Analytics for builders
See which analytics scripts a page loads without reading its source
By Nora Blake · Nov 13, 2026 · 8 min read
You installed an analytics tag, or you are curious what some other site runs, and you do not want to scroll through a minified page source to find out. There are two ways to look, and they answer slightly different questions. One reads what the server sends. The other watches what the browser actually does. This post shows both, what each one misses, and how to read the result.
Start with what the server sends
The fastest check reads the page's HTML and looks inside its script tags. The Peeko analytics checker does this for any page and shows which of 18 analytics tools it loads, with no signup. You paste the address, press Check, and read the list. Where a tool carries a tracking ID, the ID is shown next to it.
Three limits come with that, and they matter more than the list itself:
- It checks one page per run. A tool added only to your checkout or your blog template will not show up on your home page.
- Some sites block automated requests with a firewall or a bot challenge. The checker reports that it could not read the page instead of guessing.
- It sees only what is in the initial HTML. A tool that starts running in the visitor's browser without being in that HTML is missed, so a missing tool may still be running.
The checker also flags one thing worth fixing on sight: a leftover Universal Analytics tag. Google no longer processes it, so it collects nothing you can use, and you should remove it.
What the list usually looks like
Before you read your own result, it helps to know what the web at large returns. W3Techs publishes the share of all websites running each traffic analysis tool. On Nov 10, 2026, Google Analytics was on 46.7% of sites, and the next tool, Meta Pixel, was on 8.4%.
Source: W3Techs, Usage statistics and market share of traffic analysis tools for websites, Nov 10, 2026. Retrieved Nov 10, 2026.
Two cautions on that chart. W3Techs measures a fixed list of tools, so the 42.7% of sites with none are sites with none of the tools it monitors, not sites with no measurement at all. And a tool on the page is not a tool that works: a tag can be present and misconfigured. The check tells you what is loaded, not what is recorded.
If your own result is an empty list, you are in the largest group, but that is only a comfort if you meant to be there. Right after an install it is a signal that something is off.
Why an empty list is sometimes wrong
When you expect a tool and the HTML check does not show it, the cause is almost always one of three things.
A tag manager loads it. The page source contains only the tag manager's snippet. The analytics tool is fetched after that, by rules stored in the manager's account, and it never appears in the HTML. This is not a corner case. W3Techs reports Google Tag Manager on 44.7% of websites, and 99.6% of the market share among the tag managers it tracks.
Source: W3Techs, Usage statistics and market share of tag managers for websites, Nov 10, 2026. Retrieved Nov 10, 2026.
A consent banner holds it back. The tag is written into the page only after a visitor accepts. The server sent no analytics script, so a check of the HTML has nothing to find, and neither does a visitor who declines.
The page builds its own head in the browser. A single-page app can inject a script after it starts. Again, the server's HTML is clean and the browser's activity is not.
In all three cases the answer is in the browser, not the source.
Watch the browser do it
Open the page, open DevTools, go to the Network tab, and reload. Google's own instructions for its tag say it this way: if you see requests to googletagmanager.com, your tag is firing, and to check a specific product you filter for domains such as google-analytics.com. Google also offers Tag Assistant, where your tag ID should appear under the Google tags found header.
Other vendors give the same kind of test. Plausible tells you to reload with the Network tab open and look for a request from plausible.io whose name starts with pa-, which should return status 200.
To check the consent case, run the check twice: once before you touch the banner and once after you accept. The requests that appear only in the second run are the ones the banner was holding.
You can also check for the cookie. Google Analytics sets _ga and _gid as first party cookies, and the Web Almanac's 2024 cookie chapter lists them as the two most common first party cookies it found. Open the Application tab, look at Cookies, and a _ga cookie tells you Google Analytics ran even if you could not find its script. A tool that sets no cookie leaves nothing to find there, and Peeko is one of them: it sets no cookie.
A command line version of the first check
If you would rather have the server's view in a terminal, this prints every analytics host or attribute it recognizes in a page's HTML, once each:
curl -sL https://example.com | grep -oiE 'googletagmanager\.com[^"]*|google-analytics\.com[^"]*|plausible\.io[^"]*|cdn\.usefathom\.com[^"]*|matomo\.(js|php)|data-website-id|data-site=' | sort -u
Replace https://example.com with your page. Each pattern comes from a vendor's own install instructions:
- Fathom's embed code loads
https://cdn.usefathom.com/script.jswith adata-siteattribute. - Umami's tracker is a
script.jsfile with adata-website-idattribute. - Matomo's snippet loads
matomo.jsand sends data tomatomo.php. - Google's tag loads from
googletagmanager.com.
This has the same blind spots as the checker, plus one that is easy to forget. Umami's docs show the script served from your own domain, so the host tells you nothing and only the attribute gives it away. Plausible documents a proxy that routes its script through your own domain, which it says makes the script "indistinguishable from your own files." A string match fails on that setup by design. If a page makes analytics requests to its own domain after load, you will see them in the Network tab and not in the HTML.
Checking your own install the same day
The sequence that works after you ship:
- Run the checker on the live address, not a local one. It fetches the page itself, so it needs a page the public web can reach.
- Look for exactly the tool you chose, once. Peeko is one script tag with a site key and needs no configuration once it is on the page, so one tag is the whole expectation. If you see a second tool you did not pick, or the same one twice, look at what your theme, your template or an agent added. Running several analytics tools at once can slow pages, and their numbers will not match.
- If the tool is missing, look at the Network tab, as above, before you change any code.
- Load the page once yourself and watch the request go out.
Step 4 has a trap. Peeko's ingest endpoint answers 204 to every well formed beacon, valid key or not, so a tracker can never leak whether a key is real. That is the right design for privacy, and it means a 204 in your Network tab proves the beacon was sent and well formed. It does not prove the key is yours. The proof is the visit showing up in your numbers. If it does not, the ranked causes for a missing first page view go in order of likelihood, and the key is on it.
A faster way to get that proof from a coding agent is described in installing analytics and proving it works in one agent session. If you want the numbers without opening a dashboard, reading your traffic from a terminal uses npx @usepeeko/cli, which speaks the same endpoint an agent does.
One more thing about your own test visits: they count as traffic. If you reload a dozen times to watch the Network tab, you have added a dozen page views to a new site. Keeping your own visits out of your analytics covers how to stop that before the first week's numbers mean anything.
Reading someone else's site
Out of curiosity you can run the same check on any public page. Keep the claim small. A result tells you which tools were in the HTML the server returned to an automated request. It does not tell you what the site does with the data, whether a banner gates those tools for real visitors, or what a tag manager loads once a person arrives.
Some readings are safe:
- A tool with an ID in the HTML is loaded on that page. The ID is the site's own, and you can search for it to see whether the same ID appears on other pages.
- No tool found, and a tag manager found, means nothing about whether analytics runs. It means you need the Network tab.
- No tool found and no tag manager found is consistent with no analytics. It is not proof, because of the proxy and consent cases above.
- A blocked check says the site refuses automated requests. Try another page, or open the source in your browser.
What to do with the result
If you checked your own site, three outcomes cover it. The tool is there once: you are done, and the next step is a visit that reaches your numbers. The tool is there twice or a stray one appeared: remove the extra one. The tool is missing: use the Network tab to find out whether it is not loading or loading late, then fix that single cause.
The reason to do this on launch day and not a week later is cost. A missing script costs you the visits you cannot get back, and the check takes a minute.