Document the handoff before changing settings
Draw the journey from the original landing page through cart, hosted checkout, payment authentication, return URL and confirmation page. Note every hostname, whether it is under your control, which pages run the Google tag, and whether the browser changes domain or opens a new window. Record the expected acquisition source for a controlled test visit.
A payment processor used solely to complete a transaction is not usually a meaningful new referral source. A separate affiliate, review website or marketplace might be. Do not add domains to an exclusion list merely because their referral counts are inconvenient.
Compare three distinct measurement problems
First, self-referrals can occur when traffic appears to originate from your own domain or subdomain. Second, a third-party gateway can be credited after a return redirect. Third, a true cross-domain journey between properties you own can lose user or session continuity if the linker is not configured correctly. These have different fixes.
GA4 automatically recognizes many same-domain and configured cross-domain self-referrals. Its unwanted-referral configuration is designed for specific domains that should not be treated as new sources. It is not a universal repair for duplicate tags, broken session identifiers or missing purchase events.
Use controlled browser tests and DebugView
Begin with a tagged test link whose UTM values you know. Follow the exact purchase or booking flow on a clean browser profile and inspect page locations, referrers, consent state, session-related identifiers and the GA4 events that actually fire. Check whether the confirmation event is sent once, and whether the gateway is present in the event referrer.
Test both successful and cancelled payments, desktop and mobile, and any external authentication step. A return to the site after several minutes may not behave like an uninterrupted on-site journey. Do not assume that an unwanted-referral rule can merge genuinely separate sessions or reconstruct a lost user identity.
Apply the narrowest correct configuration
For an external payment provider that is only an intermediate step, review GA4 Admin's web stream settings for unwanted referrals and add the exact appropriate domain match. Google's documentation describes an ignore_referrer signal on matching events and warns against applying it indiscriminately to all pages. Record the rule, reason, owner and date.
For two owned domains that should share a measurement context, inspect cross-domain measurement and the linker parameter rather than relying only on a referral exclusion. For duplicate or conflicting tags, fix the implementation. Make one change at a time so later attribution movements have an interpretable cause.
Interpret reports with attribution history in mind
An unwanted-referral rule is not necessarily a retroactive rewrite of every report. Google explains that a previously credited source can continue to appear for returning users under session attribution rules, even after a domain is excluded. A lingering gateway row does not automatically mean the new setting failed.
Compare new test traffic and event-level diagnostics with historical aggregates. Review session source/medium, landing pages, purchase counts and revenue, while remembering that GA4 attribution and payment-system totals use different counting and timing logic. Keep an annotation for the date of each configuration change.
Validate business outcomes, not just cleaner channels
Reconcile orders in the commerce or booking platform against GA4 purchase or lead events for the same test cases. Check transaction IDs, currency, refunds and duplicate confirmation loads. A channel report that no longer lists the gateway can still be wrong if the purchase fires twice or never fires at all.
Measure the proportion of purchase sessions credited to the gateway before and after, but do not celebrate a zero figure if other legitimate referrals have been suppressed. The real objective is trustworthy acquisition attribution alongside reliable conversion measurement.
Maintain a handoff and referral register
Create a short register of owned domains, gateway domains, expected referrer behavior, cross-domain settings, test cases and who maintains each integration. Review it when checkout providers, tag containers, consent banners or booking engines change. This prevents the same attribution issue from resurfacing during a vendor migration.
Use the UTM Validator for controlled acquisition tests and pair this guide with the GA4 cross-domain and channel-grouping articles. Those guides address related but different layers: source labels, browser continuity and reporting governance.