282 static pages, zero backend: a 19-platform ecommerce fee calculator on GitHub Pages
A 6.5% platform fee is rarely 6.5% of an order. Add 2.9% + 30ยข card processing and a 20ยข listing fee and a $30 order loses about 10%.
Sellers compare headline rates, but what you actually keep depends on the price you are charging. So I generated a site that answers one version of that question per page: 19 platforms (marketplaces, resale apps, storefronts, payment processors) ร 12 price points from $10 to $500, plus head-to-head comparisons and category pages. 282 static files, no server, no database, no client-side routing.
Live: https://gkdlwjavm9-cpu.github.io/ecommerce-fee-calculator/
Why generate pages instead of shipping one calculator with a dropdown
Two reasons, one about search and one about honesty. Search intent here is specific. People do not search for "ecommerce fee comparison tool", they search what the fee is on a $25 Etsy item. A page can match that exactly; a form cannot.
The honesty reason matters more. On one calculator page the assumptions are invisible - a footnote about spreading a monthly plan over N orders is easy to miss. On a per-price page the fee lines sit next to the payout: commission, processing, listing, plan share, and the resulting effective rate. If the reader disagrees with an assumption, they can see what to change.
The data model
One record per platform, with the source URL and the date the rate was read. Roughly the shape of one (simplified):
{
"platform": "etsy",
"fees": [
{ "type": "percent", "rate": 0.065, "on": "item+shipping" },
{ "type": "percent", "rate": 0.03, "plus": 0.25, "on": "item+shipping+tax" },
{ "type": "fixed", "amount": 0.20, "when": "listing" }
],
"monthly_plan": 0,
"source": "https://www.etsy.com/legal/fees/",
"checked": "2026-11-11"
}
The generator expands that into HTML. The fee engine is small; the hard part was deciding what belongs in the record.
The modelling choices that change the answer
The arithmetic is trivial; the assumptions are not. Three of them move the ranking more than any fee difference:
- Monthly plans. A $39/month storefront plan is $0.20 per order at 200 orders and $39.00 at one order. The site spreads plans over 200 orders by default and prints that assumption on the page, because at one order every storefront looks worse than every marketplace.
- Fulfilment. Amazon FBA carries a $4.15 fulfilment fee on a $50 item. That fee, not the referral percentage, is what pushes it to a 23.7% effective rate.
- Whose fee it is. Vinted charges sellers nothing and bills the buyer at checkout. A 0% row is technically true and practically misleading, so the page says who pays.
From effective rate to break-even ROAS
- Effective rate = every fee on one order รท order price.
- Net profit per order = revenue kept โ product cost โ shipping โ fees.
- Break-even ROAS = AOV รท net profit per order.
That last line is why the fee work was worth doing. At a 44.5% net margin a $50 product breaks even at a 2.25x ROAS, so a campaign sitting at 2x that looks respectable in an ads dashboard is losing money on every order.
The same fee model sits behind a separate margin calculator: https://gkdlwjavm9-cpu.github.io/ecommerce-margin-calculator/
The static plumbing that actually matters
sitemap.xmlwith 281 URLs, submitted per property in Search Console, plusrobots.txtand.nojekyll- without that last file Pages runs the output through Jekyll and quietly drops paths starting with an underscore.- JSON-LD on every page:
WebPage,BreadcrumbListandFAQPage. The FAQ block earns its keep, because the questions sellers ask are the ones already answered on the page. - Internal links from each platform page to its neighbouring price points, its comparisons and its category page. 282 orphan pages rank for nothing.
- A source line with a checked date under every fee figure. Rates move, and a page without a date is a liability.
What I would do differently
- Fewer price points, more per-platform quirks. 228 fee pages of near-identical template is a lot; the pages worth having are the ones where a platform does something strange (eBay charging its percentage on tax, Depop on shipping).
- Put every assumption inside the table, on every page, not just on the hub.
The site is free and needs no signup: https://gkdlwjavm9-cpu.github.io/ecommerce-fee-calculator/ - every rate was read from the platform's own fee page on 11 November 2026, and each figure links to its source.
Disclosure: there is a paid Excel workbook version linked at the bottom of the site, and the generator and pages were written with an AI coding assistant, with every rate checked against the platform's own page.
Comments
No comments yet. Start the discussion.