Google's "goto" URL Redirect Update and What it Means for SEO

Google confirmed on August 26, 2026 that it is now routing search result links through google.com/goto?url= redirects instead of linking directly to the destination site.

It has been testing this for months, and by late August it was at roughly full rollout.

If you only read one line of this entire article: this is not a ranking algorithm update. It is a data supply update. And the data being cut off is the data almost every SEO report you use is built on.

Takeaways & TL;DR

• Google replaced direct links in the SERP with encoded google.com/goto redirects. The destination URL is no longer readable in the page markup. A crawler has to follow a 302 hop for every single link to find out where it goes.
• Nozzle CEO Derek Perkins reports that resolving all the links for one five-page ranking now takes between 500 and 1,000 requests. Google also blocks HEAD requests, which kills the obvious cheap workaround.
• Expect third-party rank tracking data to get less complete, less frequent, more expensive, or all three. Some vendors are already discarding affected result pages rather than reporting bad data.
• This is the second hit in a year. Google killed the &num=100 parameter in September 2025 and Semrush confirmed a tenfold operational cost increase within a day.
• Google Search Console and GA4 become your most reliable Google data. They are also the only Google data you own.
• Competitive rank data is the real casualty. GSC covers your site. It does not cover your competitors.
• Bing matters more than it did last month, because most AI answer engines never depended on Google's index in the first place.
• Do not switch tools reactively right now. You will break your baseline twice.

What Google actually changed

Previously, an organic result in the SERP contained a plain anchor pointing at the destination. Anyone parsing the page could read the ranking URL straight out of the HTML.

Now the link points at google.com/goto?url=[encoded value]. The encoded value cannot be decoded locally. It is a 302 to the real destination, so the only way to learn the ranking URL is to make the request and see where you land. Do that for every organic and paid link on every page of every SERP you track and the math gets ugly fast.

Google's statement on it was short: "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 that as what it is. Google is protecting the SERP as an asset.

Why this costs 10x to 100x more, not 2x

The cost is not the redirect itself. It is the request volume plus the rate limiting.

Perkins put the number at 500 to 1,000 requests to fully resolve a single five-page ranking. Before this change, that was one request.

He also confirmed Google does not let HEAD requests through, so you cannot cheaply peek at the redirect header without pulling a response. You can stop at the redirect and skip the full page fetch, which helps, but the binding constraint is not bandwidth. It is how many requests Google will tolerate from you before it starts blocking.

That is the actual mechanism here. More hops means more time on Google's infrastructure, which means more signal for Google to fingerprint and block the crawler. The cost increase is the first-order effect. The blocking is the second-order effect, and it is the one that will change the industry.

Two things follow from that:

1. Large vendors will absorb the cost, and pass some of it along in pricing, refresh frequency, or tracking depth.
2. Small vendors and in-house scrapers get priced out entirely, or blocked into uselessness.

Nozzle discarded goto-only results pages during testing rather than pollute their dataset with unresolvable links. Brodie Clark of SERP Lens reported making a similar call. That is the responsible move, and it means gaps in your history rather than wrong numbers in your history. Not every vendor will make the same choice.

This is the second hit, not the first

Context matters here, because the trendline is as important as this event.

In September 2025 Google removed the &num=100 parameter, which had let a single request pull 100 results instead of 10. The fallout was immediate and measurable:

• Semrush confirmed a tenfold operational cost increase within one day.
• AccuRanker cut tracking to the top twenty results.
• SISTRIX discontinued desktop updates.
• Tyler Gargula at LOCOMOTIVE studied 319 properties and found 87.7% saw Search Console impression declines and 77.6% lost unique ranking terms, because a meaningful share of the old impression counts were bots, not people.

So we have now had two structural cuts to third-party SERP access in twelve months. Both were framed as anti-abuse measures. Both had the same practical effect: the only complete, reliable, current picture of Google's results is the one Google holds.

Plan for more.

Google is fighting this on two fronts

The technical change is happening alongside a legal one.

Google sued SerpApi in December 2025 over scraping. In July 2026 a federal judge dismissed the case, finding Google lacked standing because it does not own public search results and search results themselves are not protected by copyright (ironic).

Google confirmed it would file an amended complaint, shifting the argument toward Knowledge Panels, where licensed media and rights holder protections give it firmer ground. That amended fight was back in the news in August 2026.

Here is why that matters to you: the legal outcome is uncertain, and the technical change works either way. Google does not need to win in court to make scraping uneconomical.

The goto redirect is the strategy that does not depend on a judge.

What this actually does to your reporting

Your rank tracking history is about to have a seam in it. Whatever your tool does next, the data before this change and the data after it were collected under different conditions. Treat the changeover date as a hard annotation on every ranking chart. If your position average moves in September, the first hypothesis is methodology, not performance. That was the lesson from num=100 and most teams learned it too late.

Coverage will thin before accuracy degrades. The likely vendor responses are fewer tracked results per SERP, less frequent refreshes, higher prices, or quietly narrowed keyword sets. Watch for silent changes more than loud ones.

Vendors will not respond uniformly. Ahrefs, Semrush, SE Ranking, AccuRanker, and everyone else has a different cost structure, a different crawl architecture, and a different tolerance for gaps. Cross-tool comparisons were already unreliable. Assume they are now meaningless.

GSC and GA4 become your source of truth for Google. They are first party, they are not scraped, and they are not affected by any of this. GSC is not perfect. It samples, it caps at 16 months, it averages positions in ways that hide more than they show, and it tells you nothing about competitors. It is still the most defensible Google number you have.

Competitive visibility is the real loss. This is the part most people are underweighting. You can replace your own rank data with GSC. You cannot replace competitor rank data with anything except third-party scraping, and third-party scraping is exactly what just got expensive. Share of voice, competitor keyword gaps, and SERP feature ownership are the metrics that will quietly get worse or more expensive first.

Paid gets hit too. The redirects apply to paid links as well as organic, so competitive ad monitoring and ad copy intelligence tools are in the same blast radius.

The AI search angle

The instinct is to say this cuts off AI crawlers from Google. Partly true, but not completely.

Most major answer engines were never reading Google's SERPs directly at scale. ChatGPT Search runs on Bing's index blended with OpenAI's own crawler. The systems that were hitting Google live were agents and tools built on SERP API providers, and those are exactly the ones this change targets.

The practical implications:

• Rank tracking is not AI visibility tracking, and this change widens the gap between them. If your AEO reporting is inferred from Google rankings, it was already weak. Now it is weaker.
• Bing indexation and Bing Webmaster Tools deserve more attention than most SaaS teams give them. Not because Bing traffic matters, but because Bing is upstream of a meaningful share of AI answers.
• Google ranking still matters enormously for Google AI Overviews and AI Mode, because Google grounds those in its own index. What is changing is the relationship between your Google rank and your visibility in non-Google AI surfaces. Those were always more separable than the industry admitted.
• Dedicated AI visibility tracking, measuring citations and mentions in actual answer engine responses, moves from nice-to-have to core reporting. Paired with GSC and GA4, that is the reporting stack that survives this.

One thing to watch that nobody has confirmed yet

Nobody has published a header-level analysis of what the goto hop does to the referrer.

If the redirect preserves a clean google.com referrer, GA4 classification is fine and this whole section is moot. If it strips or rewrites it, some share of organic sessions could start landing as (direct) / (none). Redirects stripping parameters is a documented behavior in Google's own analytics guidance.

I want to be clear that this is a risk to monitor, not a thing that has happened. The check is simple and worth doing now: pull your google / organic versus (direct) / (none) session split in GA4, week over week, from mid-August forward. If direct spikes without a corresponding organic drop in GSC clicks, you have your answer.

What to do in the next 30 days

1. Export your rank tracking history now. All of it, from every tool, raw. If your vendor re-baselines, discards affected pages, or backfills differently, you want the pre-change record in your own storage. This is the single highest-value action on this list and it takes an hour.
2. Annotate the date in every dashboard and report. Late August 2026. Put it on the chart so nobody has to relitigate it in a QBR three months from now.
3. Rebuild your reporting hierarchy around GSC and GA4. Clicks, impressions, and conversions from first-party sources at the top. Third-party rank as a directional supporting metric, not the headline.
4. Cut your tracked keyword set down to what you actually act on. Tracking 5,000 keywords was already mostly theater. In a world where every tracked SERP costs 500 to 1,000 requests, a tight strategic set refreshed reliably beats a bloated set refreshed badly.
5. Get on Bing Webmaster Tools and verify indexation. Low effort, and it is upstream of AI visibility you cannot otherwise influence.
6. Add or formalize AI visibility tracking. Actual citation and mention measurement in answer engines, not rank data used as a proxy.
7. Evaluate pull-based API access like DataForSEO for the specific things you need. Paying per query for the data you actually use beats paying a platform subscription for a dashboard you skim. One caveat worth stating plainly. The API providers are themselves crawling Google, so their unit costs are rising too. Expect repricing rather than a permanent arbitrage.

What not to do

Do not build your own scraper. The economics were marginal before this and they are bad now. You would be entering a rate limiting arms race against the party that sets the rate limits.

Do not switch rank tracking vendors right now. You will break your baseline once from the goto change and a second time from the tool migration, and you will never untangle which caused what. If you are unhappy with your vendor, wait for the dust to settle and switch on a clean quarter boundary.

Do not overcorrect away from organic rankings. Ranking in Google still drives Google traffic, still feeds AI Overviews and AI Mode, and still compounds. What changed is your ability to measure it through third parties, not its value.

Where this goes

The direction of travel has been obvious for a year and this makes it undeniable. Google is closing the SERP to everyone except Google. Every measurement layer that sat between you and Google's results is getting thinner, slower, and more expensive, and it is not coming back.

The reporting stack that survives is first party at the core. Search Console and GA4 for what happened on your site. Dedicated AI visibility measurement for what is happening in answer engines. Rank tracking kept deliberately high level, treated as directional context rather than the scoreboard. And business outcomes, pipeline and revenue, as the metric that was always supposed to be at the top anyway.

Teams that already reported that way will barely feel this. Teams whose entire SEO narrative rests on a rank tracking screenshot are about to have a hard quarter of conversations.

Fix the reporting before you have to explain it.

We publish a monthly breakdown of changes like this and what they mean for SaaS search performance in The AI Search Report. If you want us to look at how your current reporting stack holds up under this shift, book an Opportunity Assessment.

‍