A year on Gumroad: 19 products, one verified sale, and what it taught me

I ran a Gumroad shop with 19 digital products for twelve months. Public URL: modernwebseo.gumroad.com. Public number of verified sales over that period: one.

This is the postmortem.

What I actually did

In early 2025 I built a Next.js shop at toolgenx.com that listed nineteen products. The shop was real. The product pages had JSON-LD, OG images, FAQ sections, the works. The buy buttons sent you to modernwebseo.gumroad.com.

The thinking was reasonable. Build the storefront fast. Outsource the cart, the receipts, the VAT, the file delivery, the license keys, the chargeback handling — to Gumroad. Validate the offers first. Move to direct checkout when there is something worth moving.

The plan worked at every step except validation.

The numbers

Twelve months, nineteen products, one verified sale. About $23 net. Worth less than my electric bill.

The full picture from Plausible and Gumroad combined:

• Product page views: roughly 8,400 over the year
• Buy button clicks: roughly 340 (4% click rate, decent for a storefront)
• Gumroad checkouts started: unclear (Gumroad does not surface this for sellers on the free tier)
• Verified sales: one

The drop happened somewhere between my domain and Gumroad's. I could see traffic to toolgenx.com and clicks on the buy button. I could see one paid order in Gumroad. I could see nothing in between.

That blind spot is the actual story.

What the blind spot cost

A direct checkout funnel — Stripe Checkout, your own page, your own analytics — would have told me at least four things I did not have:

1. Who landed on the checkout (so I would know if the redirect bounced)
2. Who entered an email and got stuck (so I would know if a form field broke)
3. Who selected Apple Pay vs card (so I would know about mobile friction)
4. Who completed but then refunded within an hour (so I would know about post-purchase regret signal)

With a Gumroad redirect I had none of those. I had "page view" upstream and "order" way downstream. Everything between was a black box owned by someone else.

The 10% platform fee was actually the cheapest cost of using Gumroad. The cost I should have been counting was the cost of not knowing what was breaking.

Why I did not switch sooner

Two reasons, both honest.

Reason one: the fix felt expensive. Wiring up Stripe Checkout + Supabase + webhook + email + license + refund flow sounded like a week of work that might still not move the conversion needle. The math of "definite week of engineering vs uncertain conversion lift" never pencils out cleanly.

Reason two: I was protecting an assumption. I had assumed the product offer was the problem ("I just need better positioning, better landing pages, better SEO"). Switching the checkout would force me to confront the possibility that the checkout itself was the problem. Easier to write another blog post.

The whole postmortem is a lesson in confirmation bias.

What changed my mind

A friend who runs a small SaaS shop asked me a single question over coffee:

"If your conversion rate is 0.01% over twelve months and the average D2C software conversion is 1-3%, you have either a wildly wrong offer or a wildly wrong funnel. Which one have you actually ruled out?"

I had ruled out neither.

The product offer side was harder to test (it would mean retiring products, repositioning, all the things I had been avoiding). The funnel side was concrete. Two days of engineering and I would know.

I committed to the rewrite the next morning. Three weeks later, the new site shipped with direct Stripe + Iyzico checkout. The conversion data over the next 90 days is what will tell me whether the offer was the problem or the funnel was.

Five things I would tell past-me

If I were starting over with the same nineteen-product catalog:

1. Set a sales-count tripwire. "Switch to direct checkout the moment cumulative sales hit ten." Do not negotiate with it.
2. Instrument the funnel before launching. Even the simplest Plausible custom events (product_view, buy_click) would have surfaced the redirect bounce months earlier.
3. Treat platform fees as the cheap part. They are. The expensive part is the data you do not see.
4. Pick one channel and one offer to test. I had nineteen products diluting the signal. With fewer products on the same traffic, the data would have been louder.
5. Stop polishing the storefront. It was the wrong axis. The right axis was distribution and checkout instrumentation.

What I kept

Three things from the Gumroad year are worth keeping:

• The product catalog itself. All nineteen products still exist, ported into the new site. The work was not wasted, just under-monetized.
• The Gumroad shop as a fallback. modernwebseo.gumroad.com is still live for buyers who prefer it. The redirect is gone from the new site, but the shop runs as a parallel surface.
• A clearer picture of what an MVP storefront actually requires. Not a polished site. A working funnel with visible data.

What I changed in the rewrite

Three structural things, all directly traceable to the Gumroad lesson:

• Direct Stripe + Iyzico instead of redirect. Buyer stays on toolgenx.com from arrive to receipt. (Full comparison of the three options here.)
• Funnel instrumentation with Plausible custom events at every step (product_view, buy_click, checkout_start, purchase).
• EU CRD consent flow with logged evidence. Not just a refund policy on a page — an actual database row with timestamp, IP, user-agent, and the exact text the buyer agreed to. The Gumroad redirect had hidden that all from me.

Whether the new funnel converts better than the Gumroad redirect is an open question. The difference is that this time, when it does or does not, I will be able to see why.

Closing the loop

If you are running a Gumroad shop and recognise any of this, here is the test I should have run a year earlier:

• Plot product page views per week for the last three months.
• Plot completed Gumroad sales for the same period.
• Calculate the ratio.

If the ratio is wildly outside the 1-3% range you would expect for D2C software, the problem is either the offer or the funnel. The funnel is the cheaper one to test. Switch the checkout. Two days of work. Stop waiting.

---

This post is part of the solo founder series. The technical decisions behind the rewrite are in Why I rewrote my 19-product site in two weeks. The payment-stack decisions are in Stripe vs Iyzico vs Gumroad. The 19 products are here.