How cross-domain tracking works
Analytics recognises a returning visitor by an identifier stored in the browser, and browsers keep that storage separate for every domain. So the moment a visitor crosses from your website to a booking engine, a payment page or a shop at a different address, the second site cannot read what the first one wrote. It issues a fresh identifier, opens a new session, and records the arrival as a referral from your own site.
Cross-domain tracking solves this by handing the identifier across inside the link. When both domains are listed in the tag configuration, the tag adds a linker parameter to outbound links between them; the receiving site reads that parameter, adopts the same identifier, and carries on the session that was already running. Domains you have not named are unaffected, so nothing leaks to sites outside the list.
Why cross-domain tracking matters
Without it, one journey becomes two and both halves mislead. The first site appears to lose everybody at exactly the moment they were about to buy, so its conversion rate looks poor and its landing pages look weak. The second site appears to produce purchases out of its own referral traffic, so the campaign that actually paid for the visitor receives no credit at all.
The pattern is very common in Nepal and in travel generally: hotels and trekking agencies that hand booking to an external engine, clinics running a third-party appointment system, and any checkout that redirects to a payment provider on its own domain. In each case the money is taken somewhere the original tag cannot see, unless it has been configured to follow the visitor there.
Where cross-domain tracking goes wrong
The most frequent fault is listing only one of the domains, which works in one direction and fails quietly on the way back. Next comes a redirect that drops the linker parameter before the destination can read it — a link shortener, a strict security rule, or a redirect chain sitting in the middle of the checkout will all do it.
Then there are the parts you do not control. Some third-party checkouts strip parameters or forbid extra scripts altogether, in which case the honest answer is to measure differently rather than keep tuning the tag. Subdomains are a separate matter and normally need no configuration at all, because they share a domain; people add them anyway and assume they have fixed something.
How to act on it
List every domain a visitor can legitimately cross into, in the tag settings on both sides, then walk the journey yourself and watch the address bar — the linker parameter should be visible on the destination URL. Follow that with a live check that one visit now produces a single session rather than two, and that your own domain has stopped appearing in the referral report.
Where a third-party system genuinely cannot be tagged, agree the fallback before anybody reports a number: a server-side confirmation, an order identifier handed back, or reconciliation against the payment provider’s own totals. Whichever route you take, write it into your GA4 setup documentation, because this is precisely the configuration that disappears in a site rebuild and leaves nobody able to explain why the numbers used to add up.
