The observed request mix
The Web Almanac classifies observed third-party requests by content type. Scripts made up 24.8%, images 19.9%, and the source's ‘other’ category 13.9%. The chart shows those three leading categories; they are a subset of all third-party requests, so the bars do not add up to 100%.
Content type is different from vendor purpose. An analytics provider may send a script and later an image or other request. The chart therefore describes network traffic in the crawl, not the market share of analytics tools or a recommended mix for your site.
| Content type | Share of requests |
|---|---|
| Scripts | 24.8% |
| Images | 19.9% |
| Other | 13.9% |
Scale: 0–100%
Percent of observed third-party requests by content type. Only the three leading categories reported in Figure 3.5 are shown; ‘other’ is the source's category label.
Why a shorter vendor list may still mean more requests
Across all sampled sites, the 2025 chapter reports a median of 83 third-party requests on desktop pages and 79 on mobile pages. It says those medians rose by five requests from 2024, even though the median number of distinct third-party domains fell. One vendor can generate many requests, and a tag manager can load other services after the initial page request.
A request count is an inventory signal, not a performance score. Some requests are small or deferred; others compete with critical content, run expensive JavaScript, or fire before consent. The useful next step is to identify the specific request chain and timing on a page that matters to your business.
What the crawl can miss
The report analyzes third-party content visible to its browser crawl. It does not interact with pages or scroll, so a widget that loads after a click or farther down the page may be absent. The report also treats CNAME-cloaked and server-side routed services as first-party interactions when their third-party traffic is not visible to the browser crawl. The authors describe their measurement as a lower bound on third-party prevalence.
That boundary matters for marketing analytics. Moving a tag to a server-side setup can change what a crawler sees without removing the downstream vendor, its data flows, or its business purpose. Review the actual implementation and data destinations instead of treating fewer visible browser domains as proof that measurement is simpler.
Sources: 2025 Web Almanac: third-party definitions and limitations
A useful tag audit
Start with a high-value landing page and checkout or lead form. Record every externally loaded vendor, the person who owns it, its purpose, the event that triggers it, and the consent state required. Check whether the tag still supports an active decision or campaign. Then inspect its requests and main-thread activity in a browser trace under mobile conditions.
Test the page before and after removing or delaying a candidate tag. Look at field performance where available, and verify that conversions and consent behavior still work as intended. If a tag is essential, optimize when it loads and what it loads. A global statistic gives you a reason to audit; your own trace tells you what to change.