Is your competitor’s site better, or just easier?

What a machine sees there

A product page carries text for the buyer, and sometimes a block of data placed in the page for machines, written in a vocabulary search engines all read — it is called schema.org, and that is the word to give your web agency. That block is the first thing a comparison site, an assistant or an answer engine reads.

Our own measurement: five pages read on 20 October 2026, one per site, taken without sorting — the first address in each shop’s own site map that has the shape of a product page. On one of the five, that first address was the shop’s front page: we had to go down to the next one, and that is the rule that was applied everywhere. One page at a time, at most one a second, with a reader that gives its name and a contact address.

Four things out of five turn up identically everywhere. Only one separates the five shops.

What comes from the platform, not from a decision

Here is the part that should put your mind at rest. Those four lines are not the result of a project, an agency or a budget: they are laid down by the merchant platform, at installation, for everyone at once. Two of the five sites run on the same platform, widely used in France, recorded on 9 October 2026 across the same panel.

In other words, what you admire at your competitor’s, you most likely already have. Easier and better are not the same thing, and by far the largest part of what a machine finds on these five pages belongs to the first.

The one place where the five part ways

One line of the table is not five out of five, and it is the interesting one. On all five pages the reference never appears in the product name: it lives in its field, or nowhere. And on the fifth, the one without that field, the product name is the reference — eight characters, with not one word of description.

That shop does not have a worse site. It has a different catalogue convention, one that works perfectly for a buyer who knows the references and says nothing at all to someone searching for “brass fitting 15”. What separates the five is not a question of means: it is a catalogue decision, and it is within your reach without changing supplier.

What five pages do not tell you

We have to be blunt here, because this is exactly the trap this article is about. Five pages across six sites are not a sample. They describe neither your market, nor those six shops, nor even a second page of any of them. They describe five pages, read on one given day, with an instrument whose protocol is published.

And the instrument itself got it wrong along the way: the very first version of the table announced four out of five on product markup. It was false. The first address in one shop’s site map was the shop’s own front door, not a product page — a page with no reason to carry a block of product data. Corrected, then read again: five out of five. Looking quickly at someone else’s site tells you less than you think, even when a program does the looking.

What you can look at tonight

Take one of your own pages, and one from the competitor who worries you. Three questions, in this order: is the reference a field of its own, or buried in the title? Is the price written in the page, or worked out once the page is open? Is availability declared?

If the answers are the same on both sides — and on the evidence of these five pages, that is the likeliest case — then what separates you from them is not their site. It is what your engine does with your catalogue, and that can be measured: what a machine reads from your pages on one side, and the point where the original search gives up on the other.

And if the answer is “neither”

That is the most frequent outcome of this exercise, and the most useful. Your competitor is neither better nor better equipped: they are served by the same platform, with the same default settings, and they have decided nothing more than you have. What is left to decide is therefore entirely yours, and none of those decisions starts with rebuilding a site.