If GA4 treats two websites you operate as disconnected journeys, check whether the real browser navigation carries the _gl linker parameter and whether the destination uses the intended measurement setup. A redirect can remove the parameter after the link was decorated, too quickly to diagnose from the final address bar alone.
Confirm this is a cross-domain problem
Record both domains and the web stream or Google tag used on each. Google recommends configuring cross-domain measurement through Analytics for participating domains using the appropriate shared setup. A third-party payment provider is a different case: start with unwanted payment referrals, not an attempt to install your tag on a provider you do not control.
Read Google's cross-domain setup guide and compare the configured domains with the actual destination host. Include real subdomain and redirect behavior in that comparison. A setting for one host does not prove every partner or alternate hostname is part of the measurement boundary.
Trace a newly generated navigation
Open the source page in an authorized test session with the required consent. In browser Network tools, preserve the log. Click the actual link or submit the actual form. Do not copy a decorated address from an old session and treat it as a durable identifier.
Record the outgoing navigation, every redirect and the first destination document request. For each hop, mark whether _gl is present without publishing its value. If the source never decorates the link, inspect domain matching, tag initialization and the type of navigation. A custom JavaScript redirect may not behave like a normal anchor click.
Domain: Correct host configuration. Transport: Real linker reaches destination. Measurement: Intended tag and consent. Regression: Links and forms both tested. Explanatory checklist, not a customer test result.
Assign the first break
| First failure |
Responsible area |
| No decorated source navigation |
Tag configuration or custom click handler |
| Redirect removes parameter |
Redirect owner or URL allowlist |
| Parameter reaches destination but no tag |
Destination deployment or consent integration |
| Tag reads different stream |
Measurement configuration across sites |
Google documents redirect removal as a known failure mode. Fix the specific redirect through a structured URL operation that preserves approved measurement parameters. Do not forward arbitrary sensitive query values across domains, and do not append duplicate question marks to an existing query string.
Never generate your own _gl value or replace it with a GCLID. They have different purposes. A fabricated value surviving the redirect proves only string transport, not successful cross-domain measurement.
Verify the full path, not just the URL
Use Tag Assistant to inspect both ends of a fresh journey. Confirm the intended tag runs under the approved consent state and that the destination consumes the linker's information as expected. Repeat with an existing query parameter, a language redirect and a form navigation if those are real customer routes.
Check the processed acquisition reports after normal reporting delay, using the same dimensions and a dated test window. Do not claim a single continuous session solely because _gl remains in the address bar. Transport is necessary evidence, not the entire attribution contract.
For advertising landing redirects, use the separate GCLID hop-by-hop guide. HandL WP can audit the two-site measurement path without replacing unrelated tags or changing customer consent choices.
References reviewed October 10, 2026. Examples are explanatory, not customer test results.