Shopify or a custom storefront
Shopify, for most brands. If you sell products with prices through a cart, its checkout is better tested than anything built for you, and the money saved buys photography and ads. A custom site pays when the site is part of how the business runs: a catalogue the platform cannot model, or a process that happens on the site. You keep the code either way.
In detail
Shopify and custom, row by row
| Shopify | A custom build | |
|---|---|---|
| Right for | Products with prices, sold through a cart | A catalogue or a process a store platform has no concept of |
| The checkout | Shopify’s, tested at a scale no single build can match | Yours to design, and yours to keep working after launch |
| Time to launch | Weeks | Months, and the scope decides how many |
| Speed | Fast on day one; slows as the apps pile up | Fast on day one; slows when nobody owns it |
| Running cost | A subscription and the apps you keep | Hosting, and a developer’s time whenever something changes |
| Who runs it after we leave | Anyone on your team, after an afternoon | Someone technical, or us on a retainer |
| Who keeps the code | You: the store, the theme and the domain | You: the repository, the hosting and the domain |
| Moving away later | Products and customers export in a morning; the theme does not travel | Everything is yours to move, and nothing moves itself |
When does Shopify win?
Almost always, and we say so on the call even though the custom build is the larger invoice. Shopify wins when your products have fixed prices, your buyer pays through a cart, and the thousand small needs of a store, taxes, shipping rules, returns, the abandoned-cart email, are switches rather than projects. Seyr is that kind of store: structured around what a buyer needs to see, in the order they need to see it, documented and run by the client from the first day. The platform did not make it sell. The order of the page did, and someone has to decide that on purpose whichever way you build.
When does a custom build pay for itself?
When the site is where the work happens, not where it is shown. Frank Zoller Authentic History auctions militaria, and its catalogue is one-off pieces, each with a description, an estimate and a provenance. An AI platform we built writes the listings from the lot photographs against 32,000 past items, and zollerhistory.com with its shop sits at the end of that pipeline, built in an archival direction around the accession card. A store platform has no idea what an accession card is. The build paid because the site is where that record is seen and sold, not because the brand wanted to look different.
How to decide in an afternoon
- 01
List what the site must do
Not what it should look like. Sell products with prices? Shopify. Show a catalogue with a history per item, run a process, generate something? Keep reading.
- 02
Price the platform’s gaps
For each thing Shopify cannot do, ask whether an app does it well enough. If every gap closes with an app, Shopify. If a gap is the business, custom.
- 03
Name who runs it after launch
A store your team can edit on a Tuesday beats a site that needs a developer for a price change.
- 04
Budget the photographs and the ads
Whatever you spend on the site, the pictures and the traffic matter more. If the custom build leaves nothing for either, it is the wrong build.
- 05
Put the ownership in writing
Store, theme, domain, repository, hosting: in your name on the first day and in the contract, whichever you choose.
Speed, ownership and moving later
Both kinds of site are fast when they launch and both get slow, for different reasons. A Shopify store slows one app at a time, each adding its own script to every page, until the product image arrives after the visitor has left. A custom site slows because the person who kept it fast moved on. The fix for the first is a quarterly audit that removes what nobody uses; the fix for the second is naming the owner before you commission the build. zollerhistory.com carries an archive a collector will read for an hour, so its pages are built in advance and its images load as the visitor reaches them, which is the kind of decision that has to be made at the start.
Moving between the two is possible in both directions and tidy in neither. Products, customers and orders export from Shopify in a morning; the theme, the apps and their settings do not travel. Leaving a custom site, you own everything and have to move everything, including the parts nobody documented. Whichever you choose, keep the product data in a form you could hand to the other, and you will never be held by a platform, or by us.
What a build does not include, either way
- The product photography. No site survives supplier images.
- The ads that send people to it, which usually matter more than the platform.
- A fix for the offer. A new site presents a weak offer more beautifully.
- Speed after launch, unless somebody is named to own it.
The questions that follow.
What does a Shopify build cost with Drifted?
The set-up, the theme and the product presentation are priced on the first call once we have seen what you sell. Website development is not part of Standard; the pricing page lists it under the exclusions.
What does a custom site cost?
It is priced to the number under Tailored, after it has been scoped as software. If the justification is a feeling rather than a number, we recommend Shopify, and have done to clients who wanted the larger project.
Can you move my custom site to Shopify, or the other way?
Yes. Products, customers and orders travel; the design and the apps are rebuilt. We say how long before we start, and the data is exported in a form you keep.
Who owns the site when you are done?
You. The store or the repository, the domain, the hosting and the theme are in your name from the first day, and we work with access you can take away.
Do you build on other platforms?
Rarely, and only when your team already runs one. A platform nobody on your side can operate after we leave is a dependency, not a site.
The price is published.
Where this leads.
No deck. No discovery theatre. One metric, one date, in the contract.