Google now limits its /goto redirects: what changed on 30 October 2026 and what to do
Since 30 October 2026 Google limits how many /goto result links you can follow per page. What changed, who is affected and how to get real URLs.
CrawlerData teamPublished 4 min read
Contents
If you read Google results programmatically, you have probably relied on one thing without thinking about it: every outbound link on a results page can be turned into the real destination URL. On 30 October 2026 that got harder. Google now limits how many of those links you can follow per results page, right after the page is served.
This is what changed, what it means for anyone collecting SERP data, and how CrawlerData handles it.
What changed at Google
Every outbound link on a Google results page is an encrypted /goto link. You cannot decode it. You follow the redirect and Google answers with the real URL.
Since 30 October, Google allows only a few of those redirects per results page at first. Open the rest straight away and Google returns a “Verifying your request” page instead of the destination.
Three details matter:
- The limit is per results page, not per IP address. Rotating IP addresses does not help, because the limit follows the page you were served.
- It eases after a few minutes. A few minutes after the page was served, every link on it resolves normally again.
- It is not only us. At least one other SERP API provider posted an incident on 30 October.
What it means if you collect SERP data
The pages themselves still load. What breaks is the step after: turning each result’s link into the actual URL. If your code follows every /goto link at once, most of them will come back as the verification page, and a naive pipeline stores that as the result URL.
That is a data-quality problem more than an outage. Rank trackers, citation trackers and anything that matches results to domains need the real URL, and a wrong one fails quietly.
You have two honest options:
- Wait, then resolve. Because the limit is per page and eases after a few minutes, you can hold the page and resolve its links later, in one go.
- Resolve what you can now, and flag the rest. If you need results in seconds, you can resolve a limited number straight away, and must mark everything else as unresolved rather than guess.
Both work. They trade speed against completeness, and which one you want depends on the job.
How CrawlerData handles it
CrawlerData supports both, and you choose per request.
When you need the real URLs: resolve_links=true
On async jobs and batches (raw HTML or JSON), set resolve_links=true. Live requests and /search.json do not use the 3-minute wait. We capture the page, park it for 3 minutes, then resolve all of its links at once and send one webhook with the result. You get the real destinations in the HTML and in the JSON, about 3.5 minutes after you submit instead of seconds, and any link we could not resolve is flagged with link_unresolved. In our tests on 30 October, every link on every test page resolved.
Use it with webhooks or batches. With batches, each item sends its own webhook once its links are resolved, and the batch-complete webhook follows. Items waiting for their links do not hold up the rest of the batch.
When you need results in seconds: the default
Without the flag, nothing changes. Live requests and /search.json stay fast and resolve the organic result links they can within about 10 seconds.
Any link we could not resolve is flagged with link_unresolved: true. The link field then holds the site’s home page so your code does not break, but the flag tells you it is not the exact result URL. Check the flag before you trust a link.
Price
Unchanged. Link resolving is included in the normal 1 credit per page. See pricing.
Which one should you use
| You need | Use | Expect |
|---|---|---|
| The exact destination URLs of a page’s results | resolve_links=true with a webhook or a batch |
About 3.5 minutes per result |
| Results in seconds, such as an interactive tool or an agent | The default | Organic links resolved within about 10 seconds; check link_unresolved |
A rank tracker that runs on a schedule can almost always wait 3.5 minutes. An agent answering a question for a user usually cannot, and should treat unresolved links as unresolved.
Where to read more
The resolve_links and link_unresolved parameters are documented in the API docs. The Google Search API page covers the endpoint, and how to track rankings through an API shows where exact URLs matter most.
If you have questions about this change, write to us. You reach the engineers who built it.
Frequently asked questions
- What changed in Google's /goto links on 30 October 2026?
- Google began limiting how many /goto redirects can be followed per results page right after the page is served. Follow more than a few straight away and Google answers with a "Verifying your request" page instead of the real destination URL.
- Is the limit per IP address?
- No. It applies per results page. A few minutes after a page was served, the links on it resolve normally again.
- How do I get the real destination URL of every result with CrawlerData?
- Set resolve_links=true on an async job or a batch (raw HTML or JSON) and use a webhook. CrawlerData captures the page, waits 3 minutes, resolves all of the page's links, flags any it could not resolve with link_unresolved, and sends one webhook with the result, about 3.5 minutes after you submit. In our tests on 30 October every link on every test page resolved.
- What happens if I do not set resolve_links?
- Nothing changes. Live requests and /search.json stay fast and resolve the organic result links they can within about 10 seconds. Any link that could not be resolved is flagged with link_unresolved: true.
- What does link_unresolved mean?
- The link could not be resolved to the exact result URL. The link field then holds the site's home page so your code does not break, and link_unresolved: true tells you not to treat it as the exact result URL.
- Does link resolving cost extra?
- No. Link resolving is included in the normal price of 1 credit per page.
- Google SERP
- SERP API
- Link resolving
- Webhooks
- Scraping