{
  "version": "https://jsonfeed.org/version/1.1",
  "title": "ConsentVeil Journal",
  "home_page_url": "https://consentveil.com/blog/",
  "feed_url": "https://consentveil.com/blog/feed.json",
  "language": "en",
  "items": [
    {
      "id": "https://consentveil.com/blog/cookieyes-wordpress-google-consent-mode-default-denied/",
      "url": "https://consentveil.com/blog/cookieyes-wordpress-google-consent-mode-default-denied/",
      "title": "CookieYes on WordPress: why Google tags can run as if consent were granted",
      "summary": "A real audit of a WordPress site with CookieYes: the Reject button worked, but Google tags loaded without a Consent Mode default. What we observed, how to check it in two minutes, and what to change.",
      "content_html": "<p>The “Reject all” button worked on this WordPress site. The scan recorded no request to a known tracking host before a choice, and refusal survived a reload. But the page’s <code>dataLayer</code> contained no Google Consent Mode <code>default</code> command. That missing instruction was the finding to fix: check the CookieYes setting and make sure the denied default appears before the first Google tag entry.</p>\n<h2 id=\"what-we-scanned\">What we scanned</h2>\n<ul>\n<li>A Romanian business website on WordPress 6.x, published with the owner's written permission and anonymised here.</li>\n<li>Consent platform: CookieYes (WordPress plugin), banner in Romanian with three first-layer buttons: <em>Personalizați</em>, <em>Refuzare toate</em>, <em>Acceptare toate</em>.</li>\n<li>Google tags loaded through Site Kit, offloaded to a web worker with Partytown.</li>\n<li>Scope: home page plus two more pages, Chromium and WebKit, locale <code>ro-RO</code>, <a href=\"https://consentveil.com/monitor/methodology/0.3.2/\">methodology 0.3.2</a> at the time of the scan. The flows and tests are the published ones: <a href=\"https://consentveil.com/monitor/methodology/\">methodology</a>.</li>\n</ul>\n<h2 id=\"what-the-banner-did-right\">What the banner did right</h2>\n<p>The scan recorded the following results:</p>\n<ol>\n<li><strong>No request to a known tracking host before any choice.</strong> The clean load, with no interaction, reached no classified tracking host.</li>\n<li><strong>\"Reject all\" was on the first layer</strong>, at the same click depth as \"Accept all\", with no measurable size or weight asymmetry between the two.</li>\n<li><strong>Rejecting stopped tracking requests</strong> in the same page view, and the refusal <strong>survived a reload</strong>.</li>\n<li><strong>The banner did not come back</strong> after the refusal.</li>\n<li><strong>Consent could be withdrawn</strong> from the revisit button, and no tracking request followed the withdrawal.</li>\n</ol>\n<p>The second layer, opened with <em>Personalizați</em>, showed four categories, all off by default. Nothing was pre-selected.</p>\n<h2 id=\"what-failed-no-consent-mode-default\">What failed: no Consent Mode default</h2>\n<p>The test <code>consent-mode-default-denied</code> reads the page's <code>dataLayer</code> before any interaction and looks for a <code>consent</code> entry whose first argument is <code>default</code>, with <code>ad_storage</code> and <code>analytics_storage</code> set to <code>denied</code>, placed <strong>before</strong> the first Google tag entry. The verdict is about the order of entries in the <code>dataLayer</code>, not about network timing.</p>\n<p>On this site the <code>dataLayer</code> held <strong>zero</strong> consent commands. The Google tag was there; the instruction telling it to start in a denied state was not. Google documents the consequence plainly: when no default is set, the tag's consent types are treated as granted. In this scan no measurement request to a classified tracking host was recorded before the choice, so the practical effect was limited to what the tag was configured to send. The missing default remained a separate configuration finding even though the network observation passed.</p>\n<p>Two things made this easy to miss:</p>\n<ul>\n<li><strong>CookieYes does not push the Consent Mode default unless its Google Consent Mode option is enabled.</strong> The banner can block cookie categories perfectly well while Google's tags never receive a <code>default</code>.</li>\n<li><strong>Partytown moves the Google tag into a worker.</strong> The consent default has to be pushed to <code>window.dataLayer</code> on the main thread, before the tag is forwarded, or the tag starts without it.</li>\n</ul>\n<h2 id=\"how-to-check-this-yourself-in-two-minutes\">How to check this yourself in two minutes</h2>\n<ol>\n<li>Open the site in a private window. Do not touch the banner.</li>\n<li>Open the browser console and run:</li>\n</ol>\n<pre><code class=\"language-js\">(window.dataLayer || []).filter(e =&gt; e[0] === 'consent')\n</code></pre>\n<ol>\n<li>Expected: at least one entry <code>[\"consent\", \"default\", {...}]</code> with <code>ad_storage: \"denied\"</code> and <code>analytics_storage: \"denied\"</code>, and that entry must come <strong>before</strong> the first <code>[\"config\", \"G-...\"]</code> or <code>gtm.js</code> entry in <code>window.dataLayer</code>.</li>\n<li>Click \"Reject all\" and run the same command again. Expected: an <code>update</code> entry follows, still <code>denied</code>.</li>\n<li>If the array is empty, this expression found no consent command in the current array. Check the integration and the full sequence before drawing a wider conclusion.</li>\n</ol>\n<h2 id=\"what-to-change\">What to change</h2>\n<ol>\n<li><strong>In CookieYes</strong>, enable the Google Consent Mode option (in the plugin's advanced settings, or in the CookieYes dashboard for the site, depending on how the banner is served). CookieYes then pushes <code>default</code> with the categories you map, and <code>update</code> after the visitor's choice.</li>\n<li><strong>Load order.</strong> The CookieYes script must run before the Google tag. In WordPress that usually means keeping the CookieYes snippet in <code>&lt;head&gt;</code> ahead of Site Kit's output. If you use Partytown, keep the consent default on the main thread, outside the worker, and let the tag be forwarded afterwards.</li>\n<li><strong>Re-test</strong> with the console check above, in a private window, on at least the home page and one inner page.</li>\n</ol>\n<p>Acceptance criterion, as it appears in the developer ticket: <code>consent-mode-default-denied</code> passes on the re-scan, in both browsers.</p>\n<h2 id=\"what-our-scanner-got-wrong-first-and-what-we-changed\">What our scanner got wrong first, and what we changed</h2>\n<p>The first scan of this site reported \"no reject control found\" and six inconclusive results. The button was visible on the screenshot. The scanner had no adapter for CookieYes and matched Romanian labels only in the imperative form (<em>Refuză toate</em>), while the plugin renders the noun form (<em>Refuzare toate</em>). We added the CookieYes adapter and the missing label forms, re-ran the scan, and only then delivered the report. The methodology says detection is selector and label based and can miss unusual markup or wording; this was a plain case of it. The initial result needed correcting before the report could be delivered.</p>\n<p>Since methodology 0.3.3, the platform's own consent record (here, the <code>cookieyes-consent</code> cookie) is listed apart in the storage test and no longer counted as pre-consent storage, and the Global Privacy Control test is reported as <em>not tested</em> when the clean load reaches no tracking host, instead of a low-confidence warning. Both changes came out of this audit.</p>\n<h2 id=\"implementation-references\">Implementation references</h2>\n<p>Compare the setup with Google’s <a href=\"https://developers.google.com/tag-platform/security/guides/consent\">Consent Mode implementation guide</a> and CookieYes’s <a href=\"https://www.cookieyes.com/documentation/google-consent-mode-v2-integration-with-cookieyes-wordpress-plugin/\">WordPress integration instructions</a>. These explain the configuration; the observations above describe this particular scan.</p>\n<h2 id=\"faq\">FAQ</h2>\n<p><strong>Does a working \"Reject all\" button mean Google Consent Mode is configured?</strong>\nNo. The two are independent. The button controls what the consent platform blocks; Consent Mode is a separate signal that Google's tags read from the <code>dataLayer</code>. A banner can block cookies correctly while the tags never receive a <code>default</code>.</p>\n<p><strong>Where does the <code>consent default</code> have to be?</strong>\nIn <code>window.dataLayer</code>, before the first Google tag entry. Our test checks that order; it does not measure network timing.</p>\n<p><strong>Is this a GDPR violation?</strong>\nThat is a legal question and we do not answer it. The report records a technical observation: Google tags loaded without a Consent Mode default. What that means for a specific site is for the site owner and their adviser to decide.</p>\n<p><strong>Does Partytown break Consent Mode?</strong>\nNot by itself. It changes where the tag runs. If the consent default is pushed only inside the worker, or after the tag is forwarded, the tag starts without it. Push the default on the main thread first.</p>\n<p><strong>How was the site owner involved?</strong>\nThe audit was ordered in writing, the owner read the full report first, and agreed to this anonymised write-up. No client site is scanned without that authority.</p>\n<hr />\n<p><em>This article describes technical observations from one audit under a stated methodology version. It is not legal advice and not a certification. Methodology, prices and documents: <a href=\"https://consentveil.com/monitor/\">consentveil.com/monitor</a>.</em></p>",
      "date_published": "2026-11-17T00:00:00Z",
      "date_modified": "2026-11-17T00:00:00Z",
      "authors": [
        {
          "name": "Claudiu-Iulian Petcu, Safe and Informed S.R.L."
        }
      ]
    }
  ]
}