google.com/goto explained: what actually changed
On 26 August 2026 Google confirmed that links in its search results are being rewritten. Where a result used to point straight at your page, it now points at google.com/goto?url=[encoded token], which passes the click through to the real destination.
The change is real, confirmed, and widely rolled out. It is also being described far more dramatically than the evidence supports. Here is what actually changed, what genuinely breaks, and what does not.
What changed on 26 August 2026
What Google said
A Google spokesperson gave the same statement to Search Engine Roundtable and Search Engine Land: “We have a long history of deploying technical measures against evolving forms of abuse, and we regularly take steps to protect our services and users.”
UNCHANGED
- Your rankings
- Search Console data
- Canonicals and sitemaps
- Where visitors land
AFFECTED
- Rank tracker costs
- AI Overview link data
- Domain-level filtering
- Anything scraping the SERP
Google did not name the goto parameter, has not documented it, and did not elaborate. The widely-shared reading is that this is an anti-scraping measure, because it makes automated extraction of search results considerably more expensive.
The timeline
| Date | What happened |
|---|---|
| 23 June 2026 | First public sighting of the goto format posted on X |
| 2 July 2026 | Pattern documented across organic and paid listings |
| Early August | Becomes widely visible in results |
| 26 August 2026 | Near-full rollout reported; Google confirms |
| 27–28 August | Data providers publish resolution rates and ship fixes |
What does not change
This is the part being lost in the noise, so it is worth stating plainly.
- Your rankings. Google has not described this as a ranking update, and no source reports crawling or ranking effects on destination sites.
- Your Search Console data. Unaffected. Clicks, impressions and position are still reported the same way.
- Your canonicals, sitemap and redirects. Nothing on your server is modified by this. The rewriting happens entirely on Google’s side.
- Where the user lands. Same page as before, one extra invisible hop earlier.
There is nothing to fix on your website. Anyone selling a “goto readiness audit” is selling you nothing.
What genuinely breaks
The real impact is on measurement, and specifically on third-party tools that read the results page programmatically.
Rank trackers got much more expensive
To know the real destination, a tool now has to follow an extra redirect for every link. Nozzle reported needing 500 to 1,000 requests to resolve a single keyword’s rankings. That cost lands on the provider, and eventually on subscription prices.
AI Overviews were hit hardest
The rollout landed unevenly across the results page. DataForSEO’s own tracking, published 27 August, showed organic results almost fully resolvable while AI Overviews lagged badly — at least one unresolved link in roughly 52% of AI Overview SERPs, against 0.01% for organic. A provider update the following day addressed those cases.
Domain-level filtering can fail silently
This is the subtle one. Independent measurement by Hanns Kronenberg across 1,040 keywords and 44,072 classified URLs found that where the real destination is missing from the page source, the domain field reads google.com. Anyone filtering SERP data by domain therefore loses those results without any error — the rows simply are not there.
See it yourself in thirty seconds
Behaviour differs between sessions, which is why some people insist they cannot reproduce it.
- Open an incognito window — logged-in sessions may or may not show the substitution.
- Search for anything.
- Hover over a result and look at the status bar, bottom left.
- If it reads
google.com/goto?url=…rather than the destination domain, you are seeing it.
You can also right-click a result, copy the link address, and paste it somewhere to inspect the token.
What to actually do
- Nothing to your site. Genuinely. There is no technical change required.
- Lean harder on first-party data. Search Console and GA4 are unaffected because they are not scraping anything. This whole episode is an argument for sanctioned data sources over tools reading a page Google can restructure whenever it likes.
- Ask your rank tracker how they are handling it. Providers adapted at very different speeds. If yours has not said anything publicly, that is worth noticing.
- Annotate your reports. Mark late August so that six months from now nobody misreads a tooling artefact as a ranking event.
- Treat AI Overview link data with suspicion for now. That is where resolution has been least reliable.
The broader pattern
This is one step in a longer sequence: Google has been progressively closing the public results page as a data source for years. Removing the &num=100 parameter, restricting result counts, and now wrapping links in opaque tokens all point the same direction.
The strategic read is not “panic about goto”. It is that SEO measurement built on scraping is on a long-term downward slope, and measurement built on Search Console, analytics, and your own server logs is not. That has been true for a while. This just made it more obvious.
Related: which metrics actually matter, what belongs in an SEO report, and GA4 troubleshooting.
Check the things you can actually control
Google will keep changing its results page. Your own technical setup is the part that stays in your hands — the analyzer runs around 40 checks on any URL in about ten seconds.
No sign-up, no email, no limit.