How we test

Most sites in this subject establish authority with a face and a screenshot: here’s who I am, here’s my dashboard, therefore believe me. We don’t do that. PlainFunnel publishes under an organisation rather than a person, and we don’t show earnings.

That removes the usual shortcut, so this page is the substitute: how we get our numbers, what we do when sources disagree, what we refuse to print, and where the gaps are. It’s meant to be checkable. Everything below names something specific you can go and verify yourself.

Where the numbers come from

Every figure in an article traces to a primary source we read directly — a platform’s own fee page, help centre, or terms — not to another article that quoted it. That distinction matters more than it sounds. Most of the numbers circulating in this niche are blog posts citing blog posts, and the original claim frequently doesn’t exist.

When we quote a platform, we record the month we read it and link to the page so you can check the current version. Terms get revised without announcement, and a quote with no date attached is a quote you can’t evaluate.

What we do when sources contradict each other

We publish the contradiction instead of picking a winner. The live example, as of August 2026: Gumroad’s pricing page presents 10% + $0.50 as the per-transaction fee, while its help centre says that figure “does not include credit card processing or PayPal fees” — which would add roughly 2.9% + $0.30 on top. Those are both official pages and they don’t agree.

We could have picked the friendlier number and moved on. Instead the pricing article publishes both readings as a range, names the conflict, and tells you to check your own payout. We’ll settle it when a real sale gives us a real payout to read — and not before, because until then we’d be guessing with a confident face.

What we don’t publish

No income claims and no earnings screenshots. Not ours, not anyone’s. This is the single most common way this subject misleads people, and there is no version of it we’re willing to do.
No benchmark we can’t trace to a source. We previously used a set of conversion-rate benchmarks across several articles. When we went back to source them properly we couldn’t establish where they came from, so we removed them everywhere and replaced them with clearly-labelled scenarios that are presented as illustrations rather than data. An unverifiable number is worse than no number, because you’d anchor on it.
No restating of other companies’ policies. We don’t reproduce Etsy’s or Gumroad’s refund rules, fee schedules or seller requirements as though we maintain them. They change, this page wouldn’t, and you’d be relying on us for something we don’t control. We explain the mechanics and link you to the source for the rules.
No verdict shaped by what we sell. Where the honest answer is inconvenient for us, we print it anyway — our pricing article says plainly that Gumroad, where we sell, is worse than Etsy on a $3 product.

The arithmetic is meant to be checked

Where an article turns on a calculation, we show the calculation rather than the conclusion, and we verify it with a script before publishing rather than doing it in our heads. The point isn’t that our numbers are impressive — it’s that you can run them against your own price, your own fees and your own traffic and get a different answer if yours differs. The calculator exists for exactly that.

What we don’t know

This is the part most sites leave out, and it’s the part that should tell you the most.

We don’t know your conversion rate, and neither does anyone quoting an average at you. It varies enormously by traffic source, price and category. Every conversion figure we show is a labelled scenario, never a prediction.
We can’t currently settle the Gumroad processing-fee question — described above. It needs a real payout to resolve, and we say so rather than rounding it away.
We don’t have a sourced “normal” refund rate. The figures that circulate for this have no traceable origin, so our refund article declines to state one and tells you to watch your own baseline instead.
We don’t track platform policy changes, and we don’t pretend to. Anything policy-shaped here is dated, and the live answer is on the platform’s own help centre.
We can’t tell you whether this will work for you. Demand for a specific product in a specific niche is not something an article can know, which is why our advice is consistently to test it cheaply rather than to trust anyone’s confidence about it.

When we get something wrong

We correct it in the article itself rather than quietly editing around it, and the correction says what changed. The benchmark removal described above is the clearest example: it affected several published pieces and an email that had already gone out, and fixing it properly meant changing all of them rather than the most visible one.

We’d rather be publicly wrong once and correct it than build on a number we can’t defend. If you spot something, tell us — that’s a genuine invitation, not a formality.

Who writes this

Articles are published under PlainFunnel rather than a personal byline. That’s a deliberate choice, and not an attempt to hide: a name and a photo would be doing work that we think the evidence should do instead. A byline doesn’t make a fee table correct, and a founder photo isn’t proof of anything at all in a subject this full of invented credentials.

We sell digital products, which gives us an obvious commercial interest in this topic. How that works, including affiliate relationships, is set out on our disclosure page. Everything here is educational and general — not financial, legal or tax advice.