Google Is Suing the Supply Chain Behind Your Rank Tracker

Call it the licensed sliver.

Google’s refiled case against SerpApi runs fifteen pages, and it no longer claims the whole search results page. It claims the licensed paragraphs sitting inside it. Google filed the amended complaint on August 10 in the Northern District of California. That’s exactly three weeks after Chief Judge Yvonne Gonzalez Rogers threw out both of its original claims on July 20.

What makes that a marketing problem rather than a legal curiosity is what sits underneath it. There’s no official Google product that tells you where you rank. Search Console reports your own average position and says nothing about the live results page. Google’s Custom Search API queries a set of sites you configure rather than the web, its results don’t match google.com, and it’s a site-search tool now closed to new customers and headed for retirement on January 1, 2027. It was never the tap.

So the rank trackers, position monitors and AI visibility dashboards marketing teams buy have always run on scraped data. Not as a shortcut. As the only option going.

Counsel for SerpApi told MediaPost on August 11: “We remain confident in the strength of SerpApi’s position and look forward to responding.”

Here’s the part marketing leaders keep missing. Your rank tracker doesn’t get its numbers from Google. Neither does your position monitor, and neither does the AI visibility dashboard you bought this year. They buy that data from a handful of wholesalers who collect it at scale. One of those wholesalers is currently a defendant.

Why Google rewrote the claim

The July order wasn’t close. Judge Gonzalez Rogers found that Google “does not allege any facts that raise the inference that SearchGuard was implemented and functions to control access to copyrighted components…with the authority of the copyright owner.”

SearchGuard is Google’s anti-scraping system, built with what Google calls “tens of thousands of person hours and millions of dollars.” None of that mattered. Under the anti-circumvention law Google sued on, you can only defend a lock if the copyright owner asked you to install it.

Google didn’t have that permission on the record. So it went and got the paperwork.

The amended complaint pleads four sources of authority: Reddit by name, two content partners Google declines to identify, and Google’s own users through its privacy policy. The Reddit agreement, reported at roughly $60 million a year and signed in February 2024, tells Google “not to enable third parties to extract and independently commercialize the licensed content.” A second partner requires that its material “will not be available for download by third parties”.

That’s a much smaller claim, and a harder one to laugh off. The fight stops being about whether Google owns its results page and starts being about a contract someone actually signed.

The operator read

Jason Grad, co-founder and CEO of Massive, posted his objection on August 13. Massive sells consented web-access infrastructure, so he has money riding on this. His argument is worth reading anyway:

“Open to humans to sell ads, closed to bots so nobody builds a competing index. They want it both ways.”

He expects the narrow version to fail too, and thinks the real mechanism is cost rather than law:

“winning for them is about burying a smaller company in legal costs, not about having the law side with them”

SerpApi’s founder read the July dismissal as a matter of principle. Julien Khaleghy called it a win “not just for SerpApi, but for all who depend on an open internet”.

What it does to your stack

Search data is a supply chain, and marketing teams have been buying from it without pricing the risk. Three things move if Google’s licensed-content theory survives.

Wholesale gets more expensive. A supplier who has to strip licensed material out of every result, or carry the exposure of leaving it in, charges more for the same feed. That lands on the per-keyword price your agency or your tool passes through.

Coverage thins exactly where you’re watching. Reddit threads are the results brands now monitor most closely, because assistants lean on them so heavily, and Reddit is already selling those conversations twice. They’re also the results most likely to get stripped.

The fallback is weaker than people assume. This is the same plumbing feeding AI answers. Our reporting on ChatGPT’s retrieval stack found its paid tier draws roughly three quarters of search results from a supplier of scraped Google rankings. The tools that tell you where you show up in AI answers drink from this well too.

The case against worrying

The honest counter is strong. Google has already lost this argument once on the merits, and the claims covering results with no copyrighted material stay dismissed. What’s left is narrow by construction. Grad thinks it fails again, and he may be right.

Google has a second-order problem too. To win, it has to argue that its results page is full of copyrighted work needing protection, which cuts against two decades of its own position on everyone else’s content. And by pleading confidential contracts as the factual core, Google turned them into discoverable material. Its licensing terms are now headed toward the public record.

The outcome still isn’t the thing to plan around. Reddit is suing SerpApi separately in New York, where subpoenaed figures put its access at 784,006,306 result pages carrying Reddit data in one week of July 2025. Two suits, one supplier. Legal costs reach vendor contracts long before a judge rules.

What to do before renewal

Ask your SEO or AI visibility vendor one question: where does your search data come from, and what happens if that supplier is enjoined? A vendor who can name their sources and their backup has thought about it. A vendor who calls it proprietary has not, or would rather not say.

And drop any idea that there’s an official fallback waiting. There isn’t one and there never was. A vendor who implies they could switch to some Google API if scraping gets harder is describing a product that doesn’t exist.

This is part of what AI is doing to the martech stack, and it lands on the line item marketers treat as the safest thing they buy.

---

Correction, 2026-08-19. This piece originally ran under the headline “Google’s Official Search API Dies January 1. The Alternative Is in Court.” and framed the retirement of Google’s Custom Search JSON API as an official route to rank data closing. That was wrong. That API queries a Programmable Search Engine, meaning a set of sites you configure, and Google’s own documentation says its results don’t match google.com. It’s a site-search product being migrated to Vertex AI Search, and rank trackers never used it. Steve Toth of Notebook Agency flagged the error. The headline, dek and framing have been rewritten.

The correction does not touch the reporting on Google’s amended complaint, the licensed-content theory, or the Reddit litigation. It sharpens the point the piece was making: there’s no official alternative to lose, and that’s exactly why these tools depend on scraped data.