The two headline rates are identical to the cent, and the number that decides the choice sits three rows below them

For most of last year I answered this question with a shrug and a sentence. Both charge 2.9 per cent plus 30 cents, so pick whichever you like the dashboard of. The rates part of that was correct, which is what made the rest of it such bad advice, because the rate is not where the money goes.

Shopify publishes a second number next to the first one and it is easy to read past. It is called the third party transaction fee, it applies when you take payments through anybody other than Shopify Payments, and on the entry plan it is 2 per cent of every order on top of whatever your processor charges.

The toll is the whole comparison. So shopify payments vs stripe for a growing store is not a comparison of 2 rates. It is a comparison of 1 rate against 1 rate plus a toll, and the toll is set by which subscription you are on.

I found it the slow way, in August. A merchant sent me a Shopify invoice asking what a line she did not recognise was for, and it took me most of an afternoon, while I had the invoice open beside the pricing page, before the 2 documents agreed with each other. The line was not an error and it was not new. It had been sitting in the plan comparison the whole time, in a row I had scrolled past for a year because I was looking at the row above it.

That row had been in the plan table for as long as the table has existed. Nothing here is hidden, nothing is disclosed late, and the page says exactly what it does. I simply read the number I had gone there to read and stopped.

The published numbers, in one place

Shopify prices the card rate and the toll together, by plan, and both move at once.

Basic costs 29 dollars a month, cards run from 2.9 per cent plus 30 cents, and the third party transaction fee is 2 per cent; Grow costs 79 dollars with cards from 2.7 plus 30 and a 1 per cent fee; Advanced costs 299 dollars with cards from 2.5 plus 30 and a fee of 0.6 per cent. Plus starts at 2,300 dollars a month, the card rate is given no number at all and appears as “Most competitive rates”, and the third party fee falls to 0.2 per cent.

In person manual entry sits at 3.5 per cent plus 10 cents on the first 3 plans and does not improve as you climb. USDC is priced at the same rate as cards on each plan.

Stripe prices the same card differently in shape rather than in size. Its page offers what it calls simple, pay as you go pricing with “no setup fees, monthly fees, or hidden fees”, and then quotes 2.9 per cent plus 30 cents per successful transaction for domestic cards. There is no subscription attached to it and no plan to choose.

Shopify's own description of the toll is short enough to quote in full, and it is the sentence a merchant should read before picking a processor: “Third-party transaction fees may apply when you use a third-party payment provider (instead of Shopify Payments). The rate depends on your plan.” The same page states that “There are no setup fees on any of our plans”, which is true and is about something else.

Read those 2 paragraphs together and the shopify payments rate by plan on Basic is exactly Stripe's headline rate. They are identical to the cent.

What the toll does to the arithmetic

Put Stripe inside a Basic store and you are paying 2.9 per cent to Stripe and 2 per cent to Shopify. That is 4.9 per cent plus 30 cents on every order, against 2.9 plus 30 for the same order run through Shopify Payments.

The arithmetic below is mine rather than either company's, and it is simple enough to check on a phone. Moving from Basic to Grow costs 50 dollars a month more and cuts the third party gateway fee by 1 percentage point. Those cancel out at 5,000 dollars a month of volume through the outside processor. Below that the cheap plan wins. Above it the upgrade pays for itself. It then keeps paying, every month, quietly.

Grow to Advanced costs 220 dollars more and cuts the fee by 0.4 of a point, which balances at 55,000 dollars a month. That is a much larger jump than the first one, and it is the number I would want in front of me before anybody says the word Advanced.

I ran those divisions twice because the first result looked too tidy, and there is nothing clever in either of them: it is the extra subscription cost divided by the percentage point it buys back, done on the back of the invoice that started all this.

The pattern in that table is worth naming, because it is the whole design. The subscription buys down the toll and the card rate at the same time, so the more of your selling happens outside Shopify's own processor, the more the subscription is really a payments product wearing a software price. Whether that is a good deal for you depends entirely on where your volume sits, and the table cannot know that.

Where Stripe's price actually moves

The headline is 1 number and the invoice usually is not, because Stripe publishes its surcharges separately and they stack on the same transaction.

Manually entered cards add 0.5 per cent, international cards add 1.5, and currency conversion, where it is required, adds another 1 on top of both. A store selling to buyers abroad and pricing in their currency can therefore be paying 2.9 plus 1.5 plus 1, which is 5.4 per cent plus 30 cents, and none of those 3 lines is hidden. They are simply on a different part of the page from the number people quote.

Away from cards the shape changes completely. This is where the real money is. ACH Direct Debit is 0.8 per cent with a 5 dollar cap, so a 10,000 dollar invoice costs 5 dollars to collect rather than 290. Klarna is 5.99 per cent plus 30 cents. Stablecoin payments are 1.5 per cent of the dollar amount.

Disputes are priced as events rather than as a rate. Stripe lists 15 dollars “for each dispute you receive” and another 15 “for each dispute you respond to manually”, and its automated evidence service charges 30 per cent “of the disputed amount for each dispute you win”. That last one is a fee that only appears when you get money back, which is an unusual shape and worth understanding before it surprises you on a statement.

I want to be careful about what that identity does and does not mean. It does not mean the 2 processors are the same product, and it does not mean a merchant's effective rate will match on both. It means the comparison people usually run, which is a comparison of headline rates, has no winner at all on the entry plan, so anybody who reaches a conclusion from it has reached it from somewhere else without noticing.

An aside about the word from

Shopify writes its card rates as “from 2.9%” and “from 2.7%”. The word from is doing something and the page does not say what.

My instinct is that it covers card types and regions that price above the headline, which is how every acquirer I have looked at behaves, but the table behind it is not published and I am not going to reconstruct it from a handful of merchant statements. It has nothing to do with the toll, which is stated as a flat percentage with no from in front of it. Anyway, back to the comparison.

None of this makes either company the wrong choice. It makes the choice a different question from the one people ask at the start, which is usually a question about a rate and should be a question about where the volume sits and what shape it arrives in. A store selling one product to one country has almost nothing in common, at the level of an invoice, with a store selling to buyers across a dozen currencies, and the same pair of processors will produce completely different statements for the two of them.

The thing that is not priced at all

Shopify's built in fraud analysis carries a footnote on the pricing page. Four words long. “Requires Shopify Payments”, set in smaller type under the comparison table.

That is not a fee, so it does not appear in any arithmetic above, and it is the item I would think hardest about. A merchant moving to an outside processor loses a screening layer at the same moment it starts paying 2 per cent extra for the privilege, and the cost of that loss shows up later as chargebacks rather than sooner as a line on an invoice.

I cannot put a number on that loss, and 0 is certainly the wrong one. Nobody publishes the fraud rate difference between stores using the built in analysis and stores that are not, and I would be inventing a figure if I offered one. I suspect it matters most for the merchants least able to absorb it, meaning newer stores on the entry plan selling things that get bought fraudulently, but that is a guess and I would not defend it in front of anybody with real data.

I find it hard to write about that footnote without irritation, and the irritation is not really at the footnote. It is at how much of this decision turns on things that are stated plainly and read by almost nobody, including me, for a year, in public, while answering the question for other people.

Migrating gateway without downtime

The honest version of migrating gateway without downtime is that the switch itself is quick and the risk sits either side of it. People plan the switch and forget the edges, and the edges are where the damage is, because nothing about them announces itself on the day.

Both providers are live the moment the credentials are saved, so the window where nothing works is measured in seconds. What is not measured in seconds is everything attached: saved cards do not travel between processors by themselves, subscriptions built on stored tokens have to be recreated or migrated deliberately, and any refund you might owe on an old order has to be issued through the processor that took the original payment.

So I would keep the old processor connected and unpaid for a full refund window rather than closing it on the day of the switch. That is a habit rather than a rule, and I hold it because I have watched a merchant discover on day 4 that a 900 dollar refund had nowhere to go.

What I would actually check

Your volume through the outside processor, monthly, in dollars. That single number decides the plan question by itself against the 5,000 and 55,000 thresholds above.

The share of your orders that are international or need currency conversion, because Stripe's 1.5 and 1 point additions land only on those and a store with 5 per cent foreign orders is a different business from one with 60.

Whether any part of your volume could be ACH instead of cards, because 0.8 per cent capped at 5 dollars against 2.9 per cent uncapped is the largest single difference on either page and it has nothing to do with which company you pick.

What I cannot tell you

What Plus actually costs per transaction. Shopify does not publish the number and I am not repeating one I have heard.

Whether Stripe would price you below its published rate. It does negotiate at some size, that size is not published, and I have no source I can point you to for the threshold.

What I got wrong last year sat 3 rows below what I got right. I compared the 2 headline rates, found them identical, and told people the choice did not matter much. The 2 rates are identical. The choice is decided by a second number printed 3 rows below the first one, and I had never read it.