The Problem

If you're running a Single Page Application (SPA) built with frameworks like Nuxt.js, Next.js, React Router, or Vue Router, you might notice an unusually high percentage of Direct traffic in your Google Analytics 4 reports.

This happens because:

  1. SPAs don't trigger traditional page loads - When users navigate between pages, the browser doesn't perform a full page reload. Instead, JavaScript updates the content dynamically.

  2. document.referrer doesn't update on SPA navigation - The browser's document.referrer only captures the referrer from the initial page load. It doesn't change when users navigate within your SPA.

  3. GA4's page_referrer parameter is empty or stale - Without proper configuration, GA4 receives an empty dr (document referrer) parameter, causing it to classify sessions as "Direct" even when users came from other pages on your site or external sources.

Diagnosing the Issue

Symptoms

How to Confirm

  1. Open your browser's Developer Tools → Network tab
  2. Navigate between pages on your SPA
  3. Filter for requests to google-analytics.com or your server-side endpoint
  4. Look at the dr parameter in the GA4 collect requests

If dr= is empty on internal navigations, you have this issue:

/g/collect?v=2&tid=G-XXXXXX&...&dr=&dl=https://yoursite.com/page2/...
                                ^^^
                                Empty!

The Solution

Overview

We need to:

  1. Track SPA navigations using GTM's History Change trigger
  2. Capture the previous page URL and pass it as page_referrer
  3. Prevent duplicate page views on initial page load

Implementation in Google Tag Manager

Step 1: Create the Page Referrer Variable

Create a Custom JavaScript variable that returns the correct referrer for both scenarios:

Variable Name: CJ - Page Referrer (SPA)

function() {
  // Get the old URL from History Change event (for SPA navigations)
  var dataLayer = window.dataLayer || [];
  var historyOldUrl = null;

  // Find the most recent gtm.historyChange event
  for (var i = dataLayer.length - 1; i >= 0; i--) {
    if (dataLayer[i].event === 'gtm.historyChange' && dataLayer[i]['gtm.oldUrl']) {
      historyOldUrl = dataLayer[i]['gtm.oldUrl'];
      break;
    }
  }

  // If we have a history old URL and it's different from current URL, use it
  if (historyOldUrl && historyOldUrl !== window.location.href) {
    return historyOldUrl;
  }

  // Otherwise, return document.referrer (for initial page load)
  return document.referrer || '';
}

Step 2: Create a Helper Variable to Detect Same-URL History Changes

Some SPAs trigger replaceState on initial load where oldUrl === newUrl. We need to block these to prevent duplicate page views.

Variable Name: CJ - Is Same URL (History Change)

function() {
  var dataLayer = window.dataLayer || [];

  // Find the most recent gtm.historyChange event
  for (var i = dataLayer.length - 1; i >= 0; i--) {
    if (dataLayer[i].event === 'gtm.historyChange') {
      var oldUrl = dataLayer[i]['gtm.oldUrl'] || '';
      var newUrl = dataLayer[i]['gtm.newUrl'] || '';
      return oldUrl === newUrl;
    }
  }

  return true; // Block if no history change found
}

Step 3: Create the History Change Trigger

Create a new trigger:

Step 4: Create the Blocking Trigger

Create a trigger to prevent firing when it's not a real navigation:

Step 5: Create the SPA Page View Tag

Create a new GA4 Event tag:

Step 6: Update Your Existing Page View Tag

Add the page_referrer parameter to your existing Page View tag (the one that fires on initial page load):

How It Works

┌─────────────────────────────────────────────────────────────────┐
│                        Initial Page Load                        │
├─────────────────────────────────────────────────────────────────┤
│  1. User lands on your site from Google                         │
│  2. Original Page View tag fires                                │
│  3. page_referrer = document.referrer (google.com)              │
│  4. GA4 correctly attributes: Organic Search                    │
└─────────────────────────────────────────────────────────────────┘
                              ↓
┌─────────────────────────────────────────────────────────────────┐
│                     SPA Navigation                              │
├─────────────────────────────────────────────────────────────────┤
│  1. User clicks internal link (Home → Product Page)             │
│  2. History Change event fires                                  │
│  3. SPA Page View tag fires                                     │
│  4. page_referrer = gtm.oldUrl (your homepage)                  │
│  5. GA4 maintains session continuity                            │
└─────────────────────────────────────────────────────────────────┘

Additional Considerations

Server-Side Tagging

If you're using server-side GTM (Stape, Addingwell, etc.), ensure your server-side GA4 tag is configured to forward the page_referrer parameter from the client.

Disable Conflicting Enhanced Measurement

In GA4 Admin → Data Streams → Enhanced Measurement, consider disabling "Page changes based on browser history events" to prevent duplicate tracking. GTM will handle this instead.

Testing

  1. Open GTM Preview mode
  2. Navigate to your site
  3. Click through several pages
  4. Verify in the Preview panel:
    • Initial load: Original Page View tag fires with external referrer
    • SPA navigation: SPA Page View tag fires with internal referrer
    • No duplicates on initial replaceState events

Results

After implementing this solution:

Common Pitfalls

  1. Don't forget the blocking trigger - Without it, you may get duplicate page views
  2. Test thoroughly - SPAs vary in how they handle routing; test your specific framework
  3. Check consent timing - Ensure consent is granted before page view tags fire
  4. Coordinate with developers - For best results, have developers push a custom virtualPageview event with all necessary data

Related Resources

This solution was implemented for a Nuxt.js e-commerce site experiencing 50%+ Direct traffic. After deployment, Direct traffic dropped to expected levels (~15-20%) and campaign attribution improved significantly.