Analytics & tracking scripts

Google Analytics and Tag Manager measure different things

Google Analytics 4 has the largest recorded total among these seven queries: 9,614,187, or 7.52% of the documented index. The query looks for gtag/js?id=G- in stored homepage HTML. It is a specific loader pattern. A result gives you a source reference to inspect, along with the measurement ID and surrounding setup code when those are present.

Google Tag Manager has 7,391,424 references to googletagmanager.com/gtm.js. Its role differs from a direct analytics loader: a container can deploy several tags, including third-party services. Google’s container setup documentation describes that arrangement. The homepage’s Tag Manager snippet therefore does not enumerate everything configured inside the container.

For an implementation review, inspect these two patterns separately. A page might include a direct Google tag, a Tag Manager container, or both. Analytics loaded through a container may not expose the direct loader string in the stored HTML. Adding the two rows would count some pages twice and still miss implementations that do not expose either selected reference.

Try "gtag/js?id=G-" "googletagmanager.com/gtm.js" to investigate pages containing both references. This is a current query rather than a measured intersection in the August archive. Its results are useful for reviewing whether tags were added through multiple routes. Confirm the configuration and actual event requests before concluding that a page sends duplicate analytics events.

What the Universal Analytics references mean

The Universal Analytics row records 4,627,646 references to google-analytics.com/analytics.js. Google has replaced Universal Analytics with GA4. An old loader remaining in homepage source is therefore useful evidence for a cleanup review. It should not be read as a count of functioning Universal Analytics properties or businesses that have failed to migrate.

A site may have added GA4 while leaving a previous theme snippet in place. Another may preserve the old URL in commented code, a plugin setting or a published example. The matched source helps distinguish those situations. Record where the string occurs before assigning the result to a migration backlog, particularly when the match appears outside an executable script element.

Run "google-analytics.com/analytics.js" "gtag/js?id=G-" to inspect pages containing both loader patterns. Search "google-analytics.com/analytics.js" -"gtag/js?id=G-" for the old reference without that direct GA4 pattern. The second query is a lead for investigation, not proof that GA4 is absent: a Tag Manager implementation or another loading route may still be present.

Clarity and Hotjar source references

Microsoft Clarity has 637,831 recorded references to clarity.ms/tag/; Hotjar has 388,400 for static.hotjar.com/c/hotjar-. These are separate script-path queries. Their totals can help prioritize which source patterns to examine when building an inventory of measurement tools across a collection of websites.

Microsoft’s Clarity installation guide describes site tracking code and several installation methods. The particular path in this report only measures what is exposed in the saved homepage. A script deployed later through another loader can require runtime inspection. Treat a missing source match as an unanswered implementation question, rather than a definitive inventory of everything the browser eventually loads.

For a manual review, keep the provider, project identifier, matched path and page date together. When two tools appear on one site, investigate their intended roles with the site owner before treating either as redundant. The archive contains no session data, tool configuration, account access or usage history. It can locate references; it cannot explain how a team uses the associated product.

Meta Pixel and Cloudflare Insights

The Meta Pixel query uses fbevents.js and has 2,986,269 recorded matches. It finds the selected script filename, rather than the identity of every business advertising through Meta. If your task concerns a particular site’s setup, inspect the nearby initialization code and any visible pixel identifier. Keep a distinction between the library’s common filename and the account-specific value.

The Cloudflare Insights query has 5,814,915 references to cloudflareinsights.com. Cloudflare’s Web Analytics setup guide explains how to add its analytics snippet, including on sites that are not proxied through Cloudflare. This row therefore describes a beacon reference. It does not measure all domains using Cloudflare’s DNS, CDN or other infrastructure.

Those distinctions matter when naming an exported list. Use labels that preserve the actual selection, such as “homepages referencing fbevents.js” or “Cloudflare analytics beacon matches.” A colleague can then understand what to verify next. A broader label such as “advertisers” or “Cloudflare customers” would imply facts that are not recorded in this source snapshot.

How to audit an analytics result

Start by opening the matched HTML and finding the exact string. Identify whether it appears in a script URL, inline configuration, a comment or visible page text. Copy enough surrounding source to show the context, then record the hostname and snapshot date. This produces an evidence trail that another developer can inspect without having to repeat the initial search.

Next, separate source presence from runtime behavior. On a site you are reviewing, browser developer tools can show whether the script is fetched and whether requests follow. Google’s Google tag setup guide documents verification through Tag Assistant and developer tools. Use the provider’s own tools when checking configuration or delivery, rather than treating a source-code match as a successful event test.

For consent-related debugging, record the browser state used during your check: whether a consent choice was made, which page was loaded and which requests were observed. The archived HTML contains no record of a visitor’s choices or subsequent interactions. A useful technical handoff describes those observations directly, with request details and reproduction steps, instead of inferring behavior from a loader’s name.

Keep findings actionable. A cleanup task should identify the template or integration responsible for the reference, the proposed change and the pages to retest. A migration task should include the intended destination and event checks. A shared-identifier investigation should retain the exact identifier and source evidence; a shared value alone does not establish that two domains have the same owner.

Investigate an individual tracking identifier

A provider’s common script path finds many unrelated sites. An identifier can narrow the investigation to source containing a specific configured value. First obtain the identifier from a source you are reviewing, then search for that exact value. For example, G-EXAMPLE123 is an illustrative placeholder, not a measured ID from this report. Replace it with the value you actually observed.

Check every returned occurrence in context. A published tutorial can mention an identifier without using it, while a shared template can distribute a configuration across several domains. Record whether the value appears in initialization code, a loader URL or explanatory text. Compare other evidence before drawing a relationship between sites. A shared analytics identifier is a technical connection worth investigating; it does not, by itself, identify an owner or explain who controls the relevant account.

Build a repeatable analytics comparison

Choose the same query strings and index scope for every measurement. Preserve the count label alongside each total, and note whether you used a top-level-domain filter. If your project adds a new loader pattern, report it separately before combining hostname lists. Otherwise, a change in detection coverage can look like a change in adoption even when the underlying websites have not changed.

Do not calculate a total number of tracked websites by summing these seven rows. A homepage can contain several tools, and some implementations expose none of the listed strings. Measuring the union requires combined queries or hostname lists that can be deduplicated. The intersection between GA4 and Meta Pixel, for example, is a separate research question that this archive does not answer.

Download the original CSV before filtering or enriching the results. Retain its query, recorded count, denominator, month and measurement columns. Add your own review dates and notes separately. For research involving a specific Google Analytics identifier, the shared Analytics ID guide covers narrower searches and the source checks needed to interpret them.