Google's expanding use of internal goto redirects has pushed SISTRIX to resolve every disguised search link individually, adding a server request for each one, according to a monthly newsletter the search visibility platform sent to subscribers on August 31, 2026. The German measurement company also used the update, written by founder and managing director Johannes Beus, to detail a new chat feature attached to its AI Deep Dives product and to comment on Anthropic's move to watermark AI-generated text, a step that follows obligations already in force under the European Union's AI Act.

The newsletter arrived as a routine monthly digest, the kind SISTRIX has sent for years to summarize changes in search measurement and the wider ecosystem it tracks. But this edition landed in the middle of an unusually active month for Google's search infrastructure, one in which the company narrowed several of the channels that independent tools use to observe its results pages. SISTRIX's account, grounded in its own crawling operation, offers a measurement vendor's perspective on what those changes cost in practice.

According to the SISTRIX newsletter, Google has spent roughly a year making it progressively harder to collect search results at scale. The most visible instance of this, the newsletter states, was the retirement of the num=100 parameter, a change that let tools request 100 organic results in a single query and that Google eliminated on September 14, 2025. PPC Land reported at the time that the removal forced platforms including Semrush to make ten separate requests for the volume of data one request had previously returned.

Since then, SISTRIX writes, barely a week has passed without some smaller or larger adjustment from Google affecting how search results can be captured externally. The newsletter singles out a mechanism the company has been observing for several weeks: so-called goto-URLs. Rather than exposing a destination address directly in a search result, Google routes clicks through an internal /goto address from which the actual target cannot be read until the link is followed. According to the newsletter, this behavior had been running at small scale for some time, and Google has now begun rolling it out more broadly. SISTRIX states that Google confirmed the change to Barry Schwartz of Search Engine Roundtable.

That confirmation lines up with reporting PPC Land published in the same period. Google confirmed on August 26, 2026 that it was rolling out google.com/goto passthrough links inside search results, replacing the destination URLs that measurement tools had read directly from the page for years. Rank tracking firm Nozzle recorded the change reaching near total coverage across several residential IP providers on the same day, after roughly four months of intermittent testing. Alex Greenland first documented the pattern in a post on June 23, 2026, describing rewritten links that follow a consistent structure: a google.com/goto?url= address encoding the true destination, which redirects to it only once clicked. Brodie Clark, who publishes the SERP Alert newsletter, confirmed the pattern applied to both organic and paid listings in an analysis dated July 2, 2026.

How SISTRIX Is Adapting

SISTRIX describes its own response in direct terms. The company states it is already resolving these addresses to their real targets and is currently scaling that process, while acknowledging that every resolved URL adds an additional request to its infrastructure. Measuring search results, the newsletter states, becomes somewhat more complex and more expensive as a result, though not impossible. SISTRIX adds that its update cycles will likely run somewhat slower than usual over the following days while the transition is underway, but says the completeness and comparability of its underlying data will not change.

That framing matches what PPC Land's reporting found from the rank-tracking side of the industry. Derek Perkins, founder and chief executive of Nozzle, had been monitoring the redirect behavior for months before the August confirmation, comparing it in earlier commentary to older server-side redirect wrapping that surfaced when JavaScript was disabled. On August 26, Perkins published a chart covering roughly four months of testing that showed a sharp change in scale, moving from intermittent appearances to something close to universal coverage. Google has not issued a public statement explaining the goto rollout beyond the confirmation given to Schwartz, and the company has not commented on Nozzle's findings.

A Measurement Company's Reading of Google's Motive

Beus offers his own interpretation of why Google is tightening access this way, framing it as commentary rather than confirmed company strategy. In his view, these measures are aimed primarily at competitors to Google, naming ChatGPT as an example of a service that operates without its own search index yet still draws on Google's results. Tools such as SISTRIX, which he says bring transparency to the search market, are collateral damage in that effort rather than the actual target. He draws a direct comparison to the num=100 removal, arguing the same dynamic applied there.

Beus states he expects further measures of this kind. But he argues the larger shift is not how difficult Google makes data collection, but how the search result itself is changing: where a ranked list of links once stood, he writes, a generated answer increasingly takes its place, and the relevant unit of measurement shifts away from organic position toward what he calls mentions and, in some cases, citations. He states this is why SISTRIX has for some time been measuring both in parallel, tracking classic search results as before while its AI Check product covers responses from AI Overviews, AI Mode, ChatGPT, Gemini, Perplexity and Copilot. Beus writes that this makes the underlying data collection more demanding, but also raises the value of the insight it produces.

The framing that AI answers shift optimization away from ranked positions and toward citations recurs across recent industry commentary that PPC Land has tracked. A March 2026 survey of 45 academic studies on generative engine optimization, published as an arXiv preprint and covered by PPC Land, concluded that the strongest evidence available shows content already retrieved by a generative system can influence how it is cited, while far weaker evidence exists that any technique reliably improves whether a page is retrieved at all, or that any of it produces a durable effect on clicks or revenue.

AI Deep Dives Gains a Chat Interface

The newsletter's second major item concerns a product update rather than a Google policy shift. SISTRIX describes its AI Check tool as a feature that requires only a domain, brand or entity name as input, drawing on data the company already collects continuously and displaying visibility results immediately, without a waiting period or a dedicated project setup. Several weeks before this newsletter, SISTRIX introduced AI Deep Dives within that tool, alongside a companion product called Prompt Research aimed at keyword research for AI-driven search. According to the newsletter, Deep Dives combine SISTRIX's own deterministically calculated data with what Beus calls the interpretive strength of AI, and eleven separate analyses have been available since launch, covering areas from brand perception to competitive positioning to content opportunities.

The newsletter identifies a gap in that original launch: once an analysis finished running, a user had no way to ask a direct follow-up question about it. Anyone who wanted to know, for instance, which piece of content might address a particular finding had to take that question outside SISTRIX entirely, into a general-purpose chat tool with no knowledge of the underlying analysis.

According to the newsletter, a new chat function closes that gap. Every completed Deep Dive now displays an input field beneath it, through which users can ask targeted follow-up questions about that specific analysis. SISTRIX states that the underlying AI has full access to the complete data context behind the analysis and responds with corresponding specificity. As an example, the newsletter describes a user who finds substantial criticism on a particular topic within a brand perception analysis; that user can now ask directly what can be done about it and receive prioritized actions rather than a plain inventory of findings. SISTRIX adds that the tool also proactively suggests relevant follow-up questions through what it calls suggestion chips. Further detail on the feature, according to the newsletter, is available in the company's changelog.

The chat function is available immediately to all SISTRIX users at no additional cost, the newsletter states. SISTRIX also points subscribers to a separate tutorial describing how multiple Deep Dives can be combined into a complete action plan for what the company terms GEO analysis, framed as a five-step process.

Watermarking Enters a Wider Regulatory Context

The newsletter's third item turns to a different company entirely. SISTRIX notes that Anthropic has announced it will begin marking all text and files produced by its own AI models with a watermark, and frames the resulting attention as somewhat overstated given that, in the newsletter's assessment, most AI providers will soon be doing the same. The newsletter points out that Google already has a comparable technology in SynthID, and argues the underlying driver is regulation rather than voluntary practice: the EU AI Act, according to the newsletter, obligates providers including OpenAI, Meta, Microsoft and Google to label AI-generated content accordingly, meaning that in the coming months, AI content broadly will carry markings, not only material produced by Claude.

That regulatory context is more developed than the newsletter's brief mention suggests. PPC Land's own reporting shows Anthropic disclosed its watermarking plans in detail, stating that Claude models launched on or after August 2, 2026 support marking at launch, with generated text carrying embedded watermarks and generated files carrying signed provenance metadata under the C2PA standard for supported formats including .svg, .png and .jpg. Anthropic's own documentation states that older models are not covered at launch, and that the company is working to extend marking support to them without giving a completion date, though an EU grandfathering provision sets December 2, 2026 as the deadline for systems already on the market to bring marking and detection into conformity.

Google's own position in this regulatory sequence predates Anthropic's announcement by several weeks. Google signed the EU's Code of Practice on Transparency of AI-Generated Content on July 24, 2026, two days after the date that would have conferred an automatic presumption of compliance, according to PPC Land's coverage of the signing. Meta signed five days later, on July 28, reversing a 2025 refusal to sign an earlier, related code. Article 50 of the EU AI Act, the provision governing these transparency duties, became applicable on August 2, 2026, carrying penalties of up to 15 million euros or 3 percent of worldwide annual turnover, whichever is higher, for non-compliant providers. By the end of July 2026, roughly 190 companies and organizations had signed the code, according to PPC Land's reporting.

Beus offers three observations on the shift. First, he argues a watermark is not, by itself, a negative quality signal: anyone who has AI merely translate or correct their text will carry the same marking going forward. Second, he notes that Google's official position is that ranking depends on content quality rather than AI origin, meaning that in principle, purely AI-generated content should have the same chance at good rankings even where AI authorship is confirmed, though he states he remains skeptical of that outcome in practice. Third, he predicts that flagging and labeling AI content will become close to universal across the web, not confined to Google, and states his expectation that when people have the choice, they will prefer content made by humans over content made by machines. He concludes that human-made content will regain value in the coming months.

Other Items in the Newsletter

The newsletter also summarizes three shorter items. Google completed its August 2026 spam update, which the newsletter states began on August 18 and ended on August 21, describing it, per Google's own characterization, as a normal spam update with no particular focus on link spam or site reputation abuse. SISTRIX notes mixed reaction across webmaster forums, with some site owners reporting significant traffic drops and others observing that large generalist sites benefited more than specialized niche publishers. SISTRIX states it did not observe particular crashes or surges of its own. That description is broadly consistent with PPC Land's own coverage, which recorded the update's start at 09:27 or 09:28 Pacific time on August 18, 2026, applying globally and across all languages, following roughly seven and a half weeks without a confirmed algorithmic update since the prior spam update closed on June 26.

The newsletter also references renewed comments from Google's John Mueller on generative engine optimization, made during a discussion on Bluesky. According to the newsletter, Mueller reiterated that there is nothing special that needs to be done specifically for generative AI answers in search, and that GEO is ultimately the same discipline as SEO. Beus pushes back on the framing, noting that what Google says and what Google does have not always aligned historically, and that Google is no longer the only significant force in the search market, a qualification he says tends to be omitted from such statements. He concludes that Google's position on GEO need not be treated as final.

Finally, the newsletter covers OpenAI's expansion of ChatGPT Ads to 31 European countries, including Germany, France and Spain. According to the newsletter, ads display only to users on the Free and Go tiers, while Pro, Plus and Enterprise subscribers remain ad-free. Access initially runs through OpenAI's advertising team and agency partners, with self-service access through Ads Manager expected to follow over the summer, the newsletter states. SISTRIX links the timing to OpenAI's financial position, noting that the company is projected, according to current forecasts cited in the newsletter, to need 111 billion US dollars by 2030.

PPC Land's own coverage of the European rollout supplies additional detail the newsletter does not include. OpenAI confirmed on August 18, 2026 that ChatGPT Ads was coming to 31 European markets, describing it as the company's largest expansion to date, with sponsored placements beginning to appear across those markets on August 24. That activation followed a smaller expansion into Brazil and Mexico on August 17, and both followed the pilot's original launch in the United States on February 9, 2026. By August 29, 2026, OpenAI stated it had added 32 new markets in five days with the addition of India, bringing the publicly counted total to 41, according to a LinkedIn post from Benji Shomair, OpenAI's vice president of monetization.

Why This Matters for Marketers

For anyone who relies on rank tracking, share-of-voice dashboards or competitive intelligence tools, the SISTRIX newsletter functions as a status report from inside the plumbing that makes those tools work. When a measurement vendor states plainly that its cost structure has changed because of a change on Google's side, that has direct consequences for the accuracy and pricing of the tools marketing teams already use, even where no individual campaign or account is directly affected. PPC Land's broader coverage of the goto rollout situates SISTRIX's experience within a wider pattern: the same week Google confirmed the goto redirect expansion, it also rolled out generative AI performance reporting to every Search Console property worldwide, a report that shows how often a page appears inside an AI answer but withholds the click and query data that would let a publisher judge whether that visibility is worth anything.

The watermarking item carries a separate, longer-running significance. As more advertising and publishing platforms sign the EU's Code of Practice on Transparency of AI-Generated Content and Article 50 obligations take full effect, disclosure requirements that once applied narrowly to specific ad formats are extending across the wider content supply chain that marketers depend on for organic and paid visibility alike. Research PPC Land has covered elsewhere found that AI-labeled advertising content can see meaningfully lower click-through rates than unlabeled equivalents, a data point that gives some weight to Beus's prediction that human-authored content may see a value premium as labeling becomes routine.

Timeline

  • September 14, 2025 - Google eliminates the num=100 SERP parameter, forcing measurement tools into roughly ten times more requests to collect the same volume of search results.
  • July 18, 2025 - The European Commission publishes guidelines for general-purpose AI model obligations under the EU AI Act.
  • December 18, 2025 - Google enables SynthID-based video verification inside the Gemini app.
  • June 23, 2026 - Alex Greenland documents the first google.com/goto redirect URLs appearing in search results.
  • February 9, 2026 - OpenAI's ChatGPT advertising pilot begins running in the United States.
  • July 2, 2026 - Brodie Clark documents the goto redirect pattern across both organic and paid search listings.
  • July 20, 2026 - The European Commission publishes finalized Article 50 implementation guidelines and the Code of Practice on Transparency of AI-Generated Content.
  • July 24, 2026 - Google signs the EU Code of Practice on Transparency of AI-Generated Content.
  • July 28, 2026 - Meta signs the same code, reversing a 2025 refusal.
  • August 2, 2026 - Article 50 transparency obligations under the EU AI Act become applicable across the bloc.
  • August 18, 2026 - Google logs the August 2026 spam update on its Search Status Dashboard, applying globally and to all languages.
  • August 18, 2026 - OpenAI confirms ChatGPT Ads is expanding to 31 European markets, its largest expansion to date.
  • August 21, 2026 - The August 2026 spam update's rollout window closes, per SISTRIX's newsletter account.
  • August 24, 2026 - Sponsored placements begin appearing inside ChatGPT across the 31 European markets.
  • August 26, 2026 - Google confirms the goto redirect rollout to Barry Schwartz of Search Engine Roundtable; Nozzle records near-total coverage across residential IP providers.
  • August 29, 2026 - OpenAI adds India, bringing its publicly counted ChatGPT Ads market total to 41.
  • August 31, 2026 - SISTRIX publishes its August monthly newsletter, detailing the goto-URL resolution process, the new AI Deep Dives chat feature, and commentary on AI content watermarking.

Summary

Who: SISTRIX, the German search visibility platform led by founder and managing director Johannes Beus, alongside Google, Anthropic and OpenAI as the companies whose actions the newsletter describes.

What: SISTRIX's August 2026 monthly newsletter details how the company is resolving Google's expanding goto-URL redirects at added computing cost, describes a new chat function inside its AI Deep Dives product, and comments on Anthropic's AI content watermarking announcement in the context of EU AI Act obligations.

When: The newsletter was sent August 31, 2026, describing developments including Google's August 26, 2026 confirmation of the goto rollout and the August 2, 2026 start of EU Article 50 enforcement.

Where: The developments span Google's global search infrastructure, SISTRIX's operations from Bonn, Germany, and regulatory obligations centered on the European Union.

Why: The newsletter matters to search marketers because it documents, from a measurement vendor's own operational experience, how Google's infrastructure changes are raising the cost and complexity of independently tracking search visibility, at the same time as new regulatory labeling requirements begin reshaping how AI-generated content is disclosed across the industry.