When I launched isloved.shop on October 14, 2024, there was no master plan. I'd built an e-commerce site once before, a decade ago, from scratch with a developer — information architecture, the whole thing, learning what worked and what didn't the slow way. Back then I would have killed for something like Shopify. So when I had time and wanted to build something, I was mostly curious whether the platform was as capable as everyone claimed.
It mostly is. There's an enormous amount of functionality available the moment you sign up. There's also a lot of unnecessary complexity to navigate, and everyone pitching a solution seems to want you signed up for their service for life.
The clear-eyed moment came around day 100, when I had to pivot back toward stable income — redesigning my portfolio, interviewing, applying. I'd hoped that by then the site would be producing steady sales on its own. What I found instead was that my sales volume was tied almost directly to my ad spend. I was spending more than I was making, so scaling back became the priority. I also had to look honestly at why my SEO wasn't working, and what it would actually take to fix it.
The most expensive lesson was about automation. I'd leaned on it to place images and load product data into Shopify, and at first it felt like a gift: fast, clean, thousands of items handled at once. What I didn't see were the gaps it left behind. Google flagged nearly every clothing item in the catalog for missing gender and age attributes — the same error, replicated across the entire catalog. I downloaded the data, fixed it in a spreadsheet, tried batching the corrections. Every time I reuploaded, I lost the progress and landed back at square one.
That loop taught me the thing I now build around. Automation gets you eighty percent of the way there, and the last twenty percent is judgment. Not effort — judgment. Knowing which field actually matters. Noticing the thing that's technically populated and still wrong. Catching what no system flagged, because no system was looking for it. That twenty percent is the real work. It doesn't scale, and I've stopped trying to make it.
There were good days too. The first samples arriving was one of them — holding a finished printed piece after months of looking at screens made the whole thing real in a way nothing else had.
Then I took sample puzzles to a retailer here in Seattle, and the feedback was hard and completely correct: every piece was cut to the same shape, which made the puzzle both harder and much less fun to do. The pricing didn't survive the retailer's margin either. Neither problem was a small fix.
Practically, the next phase looks like this. I'm moving in April and setting up a laser cutter, which means cutting puzzles myself instead of ordering them in. I'm looking for visibility that doesn't depend entirely on ad spend. And I have ideas about rebuilding the storefront from the ground up rather than fighting a theme that was never meant for this.
If I could hand my day-one self a note, it would say: don't expect instant results, and don't assume every problem is yours. Platforms change underneath you. All of it is test and learn.
But the thing I keep coming back to is that conversation at the retailer. Standing there with a product that didn't quite work, at a price that didn't quite work, I had to ask what I was actually aiming for. Scale into retail on someone else's terms and margins? Or make fewer things, better, and stand behind every step of how they're made?
I've known the answer since I walked out of that store. I'm just building toward it now.
---
Comments