How to Fight the Google Analytics Referral Spam Issue

Unexpected referrals are not one problem

Unexpected referral traffic in GA4 may represent a legitimate acquisition source, a payment provider, movement between owned domains, a self-referral, broken campaign tagging, automated traffic or deliberately abusive activity. The source and journey need to be understood before configuration changes are made.

GA4’s unwanted-referrals setting affects how future events are attributed. It does not delete historical events or repair a cross-domain implementation.

Diagnose the referral before excluding it

Begin with the session source and medium, landing page, hostname, geography, engagement, event pattern and steps immediately before conversion. Reproduce the journey where possible.

A payment provider appearing as a referrer may mean the customer left the site to pay and returned without the journey being handled correctly. An owned domain may indicate missing cross-domain configuration. An unfamiliar domain with repetitive, implausible behaviour may be automated traffic, but low engagement alone is not proof of spam.

Do not visit a suspicious domain merely to identify it. Use reporting evidence, safe reputation checks and technical logs where available.

Build analytical segments from evidence

Use GA4 comparisons, explorations or segments to isolate the suspected population and compare it with a known baseline. Test dimensions such as hostname, landing page, geography, device, engagement, event sequence and date rather than relying on a single weak signal.

Use the narrowest conditions supported by evidence. Record the rule, the period it covers, and examples that should and should not match. The objective is to understand the scale and reporting impact, not to create a permanently flattering version of the data.

Choose the remedy that matches the cause

For owned domains that participate in one journey, configure and test cross-domain measurement. For eligible third-party domains such as payment services that should not begin a new attribution path, consider the unwanted-referrals list after confirming the consequence.

Correct broken campaign tagging or tag deployment at source. If traffic is generated by requests that never interact with the real site or tag, investigate the collection path, credentials and implementation rather than expecting a report filter to secure it.

Handle historical reporting honestly

Segments, comparisons and explorations can isolate suspicious patterns for analysis, but they do not alter events already stored in the property. Document the rule, affected dates, evidence and limitations so future analysts understand why a filtered analysis differs from the raw property.

Universal Analytics view filters, bot-filtering checkboxes and old referral-exclusion instructions should not be copied into a GA4 implementation. Historical procedures are useful only when clearly labelled as Universal Analytics material.