Static Host UK

How we measured: the same page, built twice

One page built twice, once with Calendly, Formspree, Google Analytics and Mailchimp, once with ours. How we tested it, the results and every source.

On 12 November 2026 we loaded two versions of the same small business page and recorded what each one cost the visitor. This page sets out how we did it, so you can check the figures or run the test yourself.

The two pages

Both pages have the same heading, text and stylesheet. Each adds a booking calendar, a contact form, analytics and a newsletter sign-up.

Page A uses each provider’s published embed code, as their own instructions give it:

  • Calendly’s inline widget, pointed at a public Calendly booking page (calendly.com/acmesales)
  • A Formspree form, with a made-up form ID: the form contacts Formspree only when someone submits it
  • The Google Analytics 4 tag, with a made-up measurement ID
  • Mailchimp’s classic sign-up form, with a made-up list

Page B uses ours:

  • The StaticHost booking widget
  • The StaticHost form widget
  • The StaticHost analytics script, with a made-up site ID
  • The StaticHost newsletter widget

Our booking, form and newsletter widgets are the live ones from be braver now’s website, which runs on StaticHost. They answer only requests from that site’s own domain, so the test script made their data requests itself, with that domain as the origin, as the live page would. Nothing was submitted or booked on either page.

How we loaded them

  • Headless Chromium, driven by Playwright, on a Mac in the UK
  • A fresh browser profile for every load, so nothing came from a cache
  • Two profiles: a desktop at 1440 by 900 pixels with no throttling, and a phone at 390 by 844 on Lighthouse’s slow 4G settings (150 ms round trip, 1.6 Mbps down, 750 kbps up) with the processor slowed four times
  • Three loads of each page in each profile, and we report the middle one
  • After the load event the script waited for the network to go quiet, then three seconds more, so late requests were counted too

What we counted:

  • Data downloaded: every response as it came over the network, headers included
  • Requests, and the separate domains they went to
  • Cookies in the browser at the end, with nothing clicked
  • Load time: the page’s load event

Results

Page APage B
Data downloaded5,011 KB45 KB
Requests3415
Domains contacted, other than the page’s own123: two of ours, and jsDelivr
Cookies set40
Load time, desktop1.4 s0.16 s
Load time, slow phone connection25.6 s0.76 s

KB here is 1,024 bytes. The phone figures for data and requests match the desktop ones within 1%.

Page A contacted calendly.com, assets.calendly.com, dfp.calendly.com, js.stripe.com, m.stripe.com, m.stripe.network, accounts.google.com, www.gstatic.com, fonts.gstatic.com, www.recaptcha.net, s3.amazonaws.com and cdn-images.mailchimp.com. Most of its weight is Calendly’s booking app: 3,285 KB from assets.calendly.com and 1,093 KB from www.gstatic.com. Mailchimp’s form-checking script is 140 KB. The four cookies were _GRECAPTCHA (www.recaptcha.net), __cf_bm and _cfuvid (calendly.com) and m (Stripe).

Page B contacted api.statichost.uk for 18 KB (the booking widget 10 KB, the form 4 KB, the newsletter 2 KB, and the data requests), t.statichost.uk for 1 KB, and cdn.jsdelivr.net for 24 KB. That last one is Altcha, the anti-spam check our form widget loads from jsDelivr, a public code host.

What the test does not show

Page A’s figures are a floor:

  • Our test network blocks googletagmanager.com, so Google Analytics never loaded. The same block stopped Calendly’s cookie banner (cdn.cookielaw.org) and an error-reporting script. A live page would load all three.
  • Calendly’s page had no free slots that day and showed “No openings at the moment” in place of a calendar, so the figure does not include a calendar full of times.
  • With Calendly’s cookie banner blocked, we can’t say which of the four cookies it would have held back until the visitor agreed.

And on both sides:

  • Formspree’s share is zero here, because its form costs nothing until someone submits it. The same goes for submitting our form.
  • Our widgets’ data requests were made by the test script, so their sizes are an estimate: each response compressed as our server sends it, plus 200 bytes for headers. They come to under 3 KB of page B’s 45.
  • One network, one day, one browser.

Prices and terms

Read on 12 November 2026 from each provider’s own pages, in the currency each one shows.

Vercel

GitHub Pages (not in our price table)

Calendly

Formspree

Plausible

  • Starter: $9 a month, one site, up to 10,000 page views a month
  • https://plausible.io/ (read from an archived copy dated 11 November 2026, because our test network blocks plausible.io)

Google Analytics

Mailchimp (not in our price table)

Where each company is based

From its own data processing terms or privacy policy:

Run it yourself

The two test pages and the script are here: same-page-test.zip. The script needs Python and Playwright. Point it at a folder holding both pages and it writes every request, cookie and timing to a JSON file.

get started

Ready to Deploy?

Join Jamstack developers across the UK hosting their static sites with privacy and performance built in. Hosting is free for good, no credit card needed, with Plus features when you want them.