing, what problem is it solving? Because there's this argument, everybody's just gonna build everything themselves. how secure is that? How scalable is that? What infrastructure is that on? What happens if the person who built it leaves? Who's gonna update it? So I think depending on the function of the content, of what the SaaS is doing, I think the answer with all of that depends. Keith Matthews: ⁓ and ⁓ know, you made an interesting point there about, you know, doing it yourself. Like, it's people like me and you that would consider doing it yourself, but that's probably only a few percent of population. You know, I see it the other way. I see it that it's a massive opportunity for SaaS that doesn't even exist right now for without funding to actually deliver it. And I think there's a massive opportunity out there, ⁓ although... John Readman: I know, ATC, are we going to see a two-person unicorn? Keith Matthews: Well, yeah, that's again, you're watching YouTube like me, you know, it will happen. It will happen. I don't know whether it's going to happen today or tomorrow, now I don't know how one or two people could do it, but certainly 10. John Readman: Yeah. So you mentioned a couple of apps like Klaviyo ⁓ but are other killer secret apps that you would recommend to larger, is there anything else out there that you'd recommend? Keith Matthews: You know what John, if you asked me that question three years ago I probably would have come up with a list but I mean there's probably 20,000 apps in the app store at the moment on Shopify and they're publishing reviews, of them are giving you better analytics, better reporting. I mean there's no secrets out there now because everybody's all in on it ⁓ to be honest with you as an agency we pride ourselves in not adding apps. We'll customize the store to be as beautiful as possible. We're not gonna put an app into, for example, to add a section or change anything on the store. We'll code it, push it live and reduce the amount of redundant code on the store. So I think the average Shopify store has, at one point, 16 apps. So like, of those 16, there's email, there's reviews. There's the backup app rewind.io that I mentioned. There's actually no secrets out there. know, everybody's all over it. I wish I could answer better, but I don't have a secret. John Readman: Yeah, it's just interesting. And yeah, it's like the Shopify app, yeah, there's tens of tens of thousands. What I think has been interesting in the last six months is this shift towards built for Shopify. And they're trying to sift out, right, these are the ones that are really enterprise level. I think that's quite an interesting ⁓ around. ⁓ think Shopify has also got a problem like dormant apps, absolutely don't work, apps have got bad. Keith Matthews: Yes. John Readman: reviews They've created this sort of cleaning up process that needs to go on. And then who are the real big brands that when you're building stuff for your clients that you sort of look up to? You mentioned Gymshark before. If you had to name two or three other brands and you can give your clients a plug if they're the ones you're looking up to. But who are the three or four inspirational brands in Shopify? Keith Matthews: Yeah. And... Yeah, like... love, ⁓ you love Gymshark. To me, it represents you know, ⁓ in the days, I don't know you remember this or not, but the promotional videos on YouTube for ⁓ Shopify ⁓ were based on, you know, originators, ⁓ people with original ideas, crafters, handmade product. And when Toby started Shopify, he genuinely did want to open it up small to medium businesses and he wanted them to be able to sell online. And ⁓ Gymshark story ⁓ me really does represent everything about Shopify that ⁓ even remember in the old days. I'm into makeup. But you have to look at the Kardashian family and their success. And there's Irish brand and they're on Shopify. They have nothing to do with Milk Bottle. I won't take any credit. There's a wonderful young woman called Aimee Connolly runs a business called Sculpted by Aimee ⁓ And she built a wonderful, a beautiful brand, scaled it on Shopify and she has competed in retail again against some of the biggest makeup brands. And when you're living in Ireland, the majority of the largest makeup brands are actually UK based. So that'll be sculpted by Aimee And then we work with a wonderful brand called Ella and Jo, which is a skincare brand. And they've done the very, very same. They've originated in the West of Ireland and they're shipping all over the world. So there's a couple of brands that I genuinely look up to. If you go onto those sites, it doesn't necessarily mean that you have to like their stores, but their stores are successful. and they're selling their own product. I have a little bit more respect for people to create it themselves, basically. John Readman: Yeah, and it's like, what can you learn from that? And so there's like so many different nuances. had a podcast recently with the guys from small who the more green like home detergents and dishwasher tablets and and just how they're at like. customer flows and process, customer acquisition and everything. It's just fascinating getting really into the detail of everything. then, if I'm a big brand and I'm thinking I need to ⁓ my agency and my Shopify support, why would I consider Milk Bottle Labs and what are you amazing at? Keith Matthews: Yeah, so the one thing I think, and I'll answer this for general agencies, like when I look at agencies, I look at successful agencies, the first agency that I ever saw that I really wanted to be like was an agency in, think they're in Ottawa, they were one of the first Shopify agencies called ⁓ Pointer Creative, and that was run by a guy called Chris Pointer, and I met him, I couldn't believe it, I met him at one of the Shopify events in Toronto a few years back, and basically, ⁓ You just need to trust people. know, ⁓ would you trust me with your bank account details? Is question that I would ask anybody. Would you trust Chris Pointer? Would you Alex and the guys ⁓ that, set you know, We websites, obviously one of the most famous agencies in London. And all of these people have something common and they're trusted. Because when you... begin to engage with the clients. If a client is doing $2,000 a month because it's a side hustle and it's scaling, or a client is doing five million a year or six million a year, at the end of the day, you have to respect that client to the same level. As an agency owner, we don't measure people based on their per month. We measure them based on their personality and their ambition and are they a good fit. The one thing that we always get back is trust because when you do start to scale and when you do start to more products, the agency that you're working with ⁓ is ⁓ embedded your business and people lean on us and depend on us. So over the years, we've evolved into Shopify only. So only Shopify, only Klaviyo. Anything to do with Google Analytics, anything to do with paid media. I would advise everybody to have a separate agency to do that. Google's ecosystem is changing so quickly and the Shopify ecosystem is changing so quickly. You need experts in each lane. That's just my opinion. Rather than giving somebody one check to do absolutely everything. That's just, you know, that's my view. ⁓ It's been a on Keith. John Readman: and then. Yeah. Okay, good. Well, it's been a pleasure having you on, Keith, ⁓ excellent insights ⁓ there. And ⁓ you'll be working with Gymshark and doing all that for them, Keith Matthews: No, I can't remember the agency that was working with them. I've off guard, but he was a great guy in the UK. And he built Victoria Beckham's site, actually. If you want to have a look it's a beauty, have a look at that Shopify store. I think of his name. His name escapes me. but a ⁓ lot of people would know exactly what I'm talking about. John. John Readman: if want to get in touch with you, where can we find Milk Bottle Labs? Where can we find you? Where would we go? ⁓ Keith Matthews: Yeah? Yeah? Very simple, milkbottle.ie or google Milk Bottle Labs. yeah, we would love to work with ambitious brands. Absolutely. John Readman: Excellent, well thanks for coming on and we'll catch up soon. Keith Matthews: Thanks, John. Thanks, the chat.