You want to sell online, and everyone has an opinion. The trouble is that most comparisons of the options are written by someone earning a commission on whichever platform they recommend, which is not a reason to distrust them exactly, but is a reason to want a second view.
So, ours, with our own interest declared below. How much an online store costs depends less on your budget than on how unusual your business is: a platform store costs $1,000 to $15,000 to set up well and a few hundred a month to run, a custom storefront on top of a platform is $15,000 to $60,000, and a fully custom build starts around $60,000 and rises with your complexity. The interesting question is not which is cheapest. It is which one your particular catalogue, pricing and stock will still fit in three years.
How much does an online store cost? The three routes
| A platform store | A custom storefront on a platform | A fully custom store | |
|---|---|---|---|
| What it is | A hosted platform with a theme, configured properly: your products, payments, shipping rules and tax set up by someone who has done it before. | The platform still runs commerce, payments and stock, but the shop your customer sees is designed and built by you, talking to the platform behind the scenes. | You own the commerce engine as well: your pricing rules, your checkout, your data model, hosted on infrastructure you control. |
| Illustrative setup | $1,000 to $15,000 | $15,000 to $60,000 | $60,000 to $250,000 and up |
| Illustrative monthly | $100 to $600, plus fees on sales | $200 to $900, plus fees on sales | $500 to $5,000, mostly infrastructure and maintenance |
| Fits when | An ordinary catalogue, standard pricing, standard checkout. Most businesses, for years. | The commerce rules fit the platform but the design, speed or content needs do not. | Your pricing, stock or checkout rules genuinely do not fit any platform, or fees at your volume have outgrown the cost of running your own. |
Setup is the one-off. The monthly figure excludes payment processing and the platform's own percentage, which scale with sales rather than sitting still, and which the next section covers.
The middle route is the one most people have never had explained to them, and it is frequently the right answer. You keep the boring, dangerous, expensive parts on somebody else's infrastructure, payments, PCI scope, stock and order management, and replace only the part your customers actually see. That is what most of our online store work is, and it exists because the choice is usually presented as a binary when it is not one.
The monthly costs people underestimate
The subscription is the number everyone compares and the smallest part of the answer. Here is the rest of it.
- Apps, and this is the big one. Platforms are deliberately spare out of the box, and the missing pieces arrive as monthly subscriptions: reviews, loyalty, subscriptions, bundles, advanced shipping rules, back-in-stock alerts, upsells, better search. Each looks trivial. Half a dozen at twenty to sixty dollars each is several hundred a month before you have sold anything, and it is the single most common reason a store's running cost surprises its owner.
- Apps that charge a percentage as well as a fee. Many combine a modest monthly figure with a cut of the transactions they touch. That structure is fine when you are small and punishing when you grow, which is exactly backwards from what you want. Read the pricing page of every app for the word "per order" before you install it.
- Apps you are still paying for. Some bill outside the platform's own billing, so uninstalling the app does not cancel the subscription. Audit the list twice a year against your card statement, not against the platform dashboard.
- Payment processing. Typically a percentage of each sale in the low single digits plus a small fixed amount, which is the cost of taking money and unavoidable on any route. What is avoidable is the additional percentage some platforms charge on top when you use a payment provider other than their own.
- The theme and whoever changes it. A bought theme is a small one-off. Making it look like your brand rather than the other stores using it is the part that costs, and every future change to it is either a developer or a compromise.
- The connectors. Accounting, shipping labels, tax calculation, warehouse, product feeds for advertising. Each is small and they add up, and this is where our connecting your systems work usually starts.
- Photography and product copy. Not software, and consistently the largest underestimate on the list. A hundred products need a hundred sets of images and words, and no platform has ever written one of them for you.
A more useful way to think about a platform store's running cost is as a percentage of revenue rather than a flat monthly fee, because most of it moves with what you sell. That framing also makes the eventual comparison with a custom build an honest one, since a custom store's costs mostly do not move with sales.
What actually forces a business off a platform
Not ambition, and not a developer's preference. Six things do it, and each one is a specific rule your business already has that the platform was not designed for.
- Trade pricing. Different customers must see different prices, with minimum order quantities, credit terms and account-specific catalogues. Platforms have grown real wholesale features, but the lower tiers cap how many separate price lists you can run, and the workarounds are apps with their own monthly cost and their own edges. If your price depends on who is logged in, in more than two or three simple tiers, this is the most common single reason to look further.
- Stock that is not one box on one shelf. Bundles and kits that must decrement their components, made-to-order items, serial or batch numbers, reservations across several locations, stock shared with a physical shop. The moment the platform's idea of an item stops matching your idea of an item, everything downstream gets harder.
- A checkout that is not a checkout. Deposits, part payment, quote requests where a human must approve, purchase orders, approval chains inside the customer's own organisation, or delivery slots that have to be booked against real capacity.
- A system that must be the master. If your ERP or warehouse system owns price and stock, and the platform must never overwrite it, the integration becomes the project. That is workable on any route, and it is a great deal easier when you control both ends.
- Catalogue scale and shape. Very large catalogues, deep variant structures, or product data with attributes the platform has no field for.
- Arithmetic at volume. When the percentage taken by the platform and its apps exceeds what it would cost to run your own, the decision has made itself. That crossover arrives later than most agencies imply and sooner than most platform advocates admit, and it is worth calculating with your real numbers rather than arguing about.
Every one of those has a platform workaround. The honest question is never whether a workaround exists, it is when the stack of workarounds costs more, in money and in fragility, than the thing they are working around. Two or three of these together is usually the signal.
What it costs to move if you outgrow it
Worth knowing before you start, because it is the cost that makes an early decision feel expensive later.
- What moves easily: products, variants, images, customers, historical orders and your content. These are exports and imports, and they are mostly a matter of care rather than difficulty.
- What needs real work: every old address mapped to its new one. Skipping redirects on a store migration is how businesses lose the search traffic that was paying for the store, and it is entirely avoidable.
- What does not move cleanly: anything that lived inside an app. Reviews, loyalty balances and points, and above all recurring subscriptions, where customers frequently have to re-authorise payment on the new system. If you sell subscriptions, ask about this before you choose a platform, not before you leave one.
- What gets rebuilt: the theme and any custom work in it, which does not transfer.
A migration is priced like a build of the tier you are moving to, plus the data work, and it is usually cheaper than the first build was because the decisions and the content already exist. It is not free, though, and the way to avoid it is to be honest at the start about which of the six triggers above your business already has.
Not for you if
The decision checklist
- Does the price a customer pays depend on who they are, beyond a simple discount code?
- Does one thing you sell consume the stock of several other things?
- Does anything need approving, quoting, or paying for in parts before it ships?
- Is there another system that must own price or stock, and never be overwritten?
- Do you sell subscriptions, rentals, or anything that recurs?
- How many paid apps would the platform version need, and what do they cost together per month?
- At your realistic volume next year, what will the platform and its apps take as a percentage of sales?
- If you had to move in three years, what would you lose?
Yes to none or one of the first five, and a platform store is almost certainly right. Yes to three or more, and you should at least price the middle route. The last three questions are the ones that turn an argument about platforms into an arithmetic problem, which is a much easier thing to decide.
Whichever route you take, budget for the store getting better after it launches rather than being finished at launch. The gains in online retail come from the unglamorous work afterwards: page speed, the checkout, the product pages that were not quite right, all of which is what a continuous improvement arrangement is actually for.