The full table
These are scripts we often see on contractor Google Ads landing pages. The numbers come from third-party-web, an open dataset by Patrick Hulce, version 0.30.0, published in September 2026. It is built on the HTTP Archive crawl of March 1, 2026. Sorted from heaviest to lightest.
| Script | What it does | Pages using it | Average script execution time |
|---|---|---|---|
| YouTube | Video embed | 1,300,771 | 6,297 ms |
| Wistia | Video embed | 36,349 | 5,003 ms |
| Calendly | Booking | 13,872 | 4,305 ms |
| Vimeo | Video embed | 182,269 | 4,099 ms |
| Zendesk | Chat and support | 111,484 | 2,157 ms |
| Intercom | Chat | 58,316 | 1,433 ms |
| HubSpot | Tracking, forms, chat | 313,527 | 1,226 ms |
| LiveChat | Chat | 62,860 | 1,116 ms |
| Google Tag Manager | Tag container and Google tag | 15,022,156 | 1,066 ms |
| Hotjar | Heatmaps, recordings | 504,590 | 732 ms |
| TikTok | Pixel and embeds | 437,719 | 578 ms |
| Meta Pixel and embeds | 5,647,056 | 514 ms | |
| Tawk.to | Chat | 176,551 | 440 ms |
| Clarity | Heatmaps, recordings | 1,226,624 | 405 ms |
| CallTrackingMetrics | Call tracking | 15,854 | 282 ms |
| CallRail | Call tracking | 60,963 | 255 ms |
| Google Analytics | Analytics | 5,063,984 | 108 ms |
Google reCAPTCHA is left out on purpose. The dataset only counts it when it loads from recaptcha.net. Google's own reCAPTCHA docs load it from www.google.com, which the dataset groups with other Google code. There is no clean number to report.
How to read the numbers
HTTP Archive tests millions of pages with Google Lighthouse on mobile settings. Lighthouse records how long each script keeps the browser's main thread busy and ties it to the address the script came from. third-party-web adds that up per company and divides by the number of pages that use it. That gives the average above.
What script execution time means on a phone
A phone browser does most of its work in one lane, called the main thread. That lane builds the page, runs scripts, and answers taps. Only one job runs there at a time. Google's Lighthouse docs say that when the main thread is busy, the page can't respond to user input. Lighthouse warns when JavaScript on a page takes longer than 2 seconds to run.
Google's Interaction to Next Paint (INP), a Core Web Vital, measures how long a tap waits for the screen to react. Google rates it good at 200 ms or less and poor above 500 ms.
Three things to keep in mind
- These are averages. Your page can run higher or lower, depending on settings and what else loads.
- Rows group by company. The Facebook row covers the Pixel and Facebook embeds. The HubSpot row covers tracking, forms, and chat. The Google Tag Manager row covers everything from googletagmanager.com, including Google's gtag.js.
- Scripts add up. No single row tells you your page's speed. Call tracking plus chat plus two pixels plus a video is the real cost.
Three ways to load scripts smartly
Load after interaction
Start chat, heatmaps, and video players only when a visitor scrolls, taps, or presses play. Google's Lighthouse docs call a stand-in that loads the real thing on tap a facade.
Delay until the page loads
In Google Tag Manager, fire nice-to-have tags on Window Loaded. Google says that trigger fires after the page has fully loaded, including images and scripts.
Use the vendor's lighter option
HubSpot lets you hold its chat widget until you call it. Intercom has a boot method to control when it loads. Calendly offers a pop-up and a plain link. Hotjar suggests delaying its own script.
Know what not to delay
Call tracking has to swap your number before the visitor taps it. The Google Ads conversion tag ties leads to clicks. Leave those on time and speed up everything around them.
Keep what you actually use
The cheapest script is one that is not there. Ad pages collect leftovers: a pixel from a campaign that ended, a chat tool from a free trial, a second heatmap tool someone tried once. Go down the page's scripts and ask one question about each: who uses this, and for what?
- Brings or tracks leads? Keep it. Load it the right way.
- Nice to have? Delay it until after load or after interaction.
- Nobody can say who uses it? Remove it from the ad page.
- Two tools doing one job? Pick one.
Check your page
The table shows averages across many sites. Your page is its own case. Scan before you change anything, change one thing, and scan again.
ClickMend checks the page your Google Ad sends people to. It gives a 0 to 100 score that blends Google PageSpeed Insights performance (40%), Core Web Vitals (25%), SEO basics (20%), best practices (10%), and accessibility (5%). It names unused JavaScript and other slow parts, so you can see which scripts weigh the page down. It also checks the Google Ads conversion tag, Google Tag Manager, and the conversion linker.
Scan the landing page free, 2 scans a week, no card. For more on testing, read landing page speed test. To catch new scripts after edits, see landing page monitoring.
Frequently asked questions
Which scripts slow down a landing page the most?
In this data, video and booking embeds are the heaviest: YouTube 6,297 ms, Wistia 5,003 ms, Calendly 4,305 ms and Vimeo 4,099 ms of script execution time per page. Call tracking and Google Analytics are among the lightest.
Should I remove third-party scripts from my Google Ads landing page?
Remove the ones nobody uses. Keep the ones that bring or track leads, and load the heavy ones after the visitor interacts or after the page finishes loading.
Why is Google reCAPTCHA not in the table?
The dataset only counts reCAPTCHA when it loads from recaptcha.net. Google's own reCAPTCHA docs load it from www.google.com, which the dataset groups with other Google code, so there is no clean number for it.
Where do these numbers come from?
From third-party-web version 0.30.0 by Patrick Hulce, published September 2026. It is built on the HTTP Archive crawl of March 1, 2026, which tests pages with Google Lighthouse on mobile settings.