Selling online doesn't make your site any better
We classified every company by how much it depends on its site to sell: those that sell right there, those that use it to get enquiries, those that serve customers at a physical location, and those that only publish company information. Then we cross-referenced that against the technical state of each site.
All four groups fail Core Web Vitals at almost the same rate. Between the company that sells through its site and the one treating it as a brochure there are two and a half points, less than the spread inside either group.
| How the site is used | Fails Core Web Vitals | No structured data |
|---|---|---|
| Sells online | 63.3% | 27.4% |
| Generates enquiries | 59.1% | 57.0% |
| Serves walk-in customers | 64.9% | 51.3% |
| Company brochure | 60.8% | 54.3% |
The one column where the online seller comes out clearly ahead is structured data, and there's no credit to claim there: Tiendanube and Shopify generate it on their own. What the platform hands you is done; what requires someone to make a decision isn't.
Speed is measured with users, not with a score
There are two ways to measure a site and they give different answers. Lighthouse simulates a load on a reference machine; the Chrome UX Report collects what the site actually took across the last 28 days of visits. The first is a rehearsal, the second is the performance.
Across the 1,472 sites with enough traffic to have field data, the most common problem isn't the one usually named.
| Field metric | In the red | What the visitor feels |
|---|---|---|
| LCP | 46.3% | Content takes a while to appear |
| INP | 30.4% | You tap something and nothing happens |
| CLS | 17.1% | Content jumps around while loading |
The median Argentine LCP is 2,379 milliseconds, just under the 2.5-second threshold. The typical site isn't broken: it's right at the edge, and any new uncompressed image pushes it over. The median homepage already weighs 3.1 MB.
The platform matters more than the budget
Split by company size, the result is flat. A large company does not have a faster site than a one-person business, and the four sizes fit inside a point and a half.
| Company size | Fails Core Web Vitals |
|---|---|
| Micro | 67.1% |
| Small | 61.5% |
| Mid | 61.6% |
| Large | 62.6% |
Spending more doesn't buy speed. By platform there are real differences, and not all of them point the way you'd expect.
| Platform | Fails Core Web Vitals | No structured data |
|---|---|---|
| WordPress | 70.4% | 32.5% |
| Next.js | 68.2% | 51.8% |
| Webflow | 41.1% | 54.3% |
| Tiendanube | not enough data | 10.0% |
| Wix | not enough data | 0% |
WordPress leading the slow sites is predictable: plugins pile up over years and nobody removes them. What isn't predictable is Next.js landing two points behind, given it's the stack people choose precisely for performance.
On structured data the order flips. Wix has not a single site without it and Tiendanube almost none, because both generate it automatically. Custom-built sites are missing it half the time, because there someone has to remember to add it.
What breaks because nobody looks
Beyond speed there's a layer of problems that require rebuilding nothing. They're settings someone configured once, or never configured, and that have stayed that way since.
- 50.7% have no h1, or more than one: the heading that tells Google what the page is about
- 32.2% declare no canonical, so a page can compete against itself
- 26.6% publish no sitemap.xml
- 24.9% have no search description, so Google writes one for them
- 22% never unified www with the bare domain: two addresses for one site
- 14.1% return an error page with a 200 status code, and search engines index it as content
- 30.2% have no analytics tool installed at all
The last one is the most expensive of the cheap ones. Three in ten companies have no way of knowing how many people reach their site, where they come from or what they do there, which means they can't notice any of the other six problems either.
How this was measured, and what it doesn't say
The universe is 2,609 websites of companies based in Argentina, deduplicated by domain. The three sources are the Chrome UX Report for field data (75th percentile across 28 days of real users), Lighthouse on a mobile profile for lab data, and the same engine that powers this site's public checkup for everything read from the HTML.
Every percentage is calculated over the sites where that data point is actually measured, not over the total, which is why denominators change from table to table. Field data exists only for the 1,472 sites with enough traffic: a site with no visitors has no Core Web Vitals to measure. A missing field is an unmeasured field, never a problem found.
Two caveats that cut against the easy thesis. First: Argentina is not worse than the rest. Across the full database, including companies from Mexico, Spain and the United States, the same indicator comes out at 61.3% against Argentina's 62.8%. If you came expecting to read that Argentine websites are a disaster compared to everyone else, the data doesn't support it.
Second: this measures what a machine can read on its own. Nothing here says whether the site sells, whether the design matches the brand, whether the message lands or whether the form works. A site can score green across the board and still not produce a single enquiry.
How your site compares
- 01Paste your domain into PageSpeed Insights, which is free: the top of the page tells you whether it passes with real visits, and that is the number that counts
- 02Open it on your phone on mobile data: if the headline and main image take more than 2.5 seconds to show up, you are in the 46.3%
- 03Tap the menu on your phone: if you feel a lag between the tap and the response, Google measures that and takes points off
- 04Ask whoever runs your site whether it has structured data, which is what lets Google show your address, your prices or your reviews: 52.2% of sites do not have it
- 05Type your domain with www and without www: one has to redirect to the other, not serve the same site at two different addresses
- 06Add anything after your domain (yoursite.com/asdf): you should get your own error page, with your design and a way back
- 07If you cannot say how many people visited last month, start there: without that number no change can be judged
Save this list and run it against your own site.
Frequently asked questions
What are Core Web Vitals?
They're the three metrics Google uses to measure a page's loading experience: LCP (how long the main content takes to appear), INP (how long it takes to respond when you tap something) and CLS (how much content jumps while loading). A site passes when all three are green for 75% of its visitors.
Where does this data come from?
From three public, verifiable sources: the Chrome UX Report, which publishes what real users' browsers measured; Lighthouse, Google's auditing tool; and an analysis of each site's HTML using the same engine behind the free checkup on this page. No company was contacted or named.
Why aren't the companies named?
Because the goal is to describe the state of the market, not to point at anyone. The percentages say what the lists would say, without exposing companies that never asked to be measured.
Does a high PageSpeed score mean my site is fine?
It means it loads well, and only in simulation. 89.9% of the sites measured score under 90 on Lighthouse, but that number is a rehearsal on a reference machine: what counts is field data, which can be worse or better. And neither one tells you whether the site sells.
How often is this study updated?
The measurements are re-run and the date above shows the latest one. Since September 2026 a monthly snapshot of every indicator is stored, so future versions will be able to show the trend as well as the state.