By Cody Wise, Founder, Wise Media. Last updated 2 September 2026.
Summary
On 26 August 2026 Google confirmed it is rolling out google.com/goto passthrough links in organic search results. Result links now point at a Google address that server-side redirects to your page. Searchers are unaffected. Rank tracking tools are, because the destination cannot be read from the link itself. If your agency report looks different this month, check the measurement before you change the site.
Table of contents
- What is google.com/goto and what actually changed?
- Why does my rank report look different? A five-line diagnostic
- This is not one change. It is a programme.
- The num=100 precedent, and why the two changes compound
- The correctness point: was that volatility even real?
- What it changes for ordinary users, including a safety habit
- What to actually do, in order
- Common mistakes
- FAQ
What is google.com/goto and what actually changed?
google.com/goto is a Google-owned passthrough address. Instead of a search result linking straight to your page, the link points at google.com/goto with an encoded value attached, and Google issues a server-side redirect to the real destination. The person searching lands on the same page they always would. Nothing about your ranking, your indexing or your content has changed.
The confirmed timeline
Search Engine Roundtable first documented the behaviour as a test in July 2026. On 26 August 2026 a Google spokesperson confirmed the rollout to Barry Schwartz, saying: “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.” Read the confirmation at Search Engine Roundtable.
Note what Google did and did not say. It named abuse. It did not name scrapers, rank trackers or AI companies. Anyone telling you Google confirmed it is targeting SEO tools is filling in a blank Google left empty. Schwartz connects it to Google’s litigation against SerpApi, and that is his assessment rather than a Google statement.
How complete is the rollout?
Derek Perkins of Nozzle told Schwartz he was seeing “nearly a 100% rollout across several residential IP providers” after roughly four months of testing. Behaviour still varies by search environment, including whether the searcher is signed in. Treat it as near-universal rather than absolutely universal.
Why it is hard for tools, in one paragraph
Perkins described the mechanics plainly. The goto links cannot be decoded locally, so a tool has to follow each redirect to learn where it goes. Google refuses HEAD requests, so a provider cannot cheaply ask “where does this point” without pulling the full response. A single page of results carries a large number of links, and resolving a five-page ranking check can take hundreds of requests. The binding constraint is not bandwidth, it is rate limiting.
Practically: a job that used to cost one request now costs many. That cost lands on your rank tracking vendor, and vendors respond to cost by sampling differently, tracking fewer positions, or raising prices.
Why does my rank report look different? A five-line diagnostic
Answer first: if the change appeared in your rank tracking tool but not in Google Search Console, it is a measurement change, not a ranking change. Search Console reports from Google’s own logs and is not affected by the goto redirect at all. That single comparison resolves most cases in under five minutes.
| What you are seeing | Most likely cause | First action |
|---|---|---|
| Rank tool shows movement, Search Console positions are flat | Tool sampling change, not a ranking change | Ask your vendor what changed in their collection method since June 2026 |
| Rank tool suddenly tracks fewer keywords or shallower positions | Vendor cut coverage to control request costs | Confirm the tracked-position depth in writing before comparing months |
| Search Console positions changed, clicks changed too | A real ranking change | Diagnose properly, starting with the traffic-drop decision tree |
| Impressions and query counts fell around mid-September 2025 | The num=100 removal, not this change | Rebase your reporting to a post-September 2025 baseline |
| Referrer data in analytics looks unfamiliar | Verify before assuming; do not rebuild attribution on a guess | Compare against Search Console clicks for the same period |
If line three applies to you, work through our Google traffic drop diagnostic rather than guessing. If pages are in the index but never surfacing, the separate problem is covered in indexed but not ranking.
This is not one change. It is a programme.
Almost every page covering goto treats it as a standalone event. It is not. Google has shipped or tested at least three separate anti-automation measures in Search inside four weeks, and reading them together explains far more than reading any one of them alone.
| Date | Measure | Status |
|---|---|---|
| Early August 2026 | Prompting searchers to sign in to verify they are human before continuing | Test |
| 26 August 2026 | google.com/goto passthrough links on results | Confirmed rollout |
| 1 September 2026 | An interstitial that verifies the request, then enables a Continue button after a few seconds, with no sign-in required | Test, spotted by Glenn Gabe and reported in several languages |
The 1 September interstitial is documented at Search Engine Roundtable. Alexis Rylko reported the same screen on a third page of results in private browsing, which is a useful detail: deeper result pages and privacy-mode sessions look more automated to Google, so ordinary humans doing ordinary research are the ones most likely to see it.
Why this matters to a business owner and not just to an SEO: if Google is layering friction across several points in the results page, expect third-party search data to get thinner, later and more expensive for a while. Plan your reporting around Search Console and your own analytics, which are unaffected, rather than around a rank tracker you do not control.
The num=100 precedent, and why the two changes compound
This has happened before, and the last time it happened it broke a year of reporting comparisons for a lot of businesses. In mid-September 2025 Google disabled the num=100 parameter, which had allowed a single request to return 100 results instead of 10.
What the data showed
An analysis of 319 properties by Tyler Gargula of LOCOMOTIVE Agency, reported by Search Engine Land on 18 September 2025, found:
- 87.7% of sites lost impressions in Google Search Console.
- 77.6% of sites lost unique ranking terms.
- Short-tail and mid-tail keywords took the biggest hit.
- Average position rose sharply, because far fewer page-three-and-beyond impressions were being recorded.
The uncomfortable reading, which Brodie Clark raised at the time, is that scrapers using num=100 may have been inflating Search Console impressions for years. Clicks barely moved, because bots never clicked. So click-through rate appeared to jump. None of that was real performance in either direction. Search Engine Land’s contemporaneous coverage of the wider mess is here.
The compounding point nobody states
These two changes do not substitute for each other. They stack, and they attack different parts of the same cost equation.
| num=100 removal (Sept 2025) | goto passthrough (Aug 2026) | |
|---|---|---|
| What it multiplies | Pages fetched to see a full ranking | Requests needed to interpret one page |
| Effect on Search Console | Large. Impressions and query counts fell | None. Search Console is Google’s own log |
| Effect on third-party rank data | Coverage depth | Cost and latency per check |
| Effect on your rankings | None | None |
A vendor that absorbed the first change by fetching more pages now has to resolve hundreds of redirects on each of those pages. That is a multiplication, not an addition.
The correctness point: was that volatility even real?
This is the part worth the whole article, and it is the part your competitors’ pages are not saying.
Google announced a spam update on 18 August 2026 and confirmed it complete on 21 August 2026. That is the only confirmed Google ranking event in August 2026. There was no August 2026 core update, whatever a summary tells you. Plenty of tools and commentators reported heavy volatility across that window.
Here is the problem. During the same period, providers were dealing with goto rollout at increasing coverage. A provider that discards a results page it cannot fully resolve is sampling differently than it was in June. So some reported volatility in August 2026 may be reduced tool visibility rather than real ranking movement.
The rule to take away: datasets built before June 2026, during the test window, and after the confirmed rollout may not be measuring the same thing. Comparing them like for like produces conclusions that feel data-driven and are not. This is not a claim that August volatility was fake. It is a claim that a rank tracker alone cannot tell you which it was, and that you should not spend money on a fix until you know.
What it changes for ordinary users, including a safety habit
For a searcher, the visible change is small but not nothing. Hovering a result used to show the destination domain in the browser status bar. Now it shows a Google address. The habit of glancing at the status bar to check where a link goes, which is one of the few phishing sanity checks ordinary people actually use, no longer works on a Google results page.
Two practical consequences:
- Train your team on the new check. Verify the domain in the address bar after the page loads, not in the status bar before clicking.
- Do not treat a google.com prefix as a trust signal. The redirect says nothing about the destination beyond the fact that Google indexed it.
This matters more than it sounds if your business is in a sector where competitors or bad actors buy lookalike domains. The same reasoning applies to keeping your own site patched, which we covered in the August 2026 WordPress vulnerability runbook.
What to actually do, in order
Six steps, roughly thirty minutes, no budget required.
- Open Search Console before you open the rank tool. Compare the last 28 days against the previous 28 for clicks, impressions and average position. Search Console is unaffected by goto, so it is your control group.
- Ask your rank vendor, in writing, what changed in their collection method since June 2026. Specifically: tracked position depth, update frequency, and whether unresolvable results pages are dropped or estimated. Keep the answer with your reports.
- Mark the dates on your reporting. Add a note to your dashboard at 12 September 2025 and 26 August 2026. Any month-over-month comparison spanning those dates needs a caveat.
- Rebase your baseline. If you are still comparing against pre-September-2025 impression counts, stop. Those numbers included scraper traffic.
- Add a second source. Bing Webmaster Tools is free and reports citation-level data Google does not. We covered the split in Google’s AI search report will not tell you who cited you.
- Report on conversions, not positions. Positions are a leading indicator that Google is now actively making harder to observe. Enquiries, calls and booked jobs are not, and they are what the business actually runs on.
Checklist for reviewing your next agency report
- Does it show Search Console data, or only third-party rank data?
- Does it state the tracked position depth?
- Does it note any methodology change since June 2026?
- Does it separate impressions, clicks and conversions rather than blending them?
- Does it explain a movement, or just display one?
If a report cannot answer four of those five, the problem is the reporting, not the rankings. What a proper engagement should cost in Canadian dollars is broken down in our SEO pricing guide for Canada.
Common mistakes
| Mistake | Why it costs you |
|---|---|
| Rebuilding or de-optimising pages because a rank tool moved | You change a site that was never the problem, and lose the ability to attribute anything afterwards |
| Blocking or redirecting google.com/goto traffic at your firewall | Those are real searchers arriving via a redirect. Blocking them blocks customers |
| Reading a rank drop as a traffic drop | Four different signals, one conclusion. Check clicks before you check positions |
| Switching agencies over a single month of tool data | The tool changed. Ask what changed before you decide who is at fault |
| Assuming Google confirmed it is targeting SEO tools | Google said abuse and named nobody. Building strategy on an inference is how bad decisions get made |
Frequently asked questions
Does google.com/goto hurt my SEO?
No. It changes how the link is presented in the results page, not how the page is ranked, crawled or indexed. There is no action to take on your website and nothing to fix in your code.
Will I lose referrer data in Google Analytics?
Verify this in your own property rather than trusting a blog post, including this one. Google Search referrals have been reported as organic search for years and the redirect is server-side, so most setups should be unaffected. Compare a fortnight of analytics sessions against Search Console clicks for the same fortnight before you change any attribution model.
Why did my rank tracker suddenly show fewer keywords?
Two candidates. The num=100 removal in September 2025 permanently reduced how deep most tools track, which cut unique ranking terms for 77.6% of sites in one study of 319 properties. The goto rollout has raised the cost of resolving each results page, and some providers respond by narrowing coverage. Ask your vendor which applies.
Is Google Search Console affected?
No. Search Console reports from Google’s own systems and does not depend on reading the results page. That is exactly why it is the right control group when a third-party tool disagrees with it.
Can I decode a google.com/goto link myself?
Not locally. Derek Perkins of Nozzle reported that the encoded value cannot be decoded client-side, and that Google refuses HEAD requests, so the only way to learn the destination is to follow the redirect. For a business owner there is no reason to try.
Does this mean rank tracking is over?
No, but it should stop being the headline number. Rank tracking is getting slower, shallower and more expensive to produce. Search Console, Bing Webmaster Tools and your own conversion data are cheaper, more accurate and closer to revenue. Use rank data as a directional signal, not as a scorecard.
Is any of this specific to Canada?
The rollout is not region-specific. The practical consequence for Canadian businesses is the same as anywhere: if your agency reports in third-party rank positions only, ask for Search Console data alongside it, and make sure the reporting currency and the reporting method are both stated.
The bottom line
Your rankings did not change. The instrument you were reading them with did. Google has spent four weeks making its results page harder to read automatically, and it has said it will keep going. The businesses that come out of this fine are the ones that already report on enquiries and revenue, and treat position data as a diagnostic rather than a scoreboard.
If your current report cannot tell you whether last month was a ranking change or a measurement change, that is a fixable problem. Our Website Growth packages cover search reporting that separates the two, plus the technical work behind it.
Tell us what you are seeing and get a scoped quote. We will look at your Search Console before we look at anything else.