On August 26, 2026, Google confirmed it is rewriting the links inside its search results, replacing each destination address with an encoded passthrough through google.com/goto. Rank tracking firm Nozzle measured the change at near total coverage across several residential IP providers the same day, after roughly four months of intermittent testing.
That sounds like a plumbing story. It is a measurement story, and it lands weeks before Michigan marketing teams lock Q4 budgets using rank reports that quietly changed methodology mid-summer.
What is the google.com/goto link rewrite, and what actually changed?
Result links no longer carry the destination address in the page markup. They carry a google.com/goto?url= wrapper with the target encoded inside, and a server-side redirect resolves it on click. Nothing about your rankings changed. What changed is who can read the results page cheaply.
Alex Greenland reported the first public sighting on June 23, 2026. Brodie Clark of SERP Alert documented the pattern across organic listings and ad units on July 2, 2026, noting the wrapper issued a 302 — a temporary redirect, signaling to crawlers that it is not meant to replace the real address in any index. Barry Schwartz could not reproduce it as late as July 8, the signature of a bucketed rollout rather than a flipped switch.
Google’s confirmation to Search Engine Roundtable was one sentence about deploying “technical measures against evolving forms of abuse.” No documentation, no timetable, no rate limit disclosure.
Why does this cost rank trackers 500 to 1,000 requests per query?
Because the encoding cannot be reversed locally. To learn where a link points, a tool has to follow it. Nozzle CEO Derek Perkins put the cost of resolving every link on a single five-page ranking at 500 to 1,000 requests, and said Google refuses HEAD requests — the cheap method returning headers without a body. A GET that stops at the redirect works, but volume, not bandwidth, becomes the binding constraint.
Here is the mechanism worth internalizing, because it predicts cases nobody has written about yet: the old cost model scaled with pages fetched; this one scales with links per page. When Google removed the num=100 parameter on September 14, 2025, platforms had to issue ten requests where one had returned a hundred results, and Semrush confirmed a tenfold cost increase within a day. That was a page tax. The goto wrapper is a link tax, and the two multiply rather than substitute.
Run that forward and you can predict the damage without waiting for a vendor announcement. Query types that render dense results pages — local packs stacked with map listings, product grids, sitelink-heavy branded SERPs — carry the most links, so they cost the most to resolve and will be trimmed first. After the num=100 change, AccuRanker limited tracking to the top twenty results and SISTRIX discontinued desktop results page updates. Expect the same shape of response, aimed this time at your most link-dense queries rather than your deepest ones.
Does the goto redirect change my analytics or my actual rankings?
Your positions are unaffected, and Google Search Console is unaffected, because Search Console reports from Google’s own logs rather than from a parsed results page. Third-party rank tools are what degrade. Treat any ranking shift you cannot corroborate in Search Console clicks as unverified.
One question remains open: Perkins was asked whether a tool resolving a goto link registers as a click, and whether vendors would start injecting noise into click reports. He did not know, and neither does anyone outside Google. There is also a security cost nobody asked for — with the destination hidden, the standard advice to check a link before clicking no longer works on Google’s own results page.
Why do Michigan catalog and multi-location sites pay the most for this?
Because a link tax scales with keyword count, and Michigan’s economy is keyword-heavy on both ends: large SKU catalogs and dense multi-location service footprints, the two account structures carrying the biggest tracked keyword lists in the state.
An automotive aftermarket or industrial supply distributor in Livonia or Sterling Heights may track thousands of fitment-driven terms, because “brake rotor 2019 Silverado 1500” and “brake rotor 2020 Silverado 1500” are different queries with different intent. A marine catalog in the Grand Rapids corridor has the same problem across model years. An HVAC or dental group with fourteen metro locations tracks the same twenty service terms fourteen times over, because the local pack differs in Troy and in Wyandotte.
Both structures multiply tracked queries — and local-pack SERPs are exactly the link-dense pages that cost the most to resolve. The concrete way this costs money: your vendor renews in Q4 with a price increase, a depth cut from top 100 to top 20, or a frequency cut from daily to weekly, and you approve it without realizing your year-over-year seasonal comparison just lost its baseline. Budgeting for 2027 search work? This belongs in the same conversation as your Detroit-area search strategy line items, not filed under vendor admin.
How should a small team audit its rank reports before Q4?
Seven steps a three-person in-house team can finish this week without new tooling.
- Email your rank vendor one question in writing: did collection method, tracking depth, or refresh frequency change between June 23 and August 26, 2026, and were goto-affected results pages discarded or resolved? Nozzle publicly discarded affected pages during the test window. Discarded pages mean gaps, not zeros.
- Export your full position history through August 31 to CSV and store it outside the vendor platform. If depth gets trimmed at renewal, historical top-50 data may stop being retrievable.
- Pull Search Console clicks and impressions for the same range and set them beside the rank data. Search Console is now your ledger; third-party rank is a directional read.
- Flag every ranking “drop” logged between August 12 and August 26 as unverified pending Search Console corroboration. The August 2026 spam update landed in the same window, and instrument noise is indistinguishable from real movement in a vendor chart.
- Cut your tracked keyword list by 30 to 50 percent. Keep terms that map to revenue, drop near-duplicate fitment variants that move together anyway. This is the single lever that reduces exposure to per-link pricing.
- Add a data-continuity note to any dashboard spanning June through September 2026. Do this before a board deck cites the trendline.
- Check renewal dates on every SEO tool contract and calendar a review two weeks prior. Price and depth changes after the
num=100removal arrived at renewal, not by announcement.
What dates should be on the calendar?
- September 14, 2026 — one-year mark on the
num=100removal. Any year-over-year rank comparison crossing this date compares two different collection methods. - Late September 2026 — when vendors announced depth and pricing changes last cycle. Watch your inbox.
- September 29, 2026 — hearing on SerpApi’s second motion to dismiss. The outcome shapes how hard Google presses technical restrictions.
- September 30, 2026 — Google’s alcohol advertising policy update takes effect, relevant if you run ads for Michigan breweries or hospitality groups.
- October 1, 2026 — Q4 begins. Whatever baseline you hold on September 30 is what you compare against all quarter.
Is this really about AI scrapers, or about the SerpApi lawsuit?
The popular read is that Google built this to strangle SerpApi. That read is too small, and here is where I part company with the consensus.
The timing is suggestive. Google sued SerpApi on December 19, 2025 under Section 1201 of the DMCA, seeking $200 to $2,500 per act of alleged circumvention. Judge Yvonne Gonzalez Rogers dismissed every claim on July 20, 2026; Google amended on August 10; SerpApi moved to end the case on August 24, two days before the goto confirmation. That connection is Schwartz’s assessment, not a company statement.
But a court cannot rule that a link must contain an address, and Google needs no verdict to change its own markup. The more durable effect is on the measurement layer everyone else depends on. A vendor with capital absorbs 500 to 1,000 requests per query. A boutique competitive-intelligence tool does not. Consolidate the supply of ranking data into three or four large vendors and Google gains a smaller, more legible set of counterparties — a structural outcome, not a litigation tactic.
Note what the “abuse” framing omits: no target category, no timetable, no geography. That is optionality, not evasion. Plan for the restriction to persist and for resolution costs to reach you, because that is the outcome your budget has to survive either way. Teams rebuilding reporting around first-party signals should start with their analytics and conversion tracking setup rather than shopping for a replacement rank tool.
Frequently asked questions
Do I need to change anything on my website?
No. The rewrite happens on Google’s results page, not on your site. There is no redirect to fix, no canonical to update, no crawl directive to change. The 302 status Clark documented signals a temporary hop, not the kind of redirect substituted for your address in an index. Your work here is in reporting, not code.
Will my Google Analytics referral data break?
No evidence of that so far, and organic attribution in GA4 leans on more than the raw referrer header. That said, Google has not published a referrer policy for goto links, so verify rather than assume: pull a week of organic sessions and confirm source and medium classification looks normal on either side of August 26, 2026.
Should I cancel my rank tracking subscription?
Not yet, but stop treating it as a source of truth. Rank data still shows directional movement and competitive share of voice. Demote it to a leading indicator, promote Search Console clicks and conversions to the reported metric, and cut your tracked keyword list so a per-link price increase lands on a smaller base.
Does this affect Google Ads reporting?
Ad units already routed through a google.com/aclk address, and Clark documented related presentation changes on paid listings in May 2026. Google Ads interface reporting is unaffected, since it comes from Google’s own logs. What degrades is third-party competitive ad monitoring, which parses results pages the same way rank trackers do.
Comments are closed