Faster pages, fewer people lost on the way

Every performance vendor quotes the same conversion statistic. Almost none of them state what it actually measured, and multiplying it out gives numbers that cannot be true. Here is what we measured on our own audit base, the benchmark as its source states it, and the arithmetic we deliberately do not do.

No signup. See your before/after PageSpeed score in about a minute.

Measured, not modelled

What we measured

Our own audits — the original page against the same page served through WebSpeed. The sample and the date are part of the result, not small print.

45 audited sites that cleared the publishing gates measured 2026-08-04

median Speed Index before 6.6 s
median Speed Index after 4 s
median time removed 2 s
sites where it got slower 7

What this does not say. We measured seconds, not sales. The benchmark beside these numbers is Akamai’s retail study, and it reports a MARGINAL effect — about 7% of conversions per 100 ms at the margin. It does not license multiplying that out: taking the two seconds we typically remove and scaling the figure would predict a conversion gain of well over 100%, which is obviously false, and it is the single most common way this statistic is misused. Treat it as direction, not as a forecast. Your own conversion rate depends on price, product, trust and traffic mix — speed removes one obstacle among them, and the size of that one is specific to your funnel.

Someone else’s measurement

What the industry benchmark says

Conversion +7% per 100 ms you cut Akamai / SOASTA, 2017

Quoted in the unit its source used, linking the page that states it. An industry benchmark is a direction, not a forecast for one site — the per-site numbers are the audits above, and yours is a free check away.

What WebSpeed does about it

Each one is independent and can be switched off for your domain — nothing here is all-or-nothing.

Fill the screen sooner

Speed Index measures how quickly the page visibly fills, which is closer to what a visitor experiences than any single milestone. We move the work that blocks painting out of the way — stylesheets that gate the first frame, scripts executing mid-render — so the page looks ready earlier rather than merely finishing earlier.

SI Render-blocking resources Critical CSS

Put the thing they came for first

On a product or landing page, one element decides whether the visit continues: the hero, the price, the form. We identify it and give it priority instead of letting it queue behind assets far below the fold.

LCP fetchpriority Preload scanner

Stop the page moving under their thumb

A layout that shifts as it loads produces mis-taps — on a checkout or a call-to-action, a mis-tap is an abandoned action. We reserve real space from measured image dimensions, so nothing jumps once it arrives.

CLS Lazy loading

Keep it fast when your origin is not

Cached delivery means a slow or briefly unavailable origin does not become a blank page for the visitor. The worst conversion outcome is not a slow page; it is a page that never arrives.

TTFB Reverse proxy

The statistic everybody quotes

If you have researched site speed at all, you have met a version of this sentence: every 100 milliseconds costs you 7% of conversions. It comes from Akamai and SOASTA’s 2017 retail study, and it is a real measurement of real shopping traffic.

It is also the most abused number in this industry, and the abuse is almost always the same shape. The study measured a marginal effect — what happens to conversion around the margin of 100 ms, on retail sites, with that traffic mix. It is quoted as though it were a rate you can multiply. Take the two seconds we typically remove from a mobile page, multiply by twenty, and the claim becomes a conversion gain of over 140%: more buyers than visitors. Nobody believes that when it is written out, and yet the arithmetic behind a great many “up to X%” promises is exactly this.

So we do not do it. The benchmark appears on this page in the unit its source used, with a link to the release that states it, and the number we put our name to is the one we measured ourselves.

What we measured

The figures above are Speed Index on mobile, across the sites in our audit base that cleared the publishing gates — both devices held up, the proxy genuinely served the page rather than redirecting back to the original, the render was not degraded into something that scores well because half of it never loaded.

Speed Index rather than a single milestone, because it tracks how quickly the page visibly fills. That is the closest lab measurement to the thing a visitor is actually reacting to when they decide whether this site is working or wasting their time.

The count of sites where it got worse is printed beside the median deliberately. It is not zero. A delivery layer that defers work sometimes defers the wrong thing, and on a handful of sites in the sample the visible fill took longer afterwards. That is a real property of what we do, it shows up in your own before/after, and a page that hid it would be describing a different set of sites than the one we measured.

What we are actually claiming

Only this: on most of the sites we audited, a couple of seconds of waiting can be removed without touching what the site says or how it is built — and you can see the number on your own pages in about a minute, for free, before deciding whether it is worth anything to you.

Whether those seconds turn into orders is your funnel’s question, not ours. We would rather be the vendor that says so than the one with the more impressive graph.

Conversion — questions

How much will my conversion rate go up?

We don’t know, and neither does anyone quoting you a number. What we can show is the measured part — how many seconds come off your pages — on your own site, before you commit to anything. The step from seconds to sales depends on your funnel, and any vendor turning our measurement into your revenue figure is extrapolating a marginal retail statistic well past what it supports.

Why quote the Akamai figure at all, then?

Because it is the best-evidenced link between load time and buying behaviour that is publicly checkable, and because leaving it out would not make the question go away. We quote it in the unit its source used — 7% per 100 ms — and link the release that states it, so you can read what was actually measured rather than the version that circulates.

Does speed matter more on some pages than others?

Yes. It matters most where a visitor arrives cold and decides quickly — paid landing pages, product pages, anything reached from search or social. It matters least on pages people are already committed to, like an account area they log into daily. If your slow pages are the second kind, the honest expected gain is small.

What if speed isn’t my problem?

Then the check will show it, and we would rather you found out in a minute than after a subscription. A site already painting quickly has little to gain here, and the report says so instead of inventing a recommendation.

Pricing

Pick how hands-off you want it

The before/after check is free to run; it isn’t one of these plans. Choose a plan when you’re ready to make the speed-up permanent — or buy the report once and apply it yourself.

SaaS subscription

Growth

from €79 · /mo
For several sites at once
14-day free trial
  • Everything in Fast, across multiple sites
  • Priority cache — your pages stay warm
  • Support with a named contact
  • One invoice for the whole portfolio
Request access
Agencies and multi-site owners.
SaaS + setup service

Custom

from €149 · setup, then monthly
Hand-tuned for your site
  • Everything in Growth
  • A config built by hand for your site
  • What only you can change, written up for your team
  • Set up, measured and maintained by us
Get a plan for my site
We look at your site by hand first. One-time setup, then a monthly plan.

Subscriptions start with a 14-day free trial (Fast and Growth), then run for a minimum of 6 months — you commit only after the trial. Fast is priced by monthly visits; indicative launch pricing.

Downloadable report

Not ready for a subscription?

Buy the Audit once and apply it yourself — the same findings the pipeline acts on, written up as instructions.

  • The same fixes Fast applies, written up as a guide you run yourself
  • Server settings, code changes and ready-to-paste head snippets
  • Before/after PageSpeed from the free check (PDF / HTML / MD)
  • Re-issued free for a year as your site changes
€108 · one-time
Priced at a year of Fast (€9 × 12)
DIY: you apply it
Buy the audit See an example report →
Not useful? Email us within 14 days and we refund it. A document, not a service — you apply it; Custom is the same work done for you.

See your own before / after

Paste your URL for a free PageSpeed comparison on mobile and desktop. No signup, about a minute.

No signup. See your before/after PageSpeed score in about a minute.

Other outcomes: Agencies & studios , Search traffic , Bandwidth & weight , Carbon footprint

By platform: WordPress , Duda , WooCommerce , Wix , Tilda , Bitrix , Squarespace