What a sub-second load time actually does to conversion: our numbers, then the research

Cutting a page from several seconds to under one does two measurable things: fewer people leave before the content appears, and search sends you more of them. On the one migration where we have a full year of before-and-after, the site went from a plugin-heavy build to a static one, and organic unique visitors rose 5x with 20x the pageviews, on zero ad spend. That is our number.

The research further down explains why one migration is worth reading as a pattern.

Our number, in context

We rebuilt Fabrika Sumy, a textile group’s site, from a builder-style site onto a fully static Astro build. Server response is now 0.17 s, measured cold from the EU. In the twelve months after launch, with no paid traffic at any point:

• 5x the unique visitors from organic search, year on year
• 20x the pageviews
• orders through the site went from occasional to regular

We did not add content marketing or run a campaign in that window. The change was structural: the pages stopped making the visitor and the crawler wait.

5x visitors, 20x pageviews is what Fabrika Sumy’s rebuild brought in a year, with zero ad spend and no content marketing.

Two things happen when a page is that fast. A person who clicks from a search result sees the answer before they lose patience, so they stay. And a crawler that can fetch a hundred pages in the time it used to fetch ten indexes more of your site, more often.

What does the research say?

You do not have to take one project’s word for it. The pattern has held across large studies for over a decade:

• Google / SOASTA, 2017. Analyzing mobile sessions, they found the probability of a bounce rose 32% as page load went from 1 s to 3 s, and 90% from 1 s to 5 s. (source)

32% more bounce at 3s versus 1s, 90% more at 5s: Google and SOASTA’s 2017 mobile session data.

• Deloitte, “Milliseconds Make Millions”, 2020. A 0.1 s improvement in mobile site speed lifted retail conversion by 8.4% and average order value by 9.2% across the sites studied. (source)
• Google, Core Web Vitals. Since 2021, loading, interactivity and layout stability are a ranking signal. Slow is no longer only a UX problem; it is a visibility one.

None of these say “fast site = more sales” as a slogan. They say the curve is steep near the top: the gap between 1 s and 3 s costs more than the gap between 5 s and 7 s, because the first few seconds are where people decide whether the page is worth waiting for.

Where do the seconds actually go?

Not to the language or the hosting. To the work left to do when the visitor arrives:

The fix is rarely “compress the images”. It is removing the things the page is blocked on: render-blocking scripts, third-party embeds, fonts that arrive late, a layout that shifts while it loads.

How do you actually get there?

1. Measure the page that matters: usually your slowest landing page over the homepage. Run it through PageSpeed Insights and favor the field data over the lab score.
2. Cut what blocks the first paint. Defer scripts, self-host fonts with font-display: swap, lazy-load anything below the fold.
3. Serve finished HTML where you can. A brochure or catalog site does not need to be assembled on every request. Static generation or full-page caching removes the server from the critical path.
4. Hold the line after launch. Speed regresses one added tag at a time. A performance budget in the build catches it before a visitor does.

The ranges for a rebuild are on our pricing page, and what we build either way is on the websites page. If you want to know whether your site has a speed problem worth fixing, send us the URL of the page that matters most. We will tell you where the seconds are going and whether it is worth the work.