{
  "measured": "2026-09-15",
  "what": "Internal links carrying UTM campaign parameters, across a random Tranco sample",
  "method": "Docket audited 40 hosts drawn with seed 20260915 from the Tranco top 10,000, at most 25 pages each, with its own identified user-agent and its own backoff. Ordinary crawling at the scale a customer runs. A host counts when mar.internal_utm fires, which is when any indexable page emits a link to the same site carrying UTM parameters (seo_engine/checks/martech.py). Every host crawled successfully.",
  "ceiling_warning": "pages_affected is bounded by the 25-page crawl limit. A host showing 25 means every page we looked at carried such a link, NOT that the site has that many pages. Those hosts are a lower bound, and the share of their site affected is unmeasured.",
  "not_measured": "Whether any analytics tool actually re-attributed a session. This counts what the HTML emits. The consequence in GA4 is documented behaviour, not something this survey observed.",
  "sample": 40,
  "seed": 20260915,
  "pages_per_site": 25,
  "crawled_ok": 40,
  "hosts_with_internal_utm": 6,
  "pct_hosts": 15.0,
  "links_total": 177,
  "links_median": 17.5,
  "links_min": 1,
  "links_max": 91,
  "pages_affected_sorted": [
    1,
    1,
    4,
    24,
    25,
    25
  ],
  "hosts_at_page_cap": 2,
  "hosts_isolated_4_or_fewer": 3,
  "hosts_withheld_because": "Publishing a named site's misconfiguration to sell a tool is not a trade this project makes.",
  "what_the_check_counts": "Narrower than 'internal links with UTM parameters', verified in the engine at extract.py:1151-1157 after a writer read it: (1) a link counts as internal only when same_site() says so, and that treats ONLY www and the apex as equivalent \u2014 a link to a subdomain is external to this check, so cross-subdomain tagging is invisible here; (2) only pages in ctx.indexable are examined, so a tagged link on a noindex page never makes its host count; (3) the match is the substring 'utm_' in the href, not a parsed query parameter, so anything utm_-prefixed counts and a utm_ in a path or fragment would count too; (4) links_total counts anchors, not distinct URLs, so a repeated tagged nav item counts once per page it appears on.",
  "shape_not_claimed": "The per-host page counts sit at two ends with nothing between, but six observations cannot establish the shape of a distribution and a split that clean in a group this small is well within chance. Do not call it bimodal."
}