I Built a Headless Ecommerce Store with Shopify, HubSpot and AI

For the last few months, I've been experimenting with something I'd wanted to try for a while: a headless Shopify store, with HubSpot CMS running the frontend.

It started with a fairly simple thought.

I really like Shopify as an ecommerce backend. Products, inventory, orders, payments, checkout. It does all of that extremely well.

But I kept running into situations where I wanted more control over the actual shopping experience.

Not just changing colours or moving sections around. I wanted the freedom to rethink how someone browses products, how collections work, how search behaves, how content sits alongside commerce and what happens before someone reaches checkout.

So I decided to separate the two.

Shopify would handle the commerce.

HubSpot CMS would handle the experience.

And that sent me down a very interesting rabbit hole.

Why HubSpot?

HubSpot probably isn't the first platform that comes to mind when someone says "headless Shopify storefront".

That was partly why I wanted to try it.

I was already interested in the idea of bringing the website, content, CRM and ecommerce journey closer together.

Normally these things can feel like separate worlds.

Someone lands on a website. They interact with marketing content. Then they enter "the shop", which looks and behaves slightly differently. Behind that sits the CRM, analytics and everything else trying to piece together what happened.

Using HubSpot CMS as the frontend gave me an opportunity to think about that differently.

Why does the website need to end where the shop begins?

Why can't content, products, customer journeys and CRM sit much closer together?

Once I started building it, the idea made more and more sense.

Then Codex entered the picture

This is where the project became really interesting for me.

I'm not someone who wants to spend three weeks writing a technical specification every time I have an ecommerce idea.

I like experimenting.

"What happens if we do this?"

"Can the basket work like that?"

"Can Shopify send this information here?"

"Can we build this into the product page?"

"Why is this returning a 404?"

A lot of ecommerce development is actually made up of these small questions.

Codex and AI have completely changed the speed at which I can answer them.

I can start with an idea, build a rough version, test it, break it, figure out why it broke and try something else.

Sometimes the first approach works.

Quite often it doesn't.

There have been plenty of moments where fixing one thing has created another problem somewhere completely unexpected.

But that's also been the fun part.

I'm learning while I'm building.

The biggest change has been speed

AI writing code is obviously useful.

But I don't think that's the most interesting part.

For me, the real change is the time between having an idea and seeing whether that idea actually works.

That gap has become incredibly small.

Before, I might have thought of a feature and added it to a list for later.

Now my first instinct is often just to try building it.

Search, product pages, collections, wishlist functionality, basket behaviour, quote requests, different navigation ideas. You can experiment with things surprisingly quickly when the cost of getting the first version wrong is much lower.

And I think that changes how you approach ecommerce.

You stop treating the website as something that gets redesigned every few years.

It becomes something you're constantly playing with.

Build something.

Put it in front of people.

Look at what happens.

Change it.

Try again.

Headless also makes you appreciate everything Shopify normally does for you

There is another side to this.

Going headless has given me a new appreciation for just how much a normal Shopify theme handles without you ever thinking about it.

Once you start separating the frontend, suddenly you care a lot more about APIs, carts, checkout URLs, domains, redirects, analytics, sessions, integrations and all the little connections holding an ecommerce experience together.

There have definitely been moments where I've wondered why I decided to make life more complicated.

But then I build something that would have been awkward within a traditional storefront and remember exactly why.

The freedom is addictive.

Now I want to know if any of this actually works

This is where the experiment gets serious.

I can already say that Shopify + HubSpot CMS + AI has changed how quickly I can build.

What I can't say yet is whether it will change how customers behave.

And ultimately, that's what matters.

Does a better product page increase add-to-cart?

Does better search help people find the right product?

Does bringing content and commerce together make the journey easier?

Does removing friction actually increase conversion?

Or have I just built a technically interesting way of achieving roughly the same result?

I genuinely don't know yet.

And I like that.

Because rather than writing this once I have a nice conversion chart and pretending there was a perfectly thought-out strategy from day one, I wanted to document it while I'm still figuring it out.

The store is built.

The architecture works.

I'm still changing things constantly.

Now comes the part I'm most curious about.

Watching what real customers do with it.

I'll probably come back to this in a few months with the numbers.