Google confirmed on August 26, 2026 that it is rolling out google.com/goto passthrough links inside search results, replacing the destination URLs that measurement tools have read straight from the page for years. Rank tracking firm Nozzle recorded the change at near total coverage across several residential IP providers on the same day, after roughly four months of intermittent testing.

For most of the history of Google Search, the anchor behind an organic listing carried the destination address. A tool reading the results page could parse that address without contacting the destination site. That assumption underpins rank trackers, share of voice dashboards, competitive intelligence platforms and a long tail of scripts that marketing teams run in house.

The new behaviour replaces the address with a passthrough link on google.com. According to a X post published by Alex Greenland on June 23, 2026, the rewritten links follow a consistent pattern: "Result link URLs are now being rewritten to google.com/goto?url=[custom base64-like encoding of URL or Google ID]. Upon click, they redirect to their proper destination."

Brodie Clark, who publishes the SERP Alert newsletter, documented the same behaviour across both paid and organic listings on July 2, 2026. "There is something bizarre happening with URLs in search results right now - both ads and organic listings," he wrote. According to his post, listings that had shown a naked address, sometimes carrying auto-tagging parameters, were instead resolving through a goto parameter that issued a 302 response to the destination. The 302 status matters technically, because a temporary redirect signals to crawlers that the wrapper is not intended to replace the target address in any index.

Clark also recorded a second-order signal. According to the same post, Ahrefs was picking up the pattern in its own tracking, with the number of queries showing it climbing over a period of days.

A four-month test window

The sequence runs from late June to late August. Greenland's observation on June 23 was the first public sighting. By July 2, Clark reported the pattern appearing on every site he checked. Barry Schwartz of Search Engine Roundtable replied the same day that he could not reproduce it, and when he published on July 8, 2026 he wrote that he had identified only three people who had noticed the behaviour and still could not see it himself.

That gap between reporters and non-reporters is characteristic of a bucketed rollout. Two comments on the July article suggest the effect was already reaching commercial tooling. According to a comment from Justin Mosebach, Places Scout ranking reports were beginning to show the impact. Aleksandr Sovgir, commenting from the Serpdino rank tracker, described a different symptom: verification interstitials returning on standard search URLs rather than goto wrappers, arriving with a 200 status code and a correct page title but no results container.

Derek Perkins, founder and chief executive of Nozzle, had been watching the behaviour for longer. On July 3, 2026 he replied to Clark that his team had been seeing it for months, comparing it to older server side redirect wrapping that appeared when JavaScript was disabled, and adding that Nozzle discarded affected results pages and refetched.

The confirmation

On August 26, 2026, Perkins published a chart covering roughly four months of testing and reported a sharp change in scale. "It was widely reported the first week of August from the first major test, and as of this morning, we're seeing near 100% rollout across several residential ip providers," he wrote. He described the underlying mechanism as anti-bot work: "Google has been testing anti-bot measures that eliminate all client side links in favor of google. com/goto... server side redirects."

Nozzle had been discarding goto-only results pages so that customer data was not polluted with opaque links, according to the post, and the share affected had been low enough not to matter operationally. That is the part the August reading changed.

Schwartz published later the same day, and Google confirmed the rollout in a statement. According to a Google spokesperson quoted by Search Engine Roundtable: "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 statement is notable for what it omits. It names abuse rather than scraping, and it does not identify rank trackers, artificial intelligence firms or any specific category of automated client as the target. Google offered no timetable, no geographic breakdown and no technical documentation.

Why decoding is not the workaround

The engineering problem sits in the encoding. According to Perkins, the goto links cannot be decoded locally, which means providers have to follow the redirect rather than read the address. "While that works in small scale tests, Google seems likely to make that tough, since each SERP will have hundreds of links to decode," he wrote.

Two proposed workarounds surfaced immediately in replies to his post, and both ran into the same wall. Glenn Gabe asked whether tools could still pull the addresses. Mic King asked whether HEAD requests would be enough, since a HEAD request retrieves headers without the response body and would return the redirect target cheaply.

Perkins answered that Google does not let the HEAD request through. A GET request that stops at the redirect works, avoiding a full page fetch, but the volume is the constraint. According to his reply, resolving all links for a single five-page ranking takes between 500 and 1,000 requests. Rate limiting then becomes the binding limit rather than bandwidth or storage.

Dominic Woodman raised the scale question from the other direction, asking whether the redirect is static per page, which would allow a large operator to build a lookup table once and reuse it. That question was not answered in the thread, and it is the difference between a fixed engineering cost and a permanent per-query tax.

A measurement problem in both directions

Following a redirect is not a neutral act. Joe Manna asked in reply whether resolving a goto link registers as a click, and whether click reports would start carrying noise generated by rank trackers. Perkins responded that it would be interesting to see whether Google correlates those events with client side analytics.

That question runs in both directions. If resolution registers server side, the party doing the measuring becomes a source of the traffic being measured. If Google cross-references those events against client side signals, the wrapper doubles as a detection surface for automated clients, separate from its function as a link.

The paid side of the results page had already moved in a comparable direction. In a post published on May 11, 2026, Clark documented that hovering over an advertisement in search results had begun to show the exact destination address, where it previously had not, while interacting with the same link produced a google.com/aclk address. His working theory was that the presentation change let advertisement data filter into Google Search Console over time, potentially contributing to impression spikes on merchant listing surfaces.

Clark was explicit about the limits of that reading. According to his post, it is difficult to say with certainty whether the issue is ongoing or already patched. He tied the question to a broader pattern of unexplained Search Console impression behaviour, which Google formally acknowledged in April 2026 as a logging error dating back to May 13, 2025. Search Console has also frozen performance data for several days in October 2025, lost Discover data on individual days in May 2026, and dropped impressions and clicks again from around August 12, 2026.

The num=100 comparison

Perkins drew the historical parallel himself. If the rollout holds, according to his post, the industry is looking at something akin to the num=100 change from the previous autumn.

That comparison carries specific numbers. Google eliminated the num=100 parameter on September 14, 2025, forcing platforms to issue ten requests where one had previously returned a hundred results. Semrush confirmed the tenfold operational cost increase within a day. The wider industry response ran through late September, with AccuRanker limiting tracking to the top twenty results and SISTRIX discontinuing desktop results page updates. SerpApi built a workaround through its Google Light Fast API, and Google restricted that endpoint to three organic results within weeks. Ahrefs restored top 100 tracking at the end of October 2025 without disclosing the method.

The structural difference is worth stating plainly. The num=100 change multiplied the number of results pages a provider had to fetch. The goto change multiplies the number of requests needed to interpret a single results page once fetched. Both raise cost per tracked keyword, but they compound rather than substitute.

Peter Mindenhall, replying to Clark on July 2, noted that the srsltid parameters already in use had been sufficient to confuse rank tracking tools previously. Akansha Agarwal added that intentional deployment would carry consequences past search engine optimisation, into artificial intelligence and search data collection generally. Joe Kizlauskas framed it more bluntly in the August thread, arguing that reported ranking volatility since the spam update reflected reduced tool visibility rather than movement in results, and that the change concerns third party access to Google's index rather than bots as such.

The litigation running alongside

Schwartz linked the rollout to Google's legal position, writing that it is likely an attempt to make scraping harder for SerpApi, against whom Google has litigated without success so far. That is his assessment rather than a company statement, and Google's spokesperson made no reference to any case.

The docket supports the timing observation without confirming the motive. Google sued SerpApi on December 19, 2025under Section 1201 of the Digital Millennium Copyright Act, alleging circumvention of its SearchGuard anti-bot system and seeking statutory damages of 200 to 2,500 dollars per act. SerpApi responded publicly in January 2026 and moved to dismiss on February 20, 2026, arguing that SearchGuard is a website control rather than a copyright access control.

Judge Yvonne Gonzalez Rogers dismissed every claim on July 20, 2026, killing the general case with prejudice and leaving open only claims tied to specific licensed works. Google filed an amended complaint on August 10, 2026, dropping its Shopping and Maps theories. SerpApi asked the court to end the case on August 24, 2026, two days before the goto confirmation, with a hearing set for September 29, 2026. A parallel action brought by Reddit against SerpApi, Oxylabs, AWMProxy and Perplexity AI began on October 22, 2025 and also turns on SearchGuard.

A technical measure and a legal theory address the same access question through different instruments. The court can rule on whether reading a results page is circumvention. It cannot rule on whether a link has to contain an address.

Why this matters for marketers

The immediate consequence is arithmetic. Every organic position report that a marketing team consumes is assembled from parsed results pages, and the parsing step now requires network calls it did not require in June. Perkins put the figure at 500 to 1,000 requests for a single five-page ranking. Providers that pass costs through will price that in; providers that do not will absorb margin or reduce depth, as several did after September 2025.

The second consequence is data continuity. Nozzle discarded goto-only results pages during the test period, and Clark reported that SERP Lens users had prompted a similar decision. Discarding is viable while affected pages are a minority of fetches. At near total coverage, discarding means no data at all, so the alternative is resolution at scale, which raises the click contamination question Manna posed.

The third consequence concerns comparability across time. Ranking datasets built before June 2026, during the test window, and after full rollout may not be measuring the same thing, because sampling depended on which infrastructure a provider used and whether affected pages were kept or thrown away. Reported volatility during the transition is therefore ambiguous by construction, which is the point Kizlauskas made in the August thread.

The fourth sits with publishers and advertisers rather than tool vendors. A redirect hop between the results page and the destination adds a step controlled entirely by Google, on both organic listings and, as Clark documented in May, on advertisement units through the aclk address. Referrer behaviour, latency and the visibility of the destination address to a user hovering over a link all pass through that hop. Clark noted in the July thread that the change removes the ability to preview a destination before clicking, a point Schwartz repeated in his July article in the context of phishing and spam checks.

Google has not published documentation for the goto parameter, has not stated whether the rollout is permanent, and has not described what rate limits apply to resolution. Until it does, the operative numbers belong to the people fetching the pages.

Timeline

Summary

Who: Google confirmed the change through a spokesperson quoted by Search Engine Roundtable. Derek Perkins, founder and chief executive of the rank tracking firm Nozzle, supplied the rollout measurements. Brodie Clark of SERP Alert and SERP Lens documented the behaviour across organic and paid listings, and Alex Greenland reported the first public sighting. Affected parties include rank tracking vendors, competitive intelligence platforms, in house measurement teams, publishers and advertisers.

What: Result links in Google Search are being rewritten to a google.com/goto passthrough address carrying an encoded parameter, replacing the destination URL previously present in the markup. A server side redirect resolves the link on click. According to Perkins, the encoded links cannot be decoded locally, HEAD requests are refused, and resolving all links for a single five-page ranking requires between 500 and 1,000 requests.

When: First reported publicly on June 23, 2026. Observed across organic and paid listings on July 2, 2026. Reported as widely visible in the first week of August 2026. Measured at near 100% coverage across several residential IP providers and confirmed by Google on August 26, 2026.

Where: Google Search results pages, affecting both organic listings and advertisement units, observed across multiple residential IP providers and multiple browsers.

Why: Google described the measure as protection against evolving forms of abuse without naming a target. Practitioners including Clark, Perkins and Schwartz have connected it to scraping by third party tools and artificial intelligence systems. The change lands during active litigation between Google and SerpApi over automated access to search results, with a hearing scheduled for September 29, 2026.