<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Gavin Ballard on Medium]]></title>
        <description><![CDATA[Stories by Gavin Ballard on Medium]]></description>
        <link>https://medium.com/@gavinballard?source=rss-45db9babd79d------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*m-PblmKgeTEJm0D1WqA6-w.jpeg</url>
            <title>Stories by Gavin Ballard on Medium</title>
            <link>https://medium.com/@gavinballard?source=rss-45db9babd79d------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 17:32:59 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@gavinballard/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Moats for GOATS]]></title>
            <link>https://medium.com/@gavinballard/moats-for-goats-bea1510da826?source=rss-45db9babd79d------2</link>
            <guid isPermaLink="false">https://medium.com/p/bea1510da826</guid>
            <category><![CDATA[business-strategy]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[saas]]></category>
            <category><![CDATA[tech-startups]]></category>
            <category><![CDATA[shopify]]></category>
            <dc:creator><![CDATA[Gavin Ballard]]></dc:creator>
            <pubDate>Tue, 25 Aug 2026 05:25:07 GMT</pubDate>
            <atom:updated>2026-08-25T05:25:40.155Z</atom:updated>
            <content:encoded><![CDATA[<p>Building sustainable and defensible Shopify offerings in the age of AI <strong><em>(or, “are we f**ked yet?”)</em></strong></p><p>I was lucky enough to be invited to speak at Brightwork a few weeks ago as part of Shopify’s DotDev conference in Toronto. A recurring theme across the week was “is AI going to eat the Shopify App and Agency ecosystem?”, so I thought I’d share my take on the question as well as a few ideas I<br>have about what we can do to improve the defensibility of our apps and services in the new world.</p><p>The TL;DR of my talk<br>Fifteen minutes, in a room full of freelancers, app developers, and agency owners — the people whose businesses are most directly impacted by the threat of a wave of AI-driven copycats. My argument was simple enough: if what you think you’re selling is code, you’re in trouble — but that<br>was always the case. Most of my time was spent exploring the alternatives — six different types of real-world moats I’d seen businesses build in the Shopify ecosystem, illustrated with apps everyone in the room already<br>knew.</p><p>Why I thought it was worth saying</p><p>To me, there’s a misplaced fear of the vibe-coding copycat flooding the App Store. Five years ago, someone could hire an offshore team to clone your app over three weeks — so I feel the business risk hasn’t changed, it’s just that the timescale has compressed. If a competent clone could kill your<br>business, you were vulnerable long before any of us had heard of Claude.<br>What has changed is what sits underneath. Someone asked me a question after the talk about features — whether building a lot of them counts as a moat — and I gave an answer I’ve been thinking about ever since:</p><p>The value of features in and of themselves is the same as the value of code, which is trending very quickly to zero. And I say that as someone who writes code and loves code. Most app businesses believe they sell features. I don’t think they do. A feature list is code with marketing on it, and both of those things are getting cheaper by the month. Which is, oddly, why I’m optimistic. Every moat I talked about has one thing in common: you can’t just wake up one day and decide to have one. You have to earn one through accumulation.</p><p>It takes twenty years to get twenty years of brand trust. A data moat takes years of collecting data that’s actually worth something to<br>somebody. Our own expertise came out of the better part of a decade running an agency and living merchants’ problems before we built anything worth selling.</p><p>AI has collapsed the cost of building. It hasn’t touched the cost of accumulating, and I don’t think it’s going to. That’s the bit worth defending, and it’s why I think the pressure is real but the panic isn’t warranted. None of this is abstract for me, incidentally — it’s what I think about most weeks in our own business.</p><p>Six moats for GOATS</p><p>These are the six I thought of and discussed in the talk:</p><p>• Brand and trust — Klaviyo, Mechanic. The hardest thing on this list to fake right now, which means it’s appreciating.<br>• Data — Fairing, Gorgias. Data that has network effects and gets better for every merchant as another merchant joins.<br>• Expertise — this one’s ours, and I’ll own the bias. People come to Submarine for the API and stay for the answer to “how do we actually make this work on Shopify?”<br>• Schlep — TaxJar, Swap. County-level US sales tax and cross-border duty regimes are not a weekend build, making boring things a moat.<br>• Risk shifting — Rewind. Nobody is going to build their own Rewind. If someone’s paying you to stop owning a problem, they keep paying.<br>• Price — Judge.me. Fifteen dollars a month rounds to zero for a business. You’d spend more on tokens replicating it than you’d ever save.<br>That last one cuts against roughly every piece of pricing advice I’ve ever given, and I suspect it makes people uncomfortable because it implies downward pressure on all of us. I still think it’s one of the most interesting ones on the list.</p><p>“If it was easy, it wouldn’t be a moat”</p><p>That was my answer to the best question I got on the day, from someone building a young app: how do you build brand and trust when you’re starting from nothing? You can’t shortcut twenty years. But you can start the clock, and the Shopify ecosystem is unusually good to the people who do.</p><p>Katie Keith made a great point in her talk about building in public, and<br>there’s a real tradition here of year-in-review posts and people sharing numbers and mistakes they have no commercial reason to share.</p><p>Make friends with agencies and find out what their clients complain about, too — though be honest that what you’re getting is reflected pain. It isn’t the same as having felt it, and it doesn’t accumulate the same way.<br><br>I gave this talk to Shopify people because that’s my ecosystem, but I don’t think any of the argument is Shopify-specific. Anyone building software on somebody else’s platform is dealing with the same three pressures: customers who can increasingly build it themselves, a platform that might absorb your category, and buyers who are progressively less likely to be human.<br>It works at the level of a person, too. If a competent generalist with a good model could produce your output, what you’re left holding is your context, your relationships, and knowing why the thing<br>was built the way it was.</p><p><strong>So — are we f**ked?</strong></p><p>I don’t think so, and not just to hold up Betteridge’s Law. There’s still a lot that AI can’t replicate or shortcut, and as long as you’re focusing on those things rather than seeing the answer to your challengers as “ship more code and features”, you’re in good shape. For now, at least.</p><p>You can watch my talk, along with the entire Brightwork content run, <a href="https://brightwork.dev/vol-01/">here</a>.</p><p><em>AI was used to help outline this piece, but the words are my own — I’ve been using em dashes since before it was cool and/or robotic.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=bea1510da826" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Shopify Flow: diving deep into ecommerce automation at scale]]></title>
            <link>https://medium.com/@gavinballard/shopify-flow-diving-deep-into-ecommerce-automation-at-scale-690ae4fe3237?source=rss-45db9babd79d------2</link>
            <guid isPermaLink="false">https://medium.com/p/690ae4fe3237</guid>
            <category><![CDATA[shopify-development]]></category>
            <category><![CDATA[ecommerce]]></category>
            <category><![CDATA[shopify]]></category>
            <category><![CDATA[shopify-flow]]></category>
            <category><![CDATA[shopify-apps]]></category>
            <dc:creator><![CDATA[Gavin Ballard]]></dc:creator>
            <pubDate>Thu, 02 Jul 2026 02:36:23 GMT</pubDate>
            <atom:updated>2026-07-02T02:36:23.890Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>What Shopify Flow actually makes possible and why most merchants are only using half of it</strong></p><p><em>A conversation with the Shopify Flow product team.</em></p><p>A few months ago I had the chance to sit down with <a href="https://www.linkedin.com/in/paul-nuschke-11a6a34/">Paul Nuschke</a> and <a href="https://www.linkedin.com/in/rpenate/">Rolando Peñate</a> from the <a href="https://apps.shopify.com/flow">Shopify Flow </a>product team to talk about where Flow is headed and how we’re thinking about it at <a href="https://www.getsubmarine.com/">Submarine</a>. What came out of that conversation has shaped how we approach our own Flow integration, and I think it’s worth sharing more widely.</p><p><strong>Flow’s value is in it’s connectivity</strong></p><p>Most people think of Flow as a task automation tool. Set a trigger, define an action, save some time. That’s true, but it undersells what’s actually happening at the ecosystem level.</p><p>Paul put it better than I could:</p><blockquote><em>“What’s particularly interesting is how Flow connects the entire ecosystem without requiring point-to-point integrations. To me, this is one of the biggest long-term values of Flow — how it can connect the whole ecosystem without each of the individual apps having to write point-to-point integrations with each other. So the more people that build on Flow, the more valuable it becomes, because it just has all the things already.”</em></blockquote><p>For Submarine this has been genuinely practical. We don’t have a direct Klaviyo integration, but we have prebuilt Flow templates that connect the two. That’s a different kind of flexibility -one that scales without us maintaining a bespoke integration for every tool in a merchant’s stack.</p><blockquote><em>“For us, it’s really been a game changer in terms of avoiding the need to build a one-on-one integration with a whole bunch of other systems. It’s been really handy being able to say: we don’t have a direct integration with Klaviyo, but we do have all of these prebuilt flows that we’ve been able to put together.”</em></blockquote><p><strong>The “Run Code” feature has changed what’s possible without custom development</strong></p><p>In the past 12 months, Flow has become significantly more capable. The addition of custom code execution inside a workflow has shifted how we think about when to build something bespoke versus when to invest in the Flow integration instead.</p><blockquote><em>“So many of the customisation use cases that we’re trying to solve for can now be achieved with Flow because it has been getting more and more powerful in the background with additional triggers and actions like the Admin API request tool. 80% of the things that we might originally have turned to a custom serverless function to solve for, we can now get into Flow, which is actually really pretty exciting.”</em></blockquote><p>Our thinking has shifted accordingly. For anything that runs asynchronously, we’re better off investing in making the Flow integration as strong as possible rather than building parallel infrastructure.</p><p><strong>Metafields are an underrated use case</strong></p><p>One specific area worth calling out. Flow is one of the only ways to automatically push data into a metafield. Paul’s framing on this was sharp:</p><blockquote><em>“Flow is a superpower for metafields because you can automatically get data into a metafield using Flow, whereas you can’t really do that any other way.”</em></blockquote><p>If you’re building merchant experiences that depend on metafields -merchandising tied to inventory levels, product flags, dynamic display rules - this is the cleanest path available.</p><p><strong>The scale makes it reliable</strong></p><p>It’s worth noting how many merchants are actually running on Flow. According to the Shopify team:</p><blockquote><em>“There’s about 800,000 merchants that have active workflows. And that’s just the beginning. Effectively all merchants are running on Flow because we power abandoned checkout. Scale-wise, that’s millions of merchants running many of those events.”</em></blockquote><p>At that scale, Flow is infrastructure so it is not optional.</p><p><strong>How we treat Flow at Submarine</strong></p><p>We’ve come to view Flow as a first class citizen of the platform, the same way we think about our webhook endpoints or our GraphQL API.</p><blockquote><em>“We spend just as much time with the Flow integration.”</em></blockquote><p>For merchants, that means more is achievable without custom development than most people realise. For agencies scoping Shopify projects, Flow should be in the conversation to remain proactive in a volatile.</p><p>The Shopify team has more coming: better observability and debugging tools, expanded education resources for service partners, and continued improvements to metafield handling. The platform is still building, and the ceiling keeps moving.</p><p><em>Gavin Ballard is the founder and CTO of </em><a href="http://getsubmarine.com/"><em>Submarine</em></a><em>, a flexible commerce platform for subscriptions, preorders, memberships and crowdfunding on Shopify.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=690ae4fe3237" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>