Google's goto links push rank trackers to 500 to 1,000 requests per query

Resolving the links on a single five-page ranking report now takes between 500 and 1,000 requests, according to Derek Perkins, founder and chief executive of rank tracking firm Nozzle, in a report published by PPC Land. Perkins said on 26 August 2026, the day Google confirmed that it is rolling out google.com/goto passthrough links inside search results, that his company was seeing near total rollout coverage across several residential IP providers. The confirmation was covered here on 27 August 2026. The measurement bill attached to it is the part that is new.

What is new here, and what was already reported

This story is the tooling consequence, not the confirmation. Keeping the two apart matters, because the confirmation was a policy event and this is an engineering one.

Already reported, and covered on this blog on 27 August 2026:

• Google confirmed on 26 August 2026 that google.com/goto passthrough links are rolling out in search results, replacing the destination addresses that measurement tools have read straight from the page for years.
• A Google spokesperson, quoted by Search Engine Roundtable, said: "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."

New in the PPC Land report, and all of it attributed to Perkins of Nozzle rather than to Google:

• Nozzle recorded the change at near total coverage across several residential IP providers on 26 August 2026, after roughly four months of intermittent testing. Perkins wrote: "It was widely reported the first week of August from the first major test, and as of this morning, we're seeing near 100% rollout across several residential ip providers."
• The goto links cannot be decoded locally, so a provider has to follow the redirect rather than read the address off the page.
• Google does not let a HEAD request through, which removes the cheapest way of retrieving a redirect target.
• Resolving all links for a single five-page ranking takes 500 to 1,000 requests, according to Perkins.
• Rate limiting, rather than bandwidth or storage, becomes the binding constraint on rank tracking at that volume.

Why decoding the link locally is not the workaround

The engineering problem sits in the encoding. According to a post by Alex Greenland on 23 June 2026, quoted by PPC Land, the rewritten links follow a consistent pattern: "Result link URLs are now being rewritten to google.com/goto?url=[custom base64-like encoding of URL or Google ID]. Upon click, they redirect to their proper destination." Perkins said the links cannot be decoded locally, which forces providers to follow the redirect to learn where a listing points. He added: "While that works in small scale tests, Google seems likely to make that tough, since each SERP will have hundreds of links to decode."

Two workarounds surfaced immediately in replies to his post, and both hit the same wall. Glenn Gabe asked whether tools could still pull the addresses. Mic King asked whether HEAD requests would be enough, since a HEAD request retrieves headers without the response body and would return the redirect target cheaply. Perkins answered that Google does not let the HEAD request through. A GET request that stops at the redirect works and avoids a full page fetch, but the volume is what bites.

Dominic Woodman raised the scale question from the other side, asking whether the redirect is static per page, which would let a large operator build a lookup table once and reuse it. PPC Land reported that the question was not answered in the thread. That answer is the difference between a fixed engineering cost paid once and a permanent per query tax, and nobody has it yet.

The request arithmetic, and the historical comparison

Perkins drew the comparison to the num=100 change himself, and PPC Land set out the numbers for both events. The table below holds only figures the report stated.

PPC Land framed the structural difference plainly. The num=100 change multiplied the number of results pages a provider had to fetch. The goto change multiplies the number of requests needed to interpret one results page that has already been fetched. Those are different kinds of cost, and the second one scales with the number of links on the page rather than the number of pages tracked.

The report also recalled what the industry did after the num=100 change, which is the closest available precedent. According to PPC Land, AccuRanker limited tracking to the top twenty results, SISTRIX discontinued desktop results page updates, SerpApi built a workaround through its Google Light Fast API and Google restricted that endpoint to three organic results within weeks, and Ahrefs restored top 100 tracking at the end of October 2025 without disclosing the method. Those are the recorded responses to the previous change, not forecasts about this one.

How the behaviour was spotted over four months

The sequence runs from late June to late August 2026, and the gap between the people who could see it and the people who could not is characteristic of a bucketed rollout.

The 302 status Clark recorded on 2 July 2026 matters technically, because a temporary redirect signals to crawlers that the wrapper is not meant to replace the target address in any index. Two comments on Schwartz's 8 July article suggested the effect was already reaching commercial tooling: Justin Mosebach reported that Places Scout ranking reports were beginning to show the impact, and Aleksandr Sovgir of the Serpdino rank tracker described verification interstitials returning on standard search URLs with a 200 status code and a correct page title but no results container.

What Google said, and what it did not say

Google's confirmation on 26 August 2026 was one sentence long and it named abuse. According to the spokesperson statement quoted by Search Engine Roundtable and reported by PPC Land, Google has a long history of deploying technical measures against evolving forms of abuse and regularly takes steps to protect its services and users. Perkins described the underlying mechanism in his own words as anti-bot work, saying Google has been testing anti-bot measures that eliminate all client side links in favour of server side redirects through google.com/goto.

The distance between those two statements is the whole reporting problem. Google did not identify rank trackers, artificial intelligence firms or any other category of automated client as the target. It gave no timetable, no geographic breakdown and no technical documentation. Every operational number in this story, including the 500 to 1,000 requests, the blocked HEAD request and the near total coverage reading, comes from Perkins of Nozzle as reported by PPC Land, and should be cited that way rather than as a Google figure.

A measurement problem that runs in both directions

Following a redirect is not a neutral act, and PPC Land recorded that question being asked in the thread. Joe Manna asked whether resolving a goto link registers as a click, and whether click reports would start carrying noise generated by rank trackers. Perkins responded that it would be interesting to see whether Google correlates those events with client side analytics.

Neither man answered it, and the honest position is that nobody outside Google can. What can be said is the shape of the question. If resolution registers server side, the party doing the measuring becomes a source of the traffic being measured. If Google cross references those events against client side signals, the wrapper works as a detection surface for automated clients as well as a link. Both readings are speculation until someone publishes evidence, and this article is not going to supply the evidence.

What this means for Thai marketers

Google published no geographic breakdown, so there is no basis for saying whether Thai results pages are affected more, less or exactly as much as any other market. Nozzle's reading was taken across several residential IP providers, and PPC Land did not report which countries those providers cover.

What follows is reasoning rather than reporting. Rank tracking is an input to almost every SEO retainer in this market, and it is usually the number a client looks at first. If a provider has to spend 500 to 1,000 requests to resolve the links behind one five-page report, the tooling side has three general levers: refresh less often, track fewer keywords or fewer result positions, or change how a position is derived. Any of those would show up in a Thai client's dashboard as data that is slower, thinner, sampled differently, or simply less complete, without the dashboard necessarily saying so. That is a plausible consequence of the arithmetic in the report, not something PPC Land or Google forecast.

The practical response is unglamorous. Ask your own tooling vendor what changed, rather than assuming, and treat any sudden discontinuity in a ranking chart from late August 2026 onward as something to verify before it is explained to a client as a ranking movement. A tracked position that changes because the collection method changed is not a ranking change, and mistaking one for the other is the expensive error here. Teams that pair tracking with search performance work grounded in first party data from Search Console and analytics have a second reference point when the third party number wobbles.

What an SEO team can usefully do now

None of the following is advice from the source; it is a reasoned checklist built on what the report states.

• Ask your rank tracking vendor directly whether goto passthrough links affect their collection, and what they intend to do about it. The report describes the constraint, not any particular vendor's plan.
• Note the date. If a chart shows a break from late August 2026, check the method before you interpret the movement.
• Stop treating a single tracked position as the reporting headline. Impressions, clicks and average position from Search Console come from Google itself and are unaffected by how a third party parses a results page.
• Audit any in-house script that parses result links. Scripts written to read a destination address off the page are the ones that break first, and a technical site and reporting audit is the place to find them.
• Reconsider what your competitive reporting is actually for. If share of voice numbers become more expensive to collect, the question of how often you genuinely need them becomes a budget question rather than a technical one.

Frequently asked questions

Does this mean rank trackers stop working?

No. The report describes a cost and rate limiting problem, not a shutdown: a GET request that stops at the redirect still resolves a link, and Perkins of Nozzle put the volume at 500 to 1,000 requests for a single five-page ranking.

Did Google say it is targeting rank trackers?

No. Google's statement named evolving forms of abuse and did not identify rank trackers, artificial intelligence firms or any category of automated client, and it came with no timetable, no geographic breakdown and no documentation.

Is this affecting search results in Thailand?

The report does not say. Nozzle's near total coverage reading was taken across several residential IP providers, and neither Google nor PPC Land published a geographic breakdown.

Do I have to change anything in my own account?

Nothing in Google Ads or Search Console changes because of this. The affected layer is third party collection of organic results pages, so the action is a conversation with your tooling vendor rather than a setting to adjust.

Why can tools not just decode the goto link?

Perkins said the links cannot be decoded locally, so a provider has to follow the redirect, and Google does not let a HEAD request through, which rules out the cheap header only method of reading the redirect target.

Where this leaves reporting

Two changes in twelve months have both raised the cost of watching Google from the outside, and each one landed on the same layer of the industry. The full thread, including Perkins' chart and the replies quoted above, is in the PPC Land report.

If your monthly reporting leans on third party rank data and you want a second, first party view of how your pages actually perform in search and in AI answer surfaces, our team can build that reporting with you. It is a better time than usual to have more than one number.