Findings · 2026-08-27 · By VSNARY | Emmanuel Orta
We clicked a real Google result to test the goto redirect. The referrer cannot tell you.
Reports say google.com/goto is at near-100% rollout. From an ordinary browser we saw zero wrapped links — and even if we had, the arriving referrer would look identical.
Google has reportedly begun routing some Search result clicks through google.com/goto?url=[encoded] before redirecting to the destination. As of 26 August 2026 the rollout language firmed up considerably — one rank-tracking vendor reported seeing it at near-total coverage across residential IP providers.
Two things about that are worth measuring rather than repeating.
1. From an ordinary browser, we did not see it
We ran a normal query from a normal Chrome session on a residential connection and read the result links directly out of the DOM:
| Measure | Result |
|---|---|
| Links on the page | 34 |
Wrapped in google.com/goto | 0 |
Wrapped in the older google.com/url? | 0 |
| Pointing directly at the destination | 10 |
We then clicked one and followed it. The browser landed on the destination with redirectCount: 0 — no redirect hop occurred.
That does not disprove the reports. It is one vantage point. But it is worth noting which population the strongest rollout claims come from: rank-tracking infrastructure running through residential IP providers. That is precisely the traffic an anti-scraping measure would target first. A measure aimed at scrapers showing up at scale in scraper fleets, and not in an ordinary browser session, is consistent with the stated motive rather than evidence against it.
2. The recommended detection method cannot work
The standard advice is to search your server logs or analytics for referrers containing google.com/goto. We built exactly that check into our own visitor beacon, deployed it, and then measured a real Google click to confirm it fired.
It cannot fire. Here is what the destination page actually received:
document.referrer = "https://www.google.com/"
referrer path = "/"
redirectCount = 0Google's search results declare an origin-only referrer policy. A site receives the bare google.com origin with no path, whether or not the click passed through a redirect. And an HTTP redirect forwards the original referrer, so the hop does not add one either. Navigation Timing separately reports redirectCount: 0 for cross-origin redirects, so that will not surface it.
The redirect is invisible from the landing page. You will never see google.com/goto in a referrer header, because the path was never sent.
We removed the check the same night we shipped it, and left a comment in the source explaining why, so nobody re-adds it.
What actually would be measurable
If goto ever strips referrers entirely rather than passing the origin through, the effect is a shift in ratio: arrivals that used to carry a google.com referrer start arriving with none. Nothing errors. Reports still populate. Organic sessions fall and direct rises.
The tell is the divergence between two counts you already have: Search Console clicks measured at the SERP, and analytics sessions measured at your site. Search Console counts the click before any redirect. If its numbers hold flat while your organic sessions drop and direct climbs, that is a measurement-layer change, not an SEO failure — and it is worth annotating the date before someone concludes rankings collapsed.
That ratio is worth a weekly glance whether or not goto ever spreads. Browser referrer policies have been tightening for years; this would be one more step along a path the web is already on.
The part nobody should skip
There is a second consequence that has nothing to do with analytics. When a result link points at an encoded Google URL, a person can no longer hover a result and read where it goes. That preview is one of the few remaining checks an ordinary user has against a phishing or spam destination, and wrapping removes it.
Whatever the measurement story turns out to be, that trade-off is real and is being made without an announcement.
Every figure above came out of this scanner.
Point it at your own domain and see the same measurements, free.