Google Hid the Destination URL: Resolving One Ranking Now Costs 500 to 1,000 Requests

On August 26, 2026, Google confirmed to Search Engine Roundtable that it is rolling out google.com/goto redirects: the link behind an organic listing no longer carries the destination address. It carries an encoded Google address that issues a 302 to the real page on click. Derek Perkins, CEO of rank tracking firm Nozzle, measured “near 100% rollout across several residential IP providers” the same day, and supplied a number nobody wanted: resolving every link in a single five-page ranking takes 500 to 1,000 requests.

This is the third cut in twelve months. On September 14, 2025 Google removed the num=100 parameter, forcing vendors to issue ten requests where one had returned a hundred results; Semrush confirmed a tenfold operational cost increase the next day. January 2026 brought JavaScript gating. The legal track runs in parallel: Google sued SerpApi on December 19, 2025 under DMCA Section 1201, seeking $200 to $2,500 per act; Judge Yvonne Gonzalez Rogers dismissed every claim on July 20, 2026, and SerpApi asked the court to end the case on August 24 — two days before goto shipped. Technical measure and legal theory address the same question: may a third party read Google's results page?

Vendors split into two camps. DemandSphere published on August 27 that countermeasures are deployed, organic ranking data is unaffected, and SERP feature URLs are still being patched. Nozzle says HEAD requests are refused, leaving only a GET that stops at the redirect, with rate limiting rather than bandwidth as the binding constraint. The last round tells you how this ends: AccuRanker limited tracking to the top twenty, SISTRIX discontinued desktop results updates, and Ahrefs restored top-100 tracking in late October 2025 without disclosing the method. The category is sliding from measurement toward estimation.

For a small business, the impact is not on rankings but on the report that lands in your inbox every month. If you pay for ranking or visibility numbers, at least one of price, sampling depth, or refresh frequency will move in the next twelve months, and the report layout probably will not. Below: how these numbers are produced, which parts you should not trust, and how to build a baseline with no SaaS subscription.

What Happened: One Link, Four Months of Rolling Tests

The first public sighting was June 23, 2026, when Alex Greenland recorded links rewritten to google.com/goto?url=[encoded]; on July 2, Brodie Clark (SERP Alert) documented the same behaviour across organic and paid listings and confirmed a 302 temporary redirect. The same day Barry Schwartz could not reproduce it; by July 8 he had found only three people who could — the signature of a bucketed rollout, and the first numeric trap.

Google's statement is one sentence, describing technical measures against “evolving forms of abuse.” What it omits matters: no timetable, no geographic breakdown, no technical documentation.

The key point is that the encoding cannot be decoded locally, so providers must follow the redirect, and Perkins says Google refuses the cheaper HEAD request. Whether the redirect is static per page — which would let an operator build a lookup table once — went unanswered, and that is the line between a one-off engineering cost and a permanent per-query tax.

Where the Number Breaks Down

Denominator one: who computed 500 to 1,000? It is an estimate a vendor CEO gave on X for a single five-page ranking, with no methodology and no distribution — an order-of-magnitude reference. Converting it into “cost per keyword rises Nx” for a proposal is using a tweet as a financial assumption. “Near 100%” has the same problem: the population is that firm's own residential IP nodes, not the world's results pages.

Selection bias: everyone talking sells the same product. Both the party saying “this will get expensive” and the one saying “we are unaffected” are rank trackers: the first message supports a price increase, the second supports renewal — there is no neutral measurer in this conversation.

The time series has been cut. PPC Land notes that ranking datasets from before June, during the test window, and after rollout may not measure the same thing: sampling depended on which infrastructure a provider used and whether affected pages were discarded or kept — Nozzle admits it discarded goto-only pages during the test.

What Each Reader Should Do Now

Brand owners and SMB operators

• Ask your vendor three questions: how often is data refreshed, how deep is tracking, did sampling change after August? A report that cannot answer should not drive decisions.
• Do not sign “top three ranking” KPIs this year; shift part of the budget from buying numbers to keeping your own — Search Console and GA4 data belong to you.

Marketing and SEO practitioners

• Draw a line on 2026-08-26 and annotate cross-line comparisons as “measurement method may have changed”: reduced tool visibility and real ranking movement look identical on a chart.
• Move to impressions and clicks from the Search Console API; the denominator is at least Google's own.

Developers and agencies

• Audit every script that parses Google results pages: code assuming the anchor holds the destination will return google.com as of August 26.
• Add a pre-check to domain matching: flag goto wrappers as unresolved rather than writing google.com, which would attribute share of voice to the wrong domain.

Comparison: Five Measurement Vendors and What to Ask Them

A vendor willing to state what it changed is more credible than one claiming to be unaffected.

What Nobody Will Tell You

Google's statement never says “scraping.” It says “abuse,” naming no category of automated client: one sentence covers AI companies, rank trackers and phishing detection, and nobody can argue they are out of scope — no documentation means no predictability.

Tool vendors are not the only casualty. A user hovering over a result used to see where it led; now they see google.com, which makes phishing and spam harder to spot.

Search Console is not clean either. An impressions logging error that began on May 13, 2025 was only acknowledged in April 2026, and impressions and clicks dropped again around August 12, 2026. Use it as a baseline not because it is accurate, but because it is the same ruler.

The No-Subscription Route for SMBs and Agencies

A minimal Laravel scheduled service is enough:

• Pull the GSC API into your own database daily. Query, page, impressions, clicks, average position into MySQL — the only long series you own.
• Keep a “ruler change” events table. Store 2026-08-26 goto and 2025-09-14 num=100, and have charts draw the line automatically.
• Spot-check 10 keywords manually each week. Incognito, fixed location, screenshots archived; a fixed method beats sampling that changes monthly.
• Add goto detection to crawler scripts. On google.com/goto, flag as unresolved and alert instead of writing it in silently.
• Put a methodology line on every report. Data source, sampling depth, refresh frequency, known breakpoints — trust comes from that line.

To connect this measurement layer to your site, get in touch:

• Email: [email protected]
• Phone: 0916-224-047
• LINE: @ufv9089p

FAQ

Will the goto redirect affect my rankings?

No. It changes how links are presented, not the ranking logic; the 302 signals the wrapper is not meant to replace the target address in any index. What changed is who can read that address.

Will my Google organic traffic in GA4 change?

Google has published nothing on referrer behaviour through the redirect layer. Compare GA4 google / organic against Search Console clicks from August 26 onward; if a gap appears, assume a measurement-layer problem first.

Should I still pay for a rank tracker?

Yes, but change how you use it: relative trend and competitor comparison, not absolute truth, and no comparison across 2026-08-26. The criterion is whether the vendor will state what it changed.

Does this also hit AI visibility monitoring tools?

Same direction, different mechanism: those tools query ChatGPT, Gemini and Perplexity rather than parsing Google results; but measurement still rests on a platform permitting reads, and when it changes, your history breaks.

Our Take

The mainstream reaction is “vendors always adapt.” My read is the opposite: after this round, ranking and visibility tools will collectively slide from measurement to estimation, and the dashboard will not say so. When num=100 was removed, the industry's answer was cutting tracking depth to 20 and freezing datasets — the interface stayed identical, the meaning did not; goto is a second multiplier on top. My prediction is that within 12 to 18 months, mainstream tools will ship interpolated rankings and model-estimated visibility under exactly the same column names as measured data.

For a studio like ScriptWalker, the opportunity is not swapping a client's rank tracker but building the client's own measurement layer: daily GSC and GA4 ingestion, a ruler-change events table, auto-generated methodology notes, and a dashboard that explains why a line broke on August 26. What you sell is data ownership and comparability.

Sources

Primary

• Search Engine Roundtable — Confirmed: Google Search Rolling Out google.com/goto Tracking Parameters (2026-08-26, includes Google spokesperson statement)
• DemandSphere blog — An update from DemandSphere on Google's goto redirects (2026-08-27)
• Derek Perkins (Nozzle CEO) on X (2026-08-26, rollout chart and the 500–1,000 request estimate)
• Alex Greenland on X (2026-06-23, first public sighting)
• Brodie Clark (SERP Alert) on X (2026-07-02, 302 behaviour on organic and paid)
• Google Search Console API documentation

Third-party

• PPC Land — Google forces rank trackers into 500 to 1,000 requests per query (2026-08-27)
• Search Engine Land — Google confirms deploying goto URL redirects to search results links (2026-08-26)
• PPC Land — Google eliminates num=100 SERP parameter, forcing cost increases for SEO tools (2025-09-14)
• PPC Land — Semrush confirms significant operational impact as Google eliminates num=100
• PPC Land — Google loses DMCA bid to treat search scraping like DVD piracy (2026-07-20)
• PPC Land — Google Search Console impressions bug ran for nearly a year unnoticed
• SparkToro — In 2026, Less than One Third of Google Searches Still Send a Click