Dear Fellow CTOs: Stop Building Mobile Apps (Yes, Even With AI)

2 October, 2025 | By Jay Johnston, CEO of poq

Look, I get it. I’ve been in your shoes. Nothing makes a CTO’s heart sing quite like the phrase “we should just build it ourselves.” It’s like catnip for engineers. Mobile apps? Simple, right? How hard can it be to put together some React Native code, hook it up to our APIs, and ship something that doesn’t completely embarrass us?

As a former CTO, I now find myself on the other side of this equation, building an app platform that teams use instead of building their own. And let me tell you, every single day I see smart CTOs making the same mistake: thinking mobile commerce is just mobile development.

It’s not. And I’m here to save you from an expensive education.

The Classic “We’ll Just Build It” Delusion

Here’s how the conversation usually goes in your head:

“We’ve got a solid engineering team. Sarah knows React Native, Mike’s done some iOS work, and our backend is already API-first. We’ll bang out an MVP in a few sprints, iterate based on user feedback, and boom – we’ve got our mobile presence sorted. Plus, we’ll own the whole stack!”

Sound familiar?

The problem is that retail mobile apps aren’t just “mobile development.” They’re mobile commerce platforms disguised as apps. And there’s a world of difference between those two things that you won’t discover until you’re neck-deep in push notification infrastructure, wondering why your conversion rates are terrible and your app keeps getting rejected from the App Store.

“But it’s just a frontend talking to our Shopify API, right?” Wrong. Mobile commerce isn’t about displaying data – it’s about creating experiences optimized for thumbs, cameras, locations and completely different user behavior. Your Shopify API and front-end weren’t built for mobile app traffic patterns or rich omnichannel experiences.

The Brutal Math Nobody Wants to Hear

Let me break down what “just building it ourselves” actually costs:

Your minimum viable team:

That’s £375k+ annually before you’ve written a single line of code.

Timeline reality check:

The hidden costs nobody budgets for:

By year three, you’re looking at £1.5M+ spent, and you’ve got an app that might just be competitive with what your competitors launched in 2019.

The Opportunity Cost Reality Check

Here’s the question that should keep you awake at night: What’s the highest value use of your engineering talent?

Your team could be building:

Instead, they’re debugging why push notifications don’t work on older Android phones and trying to figure out why your app crashes when customers have more than 50 items in their wishlist.

The Platform Economics You’re Missing

Here’s what I see from the platform side that individual teams building their own apps can’t: the power of shared development and collective intelligence.

When we solve a performance issue for one client, every client benefits. When we optimize checkout flows based on data from hundreds of apps, everyone gets better conversion rates. When Apple changes their App Store guidelines (again), we handle the compliance update once, not hundreds of times across hundreds of internal dev teams.

Take App Store Optimization (ASO) – it’s incredibly domain-specific, and most IT teams have never heard of it, let alone understand how keyword density in your app description affects discoverability. Or consider Liquid Glass a native-specific and mandatory UI change from Apple coming out this month in iOS 26 that will significantly impact the user experience (in fully native apps, React Native and even hybrid web-wrappers). Handling this smoothly will require specialized knowledge and migration strategies that most internal teams don’t have.

Then there’s App Marketing itself – the life blood of your app,  driving growth and downloads through smart banner campaigns, tracking mobile attribution, optimizing push notification and Universal Linking campaigns. These are specialist skills that platforms like poq develop because it’s our core business. For your team, it’s just another thing to figure out.

Perhaps most powerfully, when we build an integration with best-of-breed providers like Yotpo for loyalty or Dynamic Yield for personalization, we productize that integration across our entire platform. You get enterprise-grade integrations without the integration work.

By building in house, you’re essentially choosing to solve the same problems that dozens of other retail CTOs are solving right now, in parallel and with no shared benefit. It’s like everyone deciding to build and maintain their own email servers instead of using Gmail.

The network effect here is real: our platform gets better faster than any individual team can improve their own app, because we’re aggregating learnings and improvements across our entire customer base (100+ apps and counting), and we’re doing it in a way that allows you to combine those composable features in unique, Brand-specific ways.

“But Wait, There’s AI Now!”

“Okay, but things are different now. GitHub Copilot, ChatGPT, Cursor – AI is making development way faster. We can probably knock this out in half the time.”

You’re absolutely right that AI is a game-changer. It’s incredible for generating boilerplate code, writing tests, and solving generic programming problems.

But here’s what AI can’t do:

AI can’t give you domain expertise. It can write you a beautiful React component, but it can’t tell you why your mobile conversion rate is 40% lower than your desktop site, or how to optimize your checkout flow for thumb navigation, or why your app’s performance tanks during Black Friday traffic spikes.

AI can’t solve integration hell. Sure, it can generate API calls faster than ever. But when your Shopify integration breaks at 2 AM on Cyber Monday, Copilot isn’t going to debug your webhook handlers or figure out why your inventory sync is showing negative stock levels.

AI can’t handle edge cases. It’s great at the happy path, terrible at “what happens when a customer tries to use a gift card, store credit, and a discount code simultaneously while their session expires during checkout.”

AI can’t navigate App Store peculiarities. Ever tried to get an expedited App Store review or run a phased roll out? Chat-GPT certainly hasn’t.

AI also can’t save you from the “web wrapper” trap. Sure, you could wrap your website in a WebView or as a PWA, add push notifications, and call it an app. Congratulations – you’ve just guaranteed terrible conversion rates and one-star reviews. Native performance and platform-specific UI patterns aren’t nice-to-haves, they are now the table stakes that today’s shopper just expects.

And here’s the kicker: AI hasn’t eliminated the build vs. buy decision – it’s made the opportunity cost even higher.

If AI makes everything faster to build, then every month your team spends building mobile commerce infrastructure is an even bigger waste. You could be using those AI productivity gains on features that actually differentiate your business instead of rebuilding shopping carts for the millionth time.

And the final plot twist? We’re using AI too. All those productivity gains you’re thinking about for your internal development team? We’re already applying them to platform development, which means faster feature releases, better optimization, and lower costs for everyone using the platform. You get the AI acceleration without having to build the infrastructure.

The Bottom Line

Your engineering team is probably fantastic. That’s exactly why you shouldn’t waste them on this.

Every retail mobile app faces the same fundamental challenges: performance, conversion optimization, platform compliance, integration complexity, and keeping up with the endless parade of iOS and Android updates. These aren’t differentiating problems – they’re solved problems.

What differentiates your business is how you serve your customers, what unique experiences you create, and how you operate more efficiently than your competitors. That’s where your team’s creativity and expertise should be focused.

The platform economics are simple: we solve these problems once, well, and everyone benefits. You solve them once, adequately (maybe), and only you benefit.

AI isn’t going to change this equation. If anything, it makes it more important to focus your AI-enhanced productivity on problems that actually matter to your business.

So do yourself (and your team) a favor: buy the platform, build the differentiation. Your future self will thank you.

And if you ignore this advice and build it yourself anyway? Well, I’ll be here when you’re ready to talk in 18 months.

P.S. – Yes, we’d love to show you what a proper mobile commerce platform looks like. But even if you go with another provider, please, for the love of all that is holy, don’t build it yourself.

Jay Johnston is the CEO of Poq Commerce and a former CTO who learned that building platforms is a little harder than it looks. He still codes occasionally but, much to the relief of his Engineering Team, he mostly sticks to strategy, drums and cycling these days.

Get in touch

Start your appcommerce journey today
Contact Us