Choosing a platform without regretting it

This decision gets more attention than it deserves and still gets made badly, because people compare feature lists when the only things that matter are what it costs at the volume you will actually reach and what happens when you want to leave.

The three kinds, honestly

• Hosted shop platforms. You pay monthly, somebody else handles hosting, updates, security and payment compliance. Fastest to a working shop, and the cost grows with you, sometimes including a cut of revenue.
• Self-hosted software. Cheaper in licence terms, and you become responsible for hosting, updates, backups and the security of a system that holds customer data and takes payments. The hours are the cost, and they are not optional.
• Marketplaces. Somebody else's shop with your products in it. Demand is already there and so is the commission, and you do not own the customer relationship.

For most people starting, hosted wins on the only dimension that matters at the start: it lets you find out whether anyone wants the product, which is the question, rather than whether you can keep a server patched.

Compare total cost at your real volume

Illustrative, because the mistake is comparing monthly fees in isolation. Suppose 100 orders a month at an average of 40, so 4,000 of revenue.

A plan at 30 a month with its own payment processing at 1.5% plus 0.25 costs 30 plus 60 plus 25, so about 115. The same plan with a third-party processor may add a platform transaction fee of a percent or two on top, which on 4,000 is another 40 to 80: the monthly fee again, twice over, for a choice that looks incidental on the pricing page.

Run that at 500 orders before deciding, not at 100. The ranking changes, and migrating later is the expensive outcome this is meant to avoid.

What actually matters in the decision

1. 01Does it support how your customers pay. Card coverage is the easy part; the local methods people in your market actually use are the part that quietly costs sales.
2. 02Does it handle VAT the way you will need it. Charging the customer's country rate, invoicing correctly, and producing figures you can file from. Retrofitting this is painful.
3. 03Does it do returns and partial refunds properly, because in the EU you will be doing them.
4. 04Can you edit the page templates. Product pages and checkout are where the money is, and a platform you cannot change is a ceiling.
5. 05Is it fast on a phone out of the box. Themes with carousels and sliders ship slow, and speed is not something you add afterwards.

What does not matter yet

Multi-currency, multi-warehouse, wholesale pricing, subscriptions, loyalty points, and a mobile app. Every one is a reason to pick a bigger plan and none of them is relevant before the shop has customers. Choose for the next twelve months and accept that a successful shop will migrate once, which is a good problem that you pay for out of profits.

The two hours that settle it

Every platform has a trial and almost nobody uses it properly. Reading the feature list tells you what the vendor is proud of; doing the following tells you what you will live with.

1. 01Add one real product with all its variants, its real photographs and its real description. Variants are where platforms differ most and where the irritation lives.
2. 02Place a real order through the real checkout, on a phone, and refund it. This exercises the three things you will do most and cannot change later.
3. 03Set a VAT rate for another country and issue an invoice. If this is awkward in the trial it will be awkward forever.
4. 04Edit the product page template. Find out whether you can move the add-to-basket control and change what appears above it, because that is where you will want to work.
5. 05Export everything and open the file. This is the leaving question answered with evidence rather than a support article.

If any of those five is unpleasant in a trial with one product, it will be unpleasant in the third year with four hundred. The feature nobody notices in the comparison is how the thing feels to use for the twentieth time.