What's Changing in Google Search Results with Google.com/goto Links?

TL;DR

• Google added a new redirect link, google.com/goto, to many search results.
• Nozzle reported a near-full rollout across residential IP providers by August 26, 2026.
• Reading each link now takes far more requests, up to 500 to 1K per check.
• Clicking results still works the same, and Search Console data stays unaffected.

How the redirect actually works

The link format is consistent: google.com/goto?url=[encoded string]. Clicking it triggers a 302 redirect, a temporary forward that sends the browser to the real page. The encoded value does not decode back into a readable URL on its own, so a tool cannot recover the destination without following the redirect.

Brodie Clark documented the pattern on both organic and paid listings in a post on SERP Alert in July 2026, and users confirmed that they've been seeing the redirects for months. Ad units carry a related redirect through a google.com/aclk address. Organic results resolve through goto instead. Both formats replace a link that used to show the destination in plain text.

The 302 status matters for how crawlers treat the link. A temporary redirect signals that the wrapper is not meant to replace the destination in an index. Google's own indexing systems still see the real page. Outside tools reading the results page lose that same shortcut.

Rollout timeline

The change built up over roughly four months instead of appearing all at once. Early reports were inconsistent. Some testers saw goto links right away while others couldn't reproduce them on the same day. That pattern usually points to a staged rollout, split by IP type or account signals, rather than a single global switch.

• June 23, 2026: Alex Greenland posts the first public sighting of the goto link format on X.
• July 2, 2026: Brodie Clark documents the pattern across organic and paid listings on SERP Alert, and notes Ahrefs picking up the same pattern in its own tracking.
• Early August 2026: The pattern becomes widely visible across search results, according to Nozzle.
• August 26, 2026: Derek Perkins reports near full rollout across residential IP providers, and Google confirms the change to Search Engine Roundtable.

Through most of the test window, the share of affected pages stayed low enough for tools like Nozzle to simply discard goto-only pages and refetch them. By August 26, that workaround stopped working, because most pages coming back were goto pages.

Why Google made this change

Google confirmed the rollout to Search Engine Roundtable but kept the explanation short. A company spokesperson gave this statement: "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."

The wording points at automated traffic without naming a category. Google has shipped anti-bot changes to search before, and this fits the same pattern – a change to how a page is delivered rather than a change to what a person sees. Derek Perkins at Nozzle and Brodie Clark at SERP Alert both describe it the same way, as a measure aimed at automated readers of the results page.

Perkins measured the rollout specifically across residential IP providers, the type of network address that looks most like ordinary consumer traffic. That detail matters for anyone running search data collection at scale. Anti-bot systems increasingly weigh IP type as one signal among several, alongside request patterns, headers, and browser fingerprints. A rollout that reaches near full coverage on residential ranges suggests Google is reading more than just obvious data center traffic now.

What it costs to read search results now

Reading a destination address used to take one step, a quick look at the page's code. That step now requires following a redirect instead, and the mechanics of that redirect are built to resist shortcuts.

Derek Perkins shared figures in a post on X. A HEAD request normally retrieves only the response headers, including a redirect target, without downloading the full page. That makes it the cheap way to resolve a link. Perkins said Google blocks HEAD requests on goto links, which forces a full GET request for every link that needs resolving.

A GET request still costs less than loading an entire page, but a single results page carries hundreds of links. Perkins estimated that resolving all the links on a standard five-page ranking report now takes 500 to 1K requests. Rate limits, not bandwidth or storage, become the real constraint at that volume.

Google made a separate change in September 2025, removing the num=100 parameter. That change forced tools to fetch 10 pages of 10 results instead of one page of 100. Semrush confirmed a tenfold jump in operational cost within a day of that change.

The two changes stack rather than replace each other. The num=100 removal multiplied the number of pages a tool has to fetch. The goto rollout multiplies the number of requests needed to read each page once it arrives. A tool built for the search results page of 2024 now pays a cost on both ends of that pipeline.

Gabriele Vitke, Product Marketing Team Lead at Decodo, noted, "This is the second big cost jump this year for teams reading Google's results pages. The num=100 change meant fetching more pages. This change makes every page more expensive to read. Most teams running a simple scraper will need to rethink their public data collection strategies from the ground up."

What changes and what doesn't?

Two things stay fixed through all of this – what a person sees when they search, and what Google actually indexes and ranks. The redirect only changes how a link is delivered in the page markup, not the underlying result.

• Clicking a result still lands on the correct page, the added redirect hop adds only a small delay for most connections.
• Google's index and rankings don't change, this is a presentation layer change to how links appear in the results page.
• Search Console data isn't tied to this change, though Search Console has had separate, unrelated reporting issues over the past year.
• The rollout raises the cost of reading results at scale, it doesn't block access outright.

Who this affects, and how

The added cost doesn't land evenly. It concentrates on tools and teams that were built to read search results pages directly, at volume, without a person clicking through each one.

Rank tracking platforms

Rank trackers rebuild ranking positions by parsing a results page for each tracked keyword, across each tracked location and device. That process depended on reading URLs straight out of the markup. Every keyword now needs its links resolved through a redirect before a report can be built, which multiplies request volume per keyword rather than per report.

AI and answer engine teams

Some AI products pull live search results to ground their answers in current information. Those systems typically query at high frequency and need fast responses. Extra redirect resolution adds both cost and latency to that pipeline, a harder trade-off for a product built around fast answers.

In-house marketing and SEO teams

Teams running their own scripts to check rankings or track competitors are the most likely to see something break with no warning. These scripts are often built once and left running, without monitoring for a change like this. A script written in 2024 to read URLs directly from a results page will now return goto links instead.

Academic research

Literature reviews and citation searches often lean on the same pattern: pulling a page of results and following the links out, whether that's manually or through a script built to speed up a systematic review. Anyone automating that step, gathering source URLs across hundreds of queries for a meta-analysis or a citation network, will hit the same wall. A goto link in place of a direct URL breaks tools built to extract and archive sources cleanly, and it's one more redirect to resolve in workflows that are already slow.

The response across these groups looks similar. Budget for a higher request volume, build or buy infrastructure that treats redirect resolution as a normal step, and monitor for the difference between a goto link and a direct link so a dataset doesn't change shape.

Gabriele Vitke added, "Search data collection keeps getting more expensive to run at scale. Teams that treat this as a one-time fix will hit the same wall again next year. The teams that build flexible infrastructure now will handle whatever change comes after this one."

How Decodo reads through this without the extra requests

Most tools hit this by treating the goto link the way they always treated a normal HREF – click it, follow the redirect, get the real URL. That's exactly the step Google made expensive: no HEAD requests, a forced GET per link, and a five-page ranking report suddenly costing 500 to 1K requests to resolve.

Decodo's Web Scraping API handles this natively – results come back with final destination URLs already resolved, so there's no extra hop and no added cost per link.

Looking at the usage data of the Web Scraping API while collecting publicly available data from Google SERP, it grew by 180% from January, mainly driven by the num=100 parameter removal and increased anti-bot systems. We're projecting that this number will continue to grow, especially within organizations that rely heavily on real-time data from Google SERP.

What can teams that rely on SERP data do?

While the changes are rolling out gradually and might not affect your data collection needs directly just now, it's best to stay prepared. Ultimately, you should:

• Plan for higher request volumes, and budget scraping or tracking costs accordingly.
• Move to proxy and request infrastructure built to handle redirect-heavy pages at scale.
• Watch official Google channels for documentation, since none has been published yet.
• Compare data collected before and after the rollout carefully, since sampling may differ.

Bottom line

Google confirmed the google.com/goto rollout on August 26, 2026. The change adds a redirect step to search result links, blocks HEAD requests on that redirect, and pushes resolution costs to 500 to 1K requests per five-page ranking report. Nothing about what a user sees or clicks has changed. What changed is the cost of reading that same page programmatically, on top of a similar cost increase from the num=100 removal less than a year earlier. Expect further adjustments from tools that depend on this data in the coming weeks, since Google hasn't published documentation or confirmed the rollout is final.