Why Your Shopify Store Is Broken (Even On World-Class Infrastructure)

Every person who shops online is affected by this problem, but very few people are actually fixing it.

Noel Mathew spent over 3 years at Fynd, one of India's major e-commerce platforms, building distributed systems and infrastructure that power millions of transactions. He architected event-driven systems, scaled real-time data processing, and built cloud-native solutions that handle massive throughput. He worked on everything from image transformation to data pipelines to metering systems - all in the e-commerce space.

Then he realized something: the e-commerce industry is solving the wrong problem.

So Noel built Shopcan. It's an AI-powered tool that analyzes 200+ signals across your entire store in 30 seconds and tells you exactly what's killing your conversions. It's already analyzed more than 2.5 million pages and helped stores improve their scores by an average of 47%.

Today, we're learning why even stores built on world-class infrastructure are fundamentally broken - and what you can actually do about it.

Key Takeaways: The Hidden Problem With E-commerce

The Infrastructure Paradox:

• Shopify has billion-dollar infrastructure, world-class AI tools, massive ecosystem - yet stores are still broken
• Using scalable systems doesn't guarantee performance if you don't configure and fine-tune them correctly
• Junior developers patching code together create unoptimized stores with no clear starting point for fixes

Performance Kills Revenue:

• Faster stores convert 2.5x better than slow ones
• Every slow interaction gives customers time to doubt their purchase intent
• Performance issues lead directly to cart abandonment and lost revenue

The Six Categories That Cost Money:

• Performance, SEO, security, conversion, mobile, and content
• Performance is the biggest money killer, followed by content
• People get told what to fix (Page Speed Insights) but don't understand how to actually fix it

Q: You spent 3 years at Fynd building massive distributed systems. Why walk away from that to start your own company?

Noel: I joined Fynd and worked on a new product we were building from scratch called Pixelbin. The opportunity I got was I could build things like authentication, authorization, SaaS metering, and data pipelines from scratch. We actually scaled it to over 10 million users, and the data pipelines were processing over 150 million messages a day.

Working at that scale actually gave me the confidence that this is a good time to try to learn how to build and sell products. Shopcan isn't the first product I built. The first product is called Beamup, and we launched it in June-July. We got our first customer last month.

Shopcan actually came out of my experience working at Fynd, some freelance work I've been doing after Fynd, and a discussion me and my brother were doing brainstorming ideas. That's how we started Shopcan.

🔥 ChaiNet's Hot Take: Building at scale teaches you what's possible. Starting a company teaches you what matters. They're not the same curriculum.

Q: Tell us about Beamup first. What does that do?

Noel: Beamup is a file uploader for getting files into your cloud storage. A lot of workflows for cloud-native companies are connected to their cloud storage - something like S3 for AWS or GCS for Google Cloud. Even BigQuery connects directly to your Google Drive.

The thing is, a lot of companies already have custom pipelines set up which connect to their storage, but people responsible for getting the data into that storage aren't that technical. It might be project managers or even customers asked to upload to shared drives.

Beamup actually connects to your storage and creates an upload portal which allows non-technical users to just drag and drop files, and then your pipeline can pick that up from there.

🔥 ChaiNet's Hot Take: The best SaaS products solve problems so boring that nobody talks about them at conferences. File uploads won't get you on TechCrunch, but they'll get you revenue.

Q: When did you realize there was an opportunity with Shopcan that people were overlooking?

Noel: While I was working at Fynd, we were building Pixelbin. Our main scope was media delivery since we were offering CDN as a core offering. We always used to get reports like slow load times for store landing pages. We used to get a lot of fingers getting pointed at us, but mainly the issue was wrongly sized images or unoptimized image formats.

The knowledge is out there. Even Shopify gives you a checklist of what you should and shouldn't do. But "is my store doing well?" - that's the question nobody's answering directly. That was the opportunity I thought we could come in and address. You just run a scan on your store and you know if your store is up to the mark.

🔥 ChaiNet's Hot Take: Checklists don't help when you don't know which items are checked. Shopcan is a mirror that shows you the truth, not just the theory.

Q: This is counterintuitive. Shopify has billion-dollar infrastructure, world-class AI tools, massive ecosystem. How are stores still fundamentally broken?

Noel: There's this ongoing joke on Twitter: "I've got 10 users, it's time for Kubernetes." Using a massive or scalable system itself doesn't guarantee you the best performance. It's all about configuring it and ensuring you're fine-tuning it the way you need to.

That's something happening with Shopify as well. I'm not pulling anyone down, but I've seen a lot of codebases - the Shopify store codebases - and it looks like juniors have patched things up. It's so unoptimized that people don't know where to start fixing it from.

🔥 ChaiNet's Hot Take: World-class infrastructure with junior implementation is like a Ferrari with a learner's permit. Technically impressive, practically terrifying.

Q: You mentioned faster stores convert 2.5x better. Walk me through what happens when a store is slow. How does it actually cost revenue?

Noel: Let's think from a user's point of view. Usually stores that are investing heavily and generating a lot of revenue do it mainly through performance marketing. The user is scrolling on Instagram, they get an ad, they're very excited about the product. But they're also very triggered from the entire scrolling thing.

The moment they land on the page and the page loads very slow, they're already off their intent. Then every click, every interaction with the website - if it feels slow, if things don't load, if it's blotchy somewhere - the user's original intent of "I want to get that" goes from "do I even need that? Maybe something of that sort."

That's where people see dropouts. That's why performance of stores is very important. Even if your stores don't load fast and don't have the right keywords and don't have enough content to put you up on SEO, you're missing out on a lot of traffic that comes to your store.

🔥 ChaiNet's Hot Take: Slow pages don't just lose sales. They give customers time to remember they don't actually need another t-shirt.

Q: You break down store problems into six categories: performance, SEO, security, conversion, mobile, and content. If you had to pick one that's costing store owners the most money, which is it?

Noel: Mainly it would be performance. Then there's content. With performance, there are a lot of tools like Page Speed Insights, but I constantly got messages from friends working at e-commerce stores saying "we see Page Speed Insights say we have to fix this. We've already fixed it, but it doesn't work - it's still giving the same error."

The thing is, people are getting what to fix, but they don't clearly understand how to do it and how to optimize it. For content, since content is cheap now with ChatGPT giving you tons of content, people are just plugging in content. It feels like a lot of high-signal words, but it's somewhere empty.

🔥 ChaiNet's Hot Take: Google tells you the symptom. It doesn't tell you the surgery. That's why Page Speed Insights makes developers cry.

Q: What about Shopify's new AI features - sidekick, the sim gym simulation thing? Are those helpful or just marketing hype?

Noel: I was really excited. I'm still going through a lot of the features myself. A lot of them are "coming soon" - they haven't been released yet. There's a CRO thing, there's sim gym. I am very excited about those features. I don't know how this is going to be for Shopcan or similar products, but I think if they've implemented it very well, you can get a lot of mileage. But it's still to be revealed.

The industry has been going through marketing hype for the last 3-4 years now. Everything's AI. We're not sure. Only when these features actually come into place will we get to know whether they're worth the hype or not. And then there will also be a learning curve of actually understanding how the products are working and getting to use them the right way.

🔥 ChaiNet's Hot Take: "Coming soon" is the corporate version of "my girlfriend goes to another school." Show me production code or show me the door.

Q: There's a debate happening - is Shopify trying to eat their partners' businesses with these AI features? What's your take?

Noel: Shopify has always been strong on the partner side. There should be a lot of opportunities there as well. For agencies, Shopcan is more like a lead generation tool. You get a structured, downloadable PDF report which is white-labeled.

You could just maybe run a monthly audit for all the stores you're managing, and you'll get a proper report you can send across. Or you can actually see all the metrics and ensure everything's working fine. Only if anything is wrong do you have to dig in.

🔥 ChaiNet's Hot Take: Platforms threaten to eat their ecosystem every year. Then they remember ecosystems are what make platforms valuable. It's a ritual, not a threat.

Q: How does Shopcan help store owners who manage stores themselves?

Noel: A lot of owners actually manage stores themselves. I've been seeing a lot of threads on Reddit like "what should I do to improve my SEO?" Shopcan helps them by having an AI-powered recommendation system. We actually go through your issues and recommend what to fix, how to fix it, and we give a priority based on what would give you the most ROI.

That gives owners some grounds to talk to the vendors they'll be talking to to get things done. These are the things which need to be done.

🔥 ChaiNet's Hot Take: Shopcan is like having a mechanic inspect your car before you take it to the dealership. Suddenly you know what's actually broken versus what they're trying to upsell.

Q: Walk me through a specific example. A store was broken, you recommended fixes, they implemented them. What changed?

Noel: I was working for a store and what happened was they had inconsistent theming across pages. Some pages were from a different theme. Their main pages were from a different theme. That actually breaks your branding. Then there was this thing - they had a "shop all" page where they put everything into one collection.

That page was serving 250 products right from the get-go. They actually thought since there was no pagination button visible, they thought they had infinite scroll. But actually there was no lazy loading involved, and they were constantly complaining "our page speeds are doing bad but we don't know why or how to fix it."

The problem is, it's very easy to do manually rather than going through all the pages and understanding what's broken. Shopcan isn't solving anything far-fetched. It's just giving you all the lowest hanging fruits that you can fix to get your store up to the mark.

🔥 ChaiNet's Hot Take: Loading 250 products at once is like opening every drawer in your house when looking for your keys. Technically everything's available. Practically you're screwed.

Q: Where do you think e-commerce is going in the next few years with AI?

Noel: People have to buy stuff. We're stuck in this consumerism thing. E-commerce isn't going anywhere - it's only going up. But honestly, I'm not too sold on AI window shopping for us because we enjoy doing that right now. I just want to see how it will work out.

Right now it looks like wealth and results. There's no core offering that actually improves the experience. If you think of it, Google - now if you search for something, there's a lot of sponsored offerings you get. So if you're looking for something relevant, you have to scroll past that. That's going to come into AI as well. Agentic storefronts as well. Will that be putting people off? How will that work out? There are things that need to be answered yet.

🔥 ChaiNet's Hot Take: AI shopping agents will have the same problem as Google: whoever pays the most gets recommended first. Meet the new boss, same as the old boss.

Q: What about clothes shopping specifically? Can AI help there?

Noel: A lot of stores - people want to buy clothes, for example. I constantly keep hearing from my wife that the collections are something she saw 3 years ago. The problem isn't that AI will make it better. You have to have a better collection to sell. People want something new. That isn't solvable by AI unless you're doing it in the backend. The storefront won't help that.

🔥 ChaiNet's Hot Take: AI can optimize your storefront. It can't make your products interesting. That's still a human problem.

Q: You've been doing this entrepreneurship thing for 10 months now. How does it feel? Would you recommend it?

Noel: I'm still getting used to the idea of salary not hitting your bank every 30th. This is not for me to recommend somebody. It's more like a calling, and I'm still in two minds of which way to go. But suddenly things are looking better for me. So I think I should stick to this a bit more. I'm still too early to make a comment here.

🔥 ChaiNet's Hot Take: Entrepreneurship isn't a career choice. It's what happens when the pain of staying comfortable exceeds the pain of being broke.

Q: What should someone do if they're 4-5 years into their job like you were before you left? What's the checklist?

Noel: I built a runway of almost 2-3 years. I would say the more runway you have, the better the sanity you'll keep going through it. One thing I did wrong was I didn't research what product to build or how I would be navigating throughout the journey before I took the plunge. That would actually help because I started working on things maybe a month or two later.

That's something that would help people if they have a plan beforehand. But not a lot of plans work out. One more thing would be learn to sell. That's the most difficult thing I'm figuring out right now. Building stuff is something that comes very naturally to us engineers, but selling - talking to people - that's one of the biggest challenges. I'm going through threads and trying to just learn.

🔥 ChaiNet's Hot Take: Engineers think the hard part is building the product. It's not. The hard part is convincing humans with money to care.

Q: What's the reality of running two products at once?

Noel: Shopcan came out of recognizing a problem that I saw repeatedly while working at Fynd and doing freelance work. Beamup came first, but Shopcan is getting more traction because it solves a more visible pain point. You have to be willing to pivot and follow where the validation is, not where your ego wants to go.

🔥 ChaiNet's Hot Take: Startups are experiments. Successful founders run multiple experiments until one starts working, then they kill the rest.

Final Thoughts: Measurement Precedes Improvement

Noel's insight: "The knowledge is out there. Shopify gives you checklists. But 'is my store doing well?' - that's the question nobody's answering directly."

The bottom line: Noel spent 3 years building infrastructure that processes 150 million messages per day for millions of users. He learned how to build things that scale. But that's not what made him start Shopcan.

What made him start Shopcan was realizing that having world-class infrastructure means nothing if you don't configure it correctly. Shopify can have billion-dollar engineering and a 12,000-app ecosystem, but if a junior developer copy-pastes unoptimized code into your store, you're still going to have slow page loads, poor SEO, and broken conversions.

The stores doing well aren't the ones with the best infrastructure. They're the ones that actually measure what's broken and fix it systematically. Performance, SEO, security, conversion, mobile, content - six categories, 200+ signals, analyzed in 30 seconds. That's what Shopcan does.

Because here's the reality: You can't improve what you don't measure. Page Speed Insights tells you there's a problem. It doesn't tell you the solution. Shopify gives you checklists. It doesn't tell you which items matter most for your specific store. Agencies will optimize your store. They charge $10K and take 3 months.

Shopcan tells you exactly what's broken, how to fix it, and what will give you the most ROI. In 30 seconds. That's not revolutionary technology. It's just solving the actual problem instead of the impressive-sounding problem.

Faster stores convert 2.5x better. Every slow interaction is revenue bleeding out. Every poorly optimized image is a customer clicking away. Every missing meta description is traffic you'll never get. The infrastructure doesn't matter if the implementation is broken.

You can't AI your way out of a poorly configured store. You can't app-install your way to good performance. You have to measure, fix, and verify. In that order. That's not sexy. It's just what works.

Q: How can people connect with you and try Shopcan?

Noel: You can find me on LinkedIn where I share insights about e-commerce performance, what we're learning from analyzing millions of pages, and the journey of building Shopcan and Beamup. I'm always happy to talk shop with other founders and engineers working in the e-commerce space.

Final words: The best solutions aren't the most impressive. They're the ones that solve real problems people are actually facing. Noel didn't build Shopcan because distributed systems are cool. He built it because stores are bleeding revenue from problems they don't even know they have. Measure first. Fix second. Optimize third. If you skip step one, steps two and three are just expensive guesses. World-class infrastructure with poor configuration is just expensive broken. Start with measurement. Everything else follows.