Shopify or PrestaShop: which e-commerce platform for a French catalogue in 2026?

The comparison doing the rounds sets a monthly subscription against “free” software. It is misleading, and it is the first thing we correct at the framing stage.

We do not publish figures here: they move too often to stay accurate, and a wrong price in a comparison discredits everything else. What counts is the structure of the cost, and that is stable. On one side, a smooth, predictable subscription, to which are added the paid extensions and, depending on the payment solution chosen, a commission on transactions. On the other, no publisher subscription, but hosting to size, modules to buy, and above all human maintenance time.

Over three years, it is that last line that decides. It is invisible at the moment of choosing because it appears on no quote, and yet it is what opens the gap. So the right question is not how much the tool costs, but how many days a year somebody will have to devote to it, and at what rate.

An online shop is not a brochure site: it takes payments, it stores customer data, and it breaks down at expensive moments.

With the hosted solution, technical maintenance leaves your scope. Security fixes, availability during a traffic peak and compatibility between components are the publisher's problem. In exchange you inherit their limits and their rhythm.

With the self-hosted solution, you keep control, which means you carry everything. Every module added is one more dependency, and an unmaintained dependency becomes a vulnerability. This model works very well with somebody competent, available and long-term. It degrades fast as soon as that person changes role.

So the question to settle at the framing stage is not technical, it is organisational: who, by name, looks after it a year from now?

This is the ground where a generic comparison becomes useless, because the constraints are local.

Both handle VAT, reduced rates and invoicing, but not with the same effort. On the hosted solution, standard cases are covered natively and particular cases go through third-party applications, with the dependence that creates. On the self-hosted solution, everything can be configured, provided you have somebody to configure it.

Two points deserve to be settled before choosing, not after: whether your invoices comply with the obligations in force, which change, and how order data flows back into your accounting. A shop that forces the sales to be re-entered every month costs more than it looks.

We do not give a date here for when regulatory changes take effect: their timetable has already slipped several times, and a wrong deadline in a comparison turns against whoever wrote it.

The number of product references is almost never the limiting factor. Both hold wide catalogues without difficulty.

What separates them is the business logic. Sliding-scale prices per customer, packaging by batch, availability per warehouse, negotiated prices, particular delivery rules: as soon as selling obeys rules that are your own, the hosted solution calls for workarounds, and those workarounds pile up.

The self-hosted solution does not have that limit, since the code belongs to you. The price of that freedom is development, and development has to be maintained.

Our marker: if the shop sells a standard catalogue at a listed price, customisation will never be the subject. If selling involves rules you have to explain to a newcomer for more than ten minutes, the subject is already there.

A shop handles payments and personal data. So it is a target, and it is one permanently, not only when business is good.

On the hosted solution, fixes are applied without your having to know. It is this model's main argument, and it is a real one.

On the self-hosted solution, the fixes exist, but somebody has to apply them. The risk does not come from the software, which is well maintained, it comes from the gap between a fix being published and being installed at your end.

If you choose this model, maintenance has to be contracted from the start, with a frequency, a named owner and verified backups. A backup that has never been restored is not a backup, it is an intention.

A very small company starting out, with a standard catalogue, no in-house technical skill and a wish to sell quickly: the hosted solution. The launch time and the absence of maintenance win by a wide margin, and the tool's frame is no hindrance at that stage.

A mid-sized company scaling up, with a large catalogue but classic selling logic: the hosted solution remains the best ratio of time to total cost, as long as the business rules fit within the frame. The switch is decided on the workarounds, not on turnover.

A team bringing things in-house, with real business logic, sovereignty constraints on the data or an information system to integrate with: the self-hosted solution becomes relevant again, provided maintenance is a budget line you accept and not a hope.

If you hesitate, start hosted. Moving from a constrained model to a free one is a project; the other way round is a migration forced on you, generally decided in a hurry.