Good afternoon and welcome to the Shopify session at Goldman Sachs Technology and Communacopia. I'm Gabriela Borges. When I was debating with Shane from Shopify how we could amp this presentation up a little bit, we thought it would be really fun to have Archie Abrams, Head of Growth, join us from Shopify, and of course, Jeff Hoffmeister, CFO. We really thought this would set us up for a little bit of leveling up in the Shopify discussion beyond all of the good stuff,
All the numbers
that Jeff typically shares with us. Look, it's such a great time for it because the investor questions that we've been getting have gotten much more technical and specific on Shopify's moats over the last six months. I wanted to start this conversation with a little bit of a reflection on the culture that Tobi has built at Shopify.
Yep.
There was a podcast that came out, I think it was a couple of weeks ago now, where Sam Altman, he opened up talking about Tobi and how he's just such a great example of a Founder CEO who actually rolls up his sleeves and gets into the technology.
Yep.
I know from speaking to you guys for a number of years now, that the first principle of thinking at Shopify is pretty unique. Tell us a little bit, maybe via an example from your day-to-day, where this plays out in practice, and Shopify essentially thinks first principles about adopting AI.
Yep.
Yeah.
Archie, you want to start?
Absolutely, yeah. Thanks, Gabriela, for having us, and excited to be here. Working with Tobi is amazing because he is so hands-on-keyboard, right? He is in the technology himself, doing it. Just the other night, we were going back and forth very late into the evening on how different AI models are discovering Shopify merchants and Shopify itself. Right? Really trying to reverse engineer a lot of that discovery and what makes a merchant appear, what makes Shopify itself appear, and using it as an opportunity to really rethink the information we are giving to merchants, and how we are even structuring ourselves. So that depth and that level of understanding I think is really unique in how we build and what we are doing.
Jeff, an example from your day-to-day?
Well, I think one of the things, and Archie alluded to it, the one point I would expand on is because Tobi is so deep in the technology, he is exceptional at making sure we constantly are questioning how we have built a product. And maybe it was built on an architecture that even though it is only two or three years old, could be rebuilt in a way. We see this a lot where he is like, "Hey, this is phenomenal, but if we want to take this to the next level, we need to do this.
Let's use this latest and greatest technology in order to enhance the product so that makes sure over the next two or three years, it is even further along." Or the other thing which I see a lot is we will be doing product reviews, and a bunch of engineers will come in and say, "Hey, we did a bunch of work. We got it down to 10 milliseconds," or, "We did this," or, "We did that." And he will be like, "Well, yeah, it needs to be two." And they are like, "No, no. We just spent two months trying to get this down to 10 milliseconds" He is like, "No, no. It has got to be two." The consumer experience demands this.
He is very good about taking that product knowledge, thinking about how we apply it, and make sure we are delivering merchants what they need, not what it is that we are kind of obsessed about building, but what they are actually going to utilize.
Jeff, there was a time, I think it was around COVID, actually, when Shopify still used to give us merchant acquisition numbers every quarter. And I remember investors used to ask us, "Look, at some point, this merchant acquisition gravy train is going to slow down because the depth of TAM just is not there." Archie, I know your role focuses a lot on merchant acquisition strategy.
Talk to us a little bit about what the strategy is. Shopify has sustained these growth rates. It does not seem like merchant acquisition is slowing down. How are you able to generate this sort of infinite TAM?
Yeah, it is a great question. I think that the key to our strategy has remained timeless over time, which is that we believe that the next large Shopify merchant is just getting started today. Our job is always to meet entrepreneurs early in their journey, wherever they are, and lower the barriers for as many people to start. We know that many of those businesses, they might not actually succeed, and that is totally fine because the businesses that do become brands like FIGS, Alo Yoga, and the like, and of course, we grow right alongside them. One thing that is really interesting about the merchant acquisition piece is one of our goals is always to build trust with entrepreneurs.
As I mentioned, a lot of those businesses, they might join, they might not be successful on their first attempt, but actually, over 40% of businesses on Shopify are started by merchants, with their second or third or fourth business, coming back to start over. If you look at one example of that, Ben Francis, the Founder of Gymshark, actually started multiple businesses before he hit it big with the apparel brand, Gymshark. That goes to us really lowering that barrier to make it really easy to start in our integrations with all of the agentic platforms so we are there, building trust along the way, and then helping more merchants come on, be successful over and over until they have hit that successful business.
There was a couple of secular shifts that happened probably around the time that Shopify was founded, where the startup cost to start and become a merchant went down. I am thinking of dynamics like dropshipping, I am thinking of Facebook ads. What does that look like in 2026 with AI? Does the barrier to the average American starting a business, or actually global, has it gone down, and in what ways?
Absolutely. I think that one thing I look at is right now, if you look at, we will focus a little bit on the U.S., but obviously our growth is very global. The 12% increase year-over-year in terms of new business starts in the U.S. But if you look at absolute numbers, it is still only 1% of the U.S. population that is starting a business. I think our fundamental belief is that there are many, many more people who want to reach for independence, particularly with AI. What AI allows one to do is, if you think about starting a Shopify business, you had to be good at a multitude of skills. Be really good at designing apparel, but then also be really good at marketing. Also be good at website design. Also be good at running your operations. Also be good at analytics.
With AI, you might only need one superpower. Be an incredible product designer, but then leverage the Sidekick or other AI tools to become an expert in marketing. Leverage Shopify's Campaign Autopilot so we can help you run your marketing for you. Use Sidekick to run your reports and run your operations. All of a sudden, the barriers for many people who want to reach for independence, they really only need one superpower. The Sidekick and other AI tools can help them with those other skills that are not as, come as naturally, I think that is really the opportunity with AI and what we are seeing.
Yeah. I think if some of you have not, as investors, just tried to build a store and use Sidekick, I think it would be really enlightening just for you to get a sense for the experience, the information, the content, all the information in there to just basically say, "I have an idea," or even, "Help me generate an idea, and with this idea, get a website up and running." Even if it is something where you are just testing it out and trying to figure out, "I want to just get a feel for how the technology works and really put it in action," I would encourage you to do it, because to Archie's point, it is pretty incredible what it enables.
Let me ask you a little bit about how agentic commerce is starting to impact the market. Let us start a little bit more macro. During COVID, people would track the e-commerce trend line, and it accelerated. I think it was three years' worth of acceleration in nine months. My question to you, do you think the e-commerce penetration trend line is changing again?
I'll start. I'm sure Archie will have plenty to add. Yes, but I also think it's early days. I think with all the product announcements, it's very easy to get focused on, "Hey, we think agentic commerce is such a huge percentage of commerce." We're really only, depending on how each of a company, a vendor, or software provider defines it. We're six to nine months into seeing real traffic where someone goes to one of the LLMs, does some discovery, and then ends up on a merchant's website and then transacts. So with that caveat that we're still in the early days of this, we are seeing an inflection. As you know, when you normalize across geographies, roughly 20% of commerce globally is done online. I should know the exact rate of change.
But in any given year, it feels like, and Gabriela, you have a view on this, too, in the last few years, it's a point or so of global e-commerce each year shifts from offline to online. We fully expect agentic commerce to inflect that. That obviously will play into our strengths, into what historically has been our traditional source of where we started with merchants and helped them grow. I think the other thing that we're going to see, just into the adjustment of the inflection of the growth rate, is we're going to see more and more discovery happen earlier in the process, which is going to help merchants. There's a couple of stats, one of which we alluded to in the last earnings call.
But when you start with an LLM and you do your searching there, you're 2.5x more likely to end up right on the product page. Rather than starting somewhere in a general spot on the merchant's website, you're going straight to the product you want, and then that leads to an uplift in conversion. We've seen roughly an 80% uplift in conversion. You're doing a lot more discovery, searching, figuring out exactly what you want before you even get to the merchant's website. But it also helps then the merchant transact and convert much faster, which is something that we've seen to be very effective.
We're also, just in terms of how we're interacting with the merchants and all the LLMs, for us, we're the only one that has the admin that allows you to work with all the different AI and LLM platforms directly from one admin, one pane of glass, and lets you decide, "Hey, I want to sell on this channel or that channel, or turn this one on, turn this one off," at any one given point in time. So it works really well. But importantly, the economics to us do not change. So whether you start on this LLM or that LLM, you're still a channel for the merchant, and you're still getting to the merchant's website.
Once you are on the merchant websites, you are still going through Shop Pay, you are still going through the process, everything you are doing in terms of obviously exchanging money and getting your goods, services, whatever, sent to you. There has been a lot of announcements around Meta and Muse, which obviously is a huge partner of ours. Exact same way in terms of everything we have done, in terms of how we work with OpenAI and everyone else.
Your point on economics is an interesting one, because I would imagine that whether it is Meta or perhaps one of the frontier models or the LLM engines, they would also want a share in the economics. How does that work? How are you able to preserve your economics and then also make it worthwhile for them to be in this path of discovery?
Well, for the LLMs, they are trying to do a couple things. The business models, the size of the businesses differ a lot, right? It is a little hard to make some generalizations across all of them. But for the most part, they are engaging with the consumer in a way largely attached to what they are trying to do on an advertising side, to funnel that consumer to a merchant and help them conduct a transaction. But if you are a merchant, the vast majority of the time, you still want the transaction to happen on your website. It may even be within one of the LLMs, it makes a reflection or a mirror, or it seems like you are actually still within the LLM, but you are really operating with the infrastructure of the merchant in a lot of cases.
Because if you are a merchant, you spend countless hours, dollars, management bandwidth trying to make sure the pictures are the way you want, the brand is the way you want, the product descriptions are what you want. This is the top level, and you click down to the second level, and you click down to the third level. All that information, a merchant has spent a ton of time working on. The LLMs are doing everything they can to engage with the consumers. One of the things is, you know, Gabriela, as we have talked about it. One of the changes that has happened is most of the LLM searches are generally when you type in the query, it is generally two dozen words or so. The old traditional search, you probably were typing in five or eight words.
To all the LLM's credit, they are generating more sophisticated questions, which allows them to give more sophisticated answers, and they are trying to compete. In addition to what they may be doing on advertising and other pieces, they are trying to compete on the quality of the referral. If you type in something very specific in terms of a product you may want, they want to get you to the right spot. I am sure we will talk about Shopify Catalog at some point, but that is the beauty of Shopify Catalog in terms of how we interface with the LLMs. They are trying to build their brand with the consumer.
We get this question, and I think you have already alluded to the answer here, but just to crystallize it. Every time a new agentic technology comes out that seems to have some way of replicating or disintermediating Shopify in the stack of merchant discovery through to conversion, we will get questions about it.
Yep.
I think that is probably a way to answer all of those questions in the same way, which gets to the heart of Shopify's moat. How do you, when investors ask you about ABC product that seems like it might abstract part of Shopify, how do you respond to that in a more comprehensive way?
Yeah. Actually, maybe Archie, I am going to let you start on this too. Part of it is just, to your point, the comprehensive element of things we do, which this is one of your near and dear points. I will let you start, and then I will pile on.
To Bill, I think the complexity and the fragmentation is amazing for Shopify and the value proposition we offer to merchants. Because if you are a merchant right now, think about if you are starting a business, you need to have a website. You need to sell on the agentic channels. You may also want to do physical retail. You want one place that can bring all of that together in a simple way, and that is what Shopify does. From a merchant's perspective, actually, as the fragmentation gets even more over time, all roads lead back to Shopify. You want one place to manage your inventory. You want one checkout. You want all of those pieces to flow back. That, I think, is the ultimate moat for what we do in the merchant value.
Always keep in mind that Shopify is a place that does, depending on how you define the products, 12, 18, 20 things for the merchant. There are companies angles that will do a piece of that. But to Archie's comment, we are helping them with the inventory. We are getting them connected to shipping. We are calculating the taxes. We are doing the payments. We are doing the consumer context in terms of everything in their wall. We are running their website. We are using Sidekick to help them figure out which new product to do next, or how to price, or what to do in terms of this initiative or that. All that stuff.
At times people are like, "Well, I want to do a little bit more of this piece or a little bit more of that piece," but no one is doing anything like we are in terms of the totality of all of it.
Let's talk about Catalog. It is one of those innovations that is so simple and obvious and yet so sophisticated because the Catalog, as I understand it, doesn't just describe, "Hey, it's a dark green dress." There is a level of how you can code that up that makes it incredibly useful to LLM discovery. Talk to us about how Catalog does that.
Yeah. So I think one of the things that we've invested a lot in is our real deep understanding of every single product that the Shopify merchant has. Creating a taxonomy that is incredibly robust around not only its the most basic facets, but also things like understanding the imagery and really understanding the tone, what type of buyer this is going to appeal to. Then in the process of that, cleaning things up so that if there are multiple products listed from different merchants, we have that unique understanding of a per product understanding versus a scattering of these products are actually the same, but they appear in different listings.
All of that, plus the live inventory data, the live pricing data, those things are extremely unique when you're an LLM who's reaching for and helping a consumer actually find the right product versus using web search or something like that, which will have all sorts of issues, not real-time, bad pricing data, no live inventory. All of those things are really compelling for the Catalog for both consumers, and then also for merchants themselves.
The statistic that I really like from your earnings call was the one about conversion rates for the long tail.
Yes.
Yes.
How do you think competition is changing in, so much of Shopify's value proposition has been empowering the rebels and-
Yep
helping the small guys compete with the power of an Amazon. How do you think share is shifting between the long tail of merchants versus the giants in the room, typically the marketplaces?
Yeah. I'll go back a little bit to one of the comments I made in terms of the LLMs being really thoughtful in terms of you're asking more sophisticated, nuanced questions, and you're getting more sophisticated, nuanced answers back. We're seeing that initially play out in a couple of spots, and we alluded this, to your point, Gabriela, a little bit on the earnings call. There's two immediate spots, and there's a third one, which is rapidly developing. The two immediate spots are some of the, I don't know if I'd want to say niche-y products. We had Harley quote a statistic regarding what percentage of the AI-driven traffic was outside of our top 100 categories. Obviously, we can define the top 100 categories a bunch of different ways.
The point is that the significant majority of the AI-driven, the AI-started searches, I should say LLM-started searches, are in some of these products which are maybe less known. It's not the huge brands. It is maybe, for example, in some product, some entrepreneur that started five or six years ago, the brand took off really well, had some phenomenal growth rates, and then just plateaued a little bit. Had kind of found an initial niche, and then with LLM, you're getting more specific questions, and you're kind of seeing a lot of these brands that had done well, slowed a little bit, and a lot of them are taking off again, which has been amazing to see.
The other category, the second category where we've seen a lot of initial impact, from LLM, are things which are very heavy in terms of product specifications, electronics, or maybe there's a piece of furniture. It needs to be a certain size or certain height or something that needs to be a certain weight or something of that where obviously the LLM is going to allow you to put in some very specific requirements that you want and then get more feedback from that and get a better result.
The third thing which we're really doing, we've talked about this a little bit when Archie was talking about Catalog, we're really trying to take some of the things that maybe did not have a lot of product-heavy searches, now because of LLM, you can take some of the things that merchants have in the inventory listing and with Catalog, present them in a way which is enhanced in terms of the product description. Maybe you're looking at, all right, so I want to look and buy this suit, is this suit a certain level of weight of fabric? Is it breathable? Is it appropriate for this type of event? Is it suitable for this?
Adding some things which historically you never would have considered within a description or a characterization of a product, now put that in there in a way where the LLMs now for the first time allow you, if you can ask the question, which of course you can, as you get in the habit of asking that question, Catalog will serve up a product which is even better suited to what you maybe would have planned.
Maybe historically you would've bought from brand X, but now the LLM will say, "Well, based on other things you bought and your style and types of things that I see you buying, actually, maybe there's a small manufacturer in Portugal or wherever, that that's the manufacturer you should be buying from." As we can help merchants do better descriptions of their products, fold that into the Catalog, it's really helping a lot of merchants be discovered in that way. It's feeding more information and content into the system and letting that content then be accessed by the consumer.
The inverse of this question is how merchants discover Shopify, which Archie, I think is at the heart of where you spend time. Jeff, I remember there was a fair amount of change at Shopify when Google Search Engine went down as a percentage of your mixed top of funnel in 2023 timeframe. The number of marketing channels that you all were able to use to land new merchants dramatically expanded as best we could tell from the outside. So Archie, tell us a little bit about how that strategy has changed. Maybe bring us up to speed over the last year.
Yeah. Great. I think the great thing about our merchant acquisition has always been a very diversified set of channels that we've used to acquire merchants. As I mentioned earlier, a lot of it has always gone back to how can we be helpful to merchants at the very top of their journey to becoming an entrepreneur. Historically, that may have been tools that we had, like a logo maker or a business name generator. What we've done is we've taken a lot of those tools and actually made skills that are accessible by agents and connectors that enable folks in Lovable, in ChatGPT, in Claude, in Manus, in Replit, in Grok, to basically be able to discover Shopify right there.
As well as really, going back to what Tobi and I were going back and forth about, at that late night, how do we understand what type of content is being served up, and how do we continue to optimize our content and how people are talking about Shopify to make sure that we are showing up right at the start when somebody is searching for the best platform to start their business. What we see is we see just an enormous amount of visibility in the AI channels. A lot of that comes back in direct to Shopify or via some other channel as well. That's been kind of the core. It's getting into the agents, being helpful at the start of the journey, and playing that out.
I think it was Harley who was talking about how the initial meetings with larger enterprises also feel different because you can very quickly show them essentially what they're missing out on. Talk about that conversation further up to the larger enterprises, the Estée Lauders of the world.
Yeah. I think a big piece of that is folks realizing some of the network power that we have, and specifically around payments and Shop Pay, that accelerated checkout really opens up a lot of conversations. More than that, I think I'd go back to a product that Estée Lauder or any of the large brands have, is that the folks are always, let's say, running A/B tests to understand how they can convert or better price their products, et cetera. Not only do we build products that make that easier to run those types of experiments, but we also leverage a lot of the network intelligence that we have. So there's a product called SimGym. What does SimGym do? Allows you to create a bunch of variants.
Could be landing pages, pricing for your product imagery, ads. Then we actually have agents come in and basically simulate which one of those variants is more likely to do well. Then actually run the experiment. Then actually launch the winning variant. So when you're an Estée Lauder or you're a large brand, you're getting both the benefits of all the complexity, but also that we handle, but also that intelligence that we're able to offer that is very unique compared to anyone else.
One of the themes that we've been exploring is this idea that switching cost goes down. I think as it's relevant to your business, it's the switching cost between some of the legacy platforms and migrating over to Shopify. Have you all observed that? Maybe give us a stat here. Perhaps the migration typically was a one-year undertaking. Have you noticed a change in pace of migrations?
Yeah, I think we're seeing that with our help, some of the largest brands are able to get launched in weeks or even shorter because of what we're able to do, both by just having the platform the way it is, and it's easy to integrate, but then also whether that's agencies that we work with and really equipping them and the integrators to be able to leverage all of the agentic workflows to be able to help move those brands over. We've seen a real ability to get folks launched increase pretty substantially over the last year or so.
Yeah. I'd say the other thing, too, just in terms of migration overall, Archie, you mentioned SimGym, but I think just in terms of having agentic roadmap, a thoughtful roadmap of, here's Sidekick, here's Catalog. We started building both of those three years ago.
Here is products that are going to help you as you really think, even if you, as a CTO of a large organization, say, "I have maybe a slightly different view or bearish view in terms of how quickly agentic commerce is going to take off or what percentage of this or that it is going to be." It is still hard for you as a CTO of a very large global brand retailer to say, "I am going to adopt the platform that does not really have a lot of AI capabilities." We are really the only ones that are presenting not just an agentic plan, but a bunch of products in place with a bunch more coming just in terms of how we think about these capabilities and really making the decision where it is hard for them not to do.
I think effectively what we have done over the last couple of years is, for the first time, given large brands the opportunity where you have a third-party software which is going to outperform, which is going to help you be more successful than something that maybe you spent, depending on how big you are and how long you have been around, years, maybe a couple of decades, in terms of building the infrastructure, building the team, and customizing all that stuff. All that is wonderful, but with the capabilities of Catalog and the capabilities of everything we are doing, you are just at a spot where you are almost at a competitive disadvantage if you are not adopting some of the things that we can do to help you propel your business.
Let me sneak in a couple of macro questions. Archie, I think this one is for you. How worried should we be, or how worried should Shopify merchants be about SEO hitting an air pocket?
I think that for our merchants, what we see is traditional search continues to grow and it continues to be strong. But the nature of that SEO, it tends to be different. It tends to be lower in the funnel, more navigational. Things like in the past, you might have had queries like black pants, and now you have Vuori pants instead. The top-of-funnel discovery is happening on social platforms, obviously still, as well as in agentic, where folks are actually researching the best pants to buy, as Jeff said, really able to get that customized thing. But then when they actually want to go discover the brands, they typically will use those navigational queries to actually have that rich experience of being on the merchant's website.
All of this, I think, will help independent brands because those queries around black pants, those are very hard and very competitive queries. Whereas in agentic, those independent brands are more discovered, on social is where with our merchants and their ability to create amazing content and the like are able to really succeed. I look at the change in SEO as very beneficial to a lot of our merchants as the new discovery channels grow.
Jeff, the other macro question for you is health of the consumer.
Yep.
I use your data point all the time on the average Shopify merchant makes on average more than $100,000. With that backdrop, any comments you are willing to share on health of the consumer?
Yeah, we try and put a lot of reliance on our own proprietary data, but I would also say that we are not seeing anything that vastly differs from what you see in the morning, whether you are reading this blog or watching CNBC or wherever you are getting from the Goldman trading desk in terms of you are getting views on the economy in terms of how things are doing. You see it in the numbers we have posted for several quarters right now. The business continues to perform very well, and I am always talking about strength across geographies and merchant sizes and industries and verticals and all that. It remains strong in our minds. Especially the last couple of days, of course, it seems like we ebb and flow in terms of what is going on with oil prices and consumer spend and all that.
Gabrielle, I got to stick to what I said at the last earnings call. That being said, it remains consistent in terms of the strength that we're seeing. You all will have your own views on consumer spend. It seems strong to us right now.
I think it's worth spending a little bit more time on Sidekick because of the amount of proprietary data that Shopify is able to leverage in Sidekick. Tell us a little bit more about the AI engine that's running behind that. How much is third party? How much is Shopify proprietary? What are the types of data that you can really put to work in something like Sidekick?
Yeah. I think that it goes back to one of the investments Tobi made right at the age of the start of AI a few years ago, which is really building a venting in the admin that was incredibly powerful. So tracking absolutely every interaction that a merchant has and a buyer has on merchants' sites and other services, which really laid the foundation for us to be able to do all sorts of next best action predictions.
I think where you see Sidekick going is leveraging that deep understanding of an entrepreneurial journey all the way from starting to scaling and making Sidekick not only good at when a merchant is doing a task like designing a website and doing analytics, but starting to become more proactive with things like Sidekick Pulse, which is your proactive recommendations of, "Hey, you might be not understanding that you could create a collection and offer a discount, and that will increase your GMV." So it all starts with really that very early investment and that very prescient investment Tobi made and we made in really understanding all of the actions.
When we have that, we understand that both the merchants and how buyers are interacting across all of our storefronts, agentic channels, in person, it allows for something very, very powerful, not only as a tool, but also as a proactive assistant to help merchants run their business, and is very unique.
Jeff, this reminds me of when I used to ask you about audiences monetization every other quarter. There are these products within Shopify that are monetized indirectly. I think Audiences is a great example, Sidekick to some effect.
Tell us more about the monetization strategy and the second derivative impact that these types of products have on GMV integrate.
Well, I'll start with Sidekick, the second half of your question. For us, we've given some statistics, we will continue to give more in terms of the impact that we see Sidekick as having on merchants and their success. Really what we've seen over the last, as I mentioned before, we started working on this over three years ago. It's been out for over two years. It's really about nine months or so ago where we really started to see diversity and explosion in the usage of Sidekick. It went from being something which we were seeing largely in terms of midsize merchants were asking, for the most part, go-to-market questions, was really the initial foothold that Sidekick has, it's really exploded in terms of merchants of all sizes and all geographies are asking all types of questions.
You can imagine that the questions being posed to Sidekick by a 10 person entrepreneur team versus one of the Lands' End or some of the biggest companies out there in terms of how they're thinking about things. They're different types of questions. What we're finding is that it's really helping people think about how they can accelerate their business. Obviously what that translates, the merchant is, the tail of the goose is selling more things. Obviously that's generally going to translate back into payments revenue or FX revenue or tax revenue and all these things.
Sidekick is there because it goes back, and you mentioned this before, in terms of the first principles of Shopify and how we're very mission-focused on a lot of things and helping the merchant be successful, us trying to be a tail into the merchant in terms of what they're trying to do. That is what Sidekick does. One of the benefits coming out of that, of course, is they're more successful. The success of our business is tied to their success. You see that in Sidekick.
The question we sometimes get when products like Sidekick are really successful is, how do you think about the token impact on gross margin?
Yep. Well, it goes a little bit back to what I just alluded to. That is, if Sidekick, and we've given some statistics, we'll give more. If Sidekick continues to drive success, then that translates into gross profit dollars. It may show up in different spots, because if it's showing up in payments, if it's showing up in tax, if it's showing up in FX, cross-border, some of that will be showing up in our, for those of you that know our financial statements really well, some of that shows up in the merchant solutions side of the business rather than the subscription solutions side of the business. But it still goes down to gross profit dollars.
As you know, as we talked about on the last earnings call for our gross margins on subscription solutions, it has been a balance as some of the LLM costs continue to rise with Sidekick usage. To be clear, we want to encourage and we are actively encouraging Sidekick usage. We want merchants to use it. We want them to use it as much as they can. The challenge for us, of course, is make sure some of those additional LLM costs and all the things that we have talked about in terms of how we manage those costs, how we also in areas of support and other things, extract costs there and just on the net, make sure that we are driving gross profit growth through balancing those costs and driving more success elsewhere.
Yeah. To build, we are seeing a lot of really great work from the distillation efforts.
Yeah
Particularly with some of the new open source models. We are getting a lot of leverage with very little or no quality trade-off, but able to leverage them very effectively to keep those costs at very good rates, yeah.
We are doing a lot of caching, too, of questions. You can imagine a lot of times merchants will be asking Sidekick similar questions, so we can be anticipatory in terms of the question they are asking. We do not have to use the latest frontier models and everything because one of the key values of Shopify is 20 years of transaction data, consumer transaction data. Not just, I sent you to a website, but you went to this website, you put this in the cart, you took this out of the cart, you added two things, then you transacted, you shipped, you did all this stuff. All that information is itself the source, then we can run models off of it, but I do not need to have the latest and greatest model to do that.
We covered a lot in 35 minutes on how the world is changing. If there were one or two themes that you want to leave this audience with that may be underappreciated in how the world is changing, I would love to end there. Archie, please.
Yeah, I think that one of the things that I'm really excited about is I think there's two really great trends for Shopify. One is entrepreneurship becomes much more accessible, which will mean a lot more interesting, unique brands and products in the world. That, combined with the ability to actually find those products and expanding the shelf space, the range of products that can even be discovered, I think will only increase with LLMs and with agentic. I think those two things of more entrepreneurs creating more interesting products, more brands able to find those 1,000 true fans, will become much more possible, which will lead to both more merchants, more growth there, but also just more buyers choosing and shopping across independent brands over the next period of time.
The only other thing I'd say on agentic commerce, as we look at what is possible in the years to come, we are very excited about the fact, and there's news about all the different LLMs, and we alluded to it, is whether you get to the merchant's website, which Shopify is running, whether you get to that website by coming through Muse or coming through OpenAI or going straight to the merchant's website or you did a Google search or whatever it is, all those are channels that lead back to the merchant and the broad suite of things that we do to help that merchant be successful. Since commerce began, it's essentially a question of, how do I get more buyers to show up for me?
However they get to me, I welcome them, and then we help them transact, and that's something that we're working really hard to do.
Hey, it's really good stuff. Please join me in thanking Jeff and Archie for their time. Thank you for being here.