← Blog

September 2, 2026 #projects

Why I Stopped Building RadResale

Last month I decided to stop working on RadResale, and I wanted to write down what I built and why I’m setting it down.

I built it for local kids’ resale shops, where the inventory is constantly changing and every item is one of a kind. A stroller, a jacket, a high chair, a toy, most of it never makes it online because photographing, describing, pricing and posting each thing, and then remembering to take it back down, is too much work for a busy shop. The idea was to reduce that whole process to one action, which was taking a photo.

Before getting to why I stopped, I want to show what came out the other side, because I like this product and it got much more complete than a small local experiment had any right to be.

The RadResale site showed the customer view: a simple, browsable storefront for a shop’s one-of-one inventory.

The site used a real-looking shop catalog to make the promise concrete: a customer could see what was in store before making the trip.

It all started with a staff member taking a photo. The intake flow was built for a phone at the counter, so staff could take up to four photos, add a price if they had one, and send the item to the catalog. RadResale then wrote a first title, description, category, condition note and price suggestion, and the staff member could edit any of it.

One-of-a-kind resale inventory is messy, especially when an item looks a lot like several versions of the same product, so I cared about the app being conservative. I didn’t want a weak guess to turn into a confident-looking listing. When the evidence was thin, the product kept the public language general and gave the staff member a note about what to check.

The RadResale product site shows the staff workflow: add photos, review a generated listing, make the item searchable, and remove it after a sale.

The product site’s four-step walkthrough. A staff member could photograph an item, review the first draft, make it findable, and take it out of the public shop once it sold.

Once an item was live, it showed up in the shop’s own storefront, with search, category filters, a detailed item page, photos, size or age information, condition, price, and a link to the current retail price when one was useful. You can still see the live demo storefront and one of its item pages.

The live RadResale demo shop, where customers could search a compact catalog of real one-of-one inventory.

The live catalog. A parent could search, filter by category, open an item, and decide whether the trip to the shop was worth making.

I wasn’t trying to turn resale into regular ecommerce. These items are often available exactly once, and I wanted the customer experience to be honest about that, so a shopper could browse what was there, open the item, and request a same-day hold.

A live RadResale item page with a stroller photo gallery, price, condition information, and a Request hold button.

The live item page for a Nuna stroller. The decision was intentionally small: look through the photos, check the condition and price, then request a same-day hold.

A hold was always the shop’s decision. The buyer sent a hold request and the shop confirmed it, and if an item was already on hold the next interested buyer could join the line. That came directly from a practical concern every shop owner has, which is that holding an item means taking it off the floor for someone who may never show.

When an item sold, staff could find it by scanning its optional QR sticker or by taking a quick photo at the register, then confirm the match and mark it sold, and the listing would disappear from the public shop. That last step mattered as much as getting something online, because a storefront full of sold inventory stops being trustworthy pretty quickly.

The app also included inventory editing, public and private fields, a simple admin for setting up shops, wait-list capture for missed items, and a master view for managing multiple stores. I tested it on my phone with real inventory and deployed it, and it was ready for a pilot.

What I learned once I started talking with shop owners is that what they wanted most was customers. I had a few meaningful conversations with local owners. One was already making Instagram and Facebook Marketplace work well enough, and another was too overwhelmed to add anything new. The clearest feedback was also the most important, which was that a better storefront only mattered if it brought shoppers with it.

RadResale made inventory easier to browse and could make the back room more organized, but it couldn’t create local demand on its own. Facebook and Instagram are messy places to sell one-of-one inventory, but they already have an audience. A cleaner destination asks an owner to promote one more place, and that’s a hard trade unless it produces a measurable outcome.

Looking back, I built too far ahead of the evidence. I had planned to talk with eight to twelve shops before investing deeply, and I got to roughly three meaningful conversations. The app became compelling quickly and building it felt productive, so I kept going.

Listing labor was a real problem, it just wasn’t the problem that determined whether an owner would adopt the product. The owners I met cared more about whether a channel already brought in sales. I knew from the beginning that customer acquisition was the major risk, and more of my early effort should have gone toward testing it. Walking into stores gave me better information than weeks of research or another round of product work.

For RadResale to create enough value it would need to bring local shoppers with it, and that makes it a different kind of business. It would mean parent communities, local promotion, partnerships, search traffic, or eventually a marketplace with inventory from multiple stores. All of those paths ask for local density and sustained consumer marketing. Shoppers want enough nearby inventory before they visit, and shops want shoppers before they contribute inventory. I don’t have the resources or the desire to build that kind of company right now.

There may be a narrower market that would find the product useful. I didn’t find a reachable, motivated market large enough to justify the customer-acquisition effort under this model, and I’m comfortable leaving that question unanswered.

The code, research, domain and product history will all stay around. I’m proud of the application, and the work around AI confidence, human editing, privacy, mobile intake and one-of-a-kind inventory will carry into whatever I build next.

I’m also keeping the lesson, which is that a polished product can make the original idea feel more real than the market has earned. The next time I see that happening, I want to spend more time with the people who would have to choose it.

I like this product, which made the decision harder. I’m glad I built it, and I feel good about setting it down.