Google lost its scraping case, then came back holding Reddit's contract

What the court actually said in July#

The fox that survives is the one that reads the fence before it reads the field. Most of the commentary on this case has read the field. The fence is a single subsection of American copyright law, and it decides whether the rank tracking industry keeps its footing.

On 20 July 2026, Judge Yvonne Gonzalez Rogers dismissed Google's claims against SerpApi in the Northern District of California, case 4:25-cv-10826-YGR. Google had sued under the Digital Millennium Copyright Act, alleging that SerpApi circumvented an access control Google calls SearchGuard, and that it trafficked in the means of doing so.

The dismissal turned on authority. In the court's own order, the reasoning is stated without hedging: "SerpApi argues that Google's claims under the DMCA fail because it does not allege that it implemented SearchGuard to protect a copyrighted work with the 'authority of the copyright owner' as required under 17 U.S.C. § 1201(a)(3)(B). The Court agrees."

The phrase the whole case hangs from#

That subsection is short and it is decisive. Section 1201(a)(3)(B) says a measure effectively controls access to a work if it "requires the application of information, or a process or a treatment, with the authority of the copyright owner, to gain access to the work". Google runs SearchGuard. Google does not own most of what its results contain.

The court required a plaintiff, in its words, "to allege and later prove that the technological measure in question was implemented and functioned with the authority of the copyright owner". Google had not. So the claim fell, and it fell in two different ways depending on what the guarded content was.

Where SearchGuard protects results containing no copyrighted content at all, the order "GRANTS, WITHOUT LEAVE TO AMEND". Where it protects licensed material, the order "GRANTS, WITH LEAVE TO AMEND". That second door is the one Google walked back through on 10 August.

Google's second attempt runs through Reddit#

The amended complaint is a different animal from the first. It does not argue that Google owns the results. It argues that the people who do own them asked Google to build the fence.

In the amended complaint filed on 10 August, the pleading states that "Copyright holders have authorized Google to implement access controls like SearchGuard for the content they license to Google, and in some cases insisted that Google do" so. On one unnamed partner, whose agreement it says has "versions of which date back to 2017", the filing goes further: Google is "not only authorized, it is obligated to use commercially reasonable efforts to safeguard the licensed content against unauthorized third-party access".

Then it names Reddit. The filing says the agreement "directs Google not to enable third parties to extract and independently commercialize the licensed content, making clear Google has the authorization to prevent such behaviour", and adds the detail that turns a contract into a narrative: "when Reddit suspected that scrapers like SerpApi were accessing, taking, and reselling the content that Reddit had licensed to Google, it specifically asked Google to employ technical measures to prevent such unauthorized appropriation".

Why this is cleverer than it looks#

The elegance is that it converts a weakness into a weapon. Google's problem in July was that it does not own the web. Its answer in August is that it has contracts with people who do, and those contracts contain obligations pointing in exactly the direction the statute requires. It is a quiet piece of legal footwork, and it outfoxes the obvious objection rather than arguing with it.

That is not a certainty. A pleading is an assertion, not a finding, and SerpApi has not yet responded on the public docket. But it is a materially stronger position than the one that failed, and anyone in the rank tracking business who reads it as a formality is reading the field again instead of the fence.

It is worth noting who is not in this case: the copyright owners themselves. Section 1203 allows that "[a]ny person injured by a violation of section 1201 or 1202 may bring a civil action", and provides statutory damages of "$200 or more than $2,500 per act of circumvention". Google is suing on its own behalf, holding contracts signed by others. That structure is the innovation here, and it is portable to any platform with licensing deals.

What this means for every rank tracking tool you pay for#

Every SEO programme in the world rests on a number somebody scraped. Position data does not come from Google in bulk. It comes from tools querying results at volume and parsing what comes back, which is precisely the activity this case is about. The whole industry has been feeding at the same hedgerow, and nobody asked who planted it.

So the exposure is not limited to SerpApi. Any rank tracking software, any google rank checker, any website ranking checker, and any agency dashboard drawing from them sits downstream of the same technique. If the licensing theory succeeds, the question stops being technical and becomes contractual: whose content was in the result, and what did they sign?

Read that chain the way a fox reads a trail. The agency at the end never touched a copyright, never signed a contract and never circumvented anything. It simply bought a number. That is the uncomfortable part: the risk is upstream, invisible, and priced into a monthly subscription nobody reads the terms of.

What would actually change, and what would not#

Be precise about the downside, because overstating it is its own failure. Nothing about this case makes measuring your own performance unlawful. Search Console reports your own impressions and clicks, first party, no scraping involved. Analytics reports your own sessions. Neither is touched by any of this.

What would change is the competitive layer: share of voice, competitor positions, SERP feature tracking, the whole apparatus that makes keyword ranking reports interesting rather than merely factual. That is the part built on querying results at scale, and that is the part with contractual weather over it.

That divergence is worth sitting with. The tools category is flat to falling while the underlying questions climb. People still want to know where they stand. They are simply less attached to any particular way of finding out, which is fortunate timing given what is in front of Judge Gonzalez Rogers.

What to change in your reporting this quarter#

None of this needs panic and none of it needs a lawyer yet. It needs the same thing every platform risk needs: knowing which of your numbers would survive if the supply stopped, and fixing the ones that would not. Catch the scent early and the work is small; catch it late and it is a rebuild.

That last step is the one people skip and the one that pays. A rank tracking series you exported is yours whatever happens to the vendor. A series living only in someone else's platform is a subscription to your own past.

The honest thing to tell a client#

Say that a court case is live, that it targets the technique the whole category uses, that the first attempt failed and the second is stronger, and that nothing about their own first-party measurement is affected. Then show them which of their numbers sit in which column. That conversation takes four minutes and buys a great deal of trust.

The measurement that survives either verdict#

There is a longer game here, and it has been coming regardless of what the Northern District of California decides. Position is a decreasingly good description of visibility, because the surface it describes keeps shrinking. The quarry moved while everyone kept watching the old clearing.

Google's own analytics documentation now concedes the shape of it. The default channel group definitions class Organic Search as arrivals "via non-ad links in organic-search results, including Google's AI Overviews and AI Mode". An AI Overview citation and a tenth blue link now land in the same bucket, and no position number distinguishes them.

Meanwhile the commercial traffic is genuinely moving. Shopify's Q2 2026 commerce data records AI-referred sessions up 197% year on year, with organic search sessions still growing 12% on a much larger base. Both channels growing, one of them invisible to any google rank checker ever built.

So the sensible response to the case is also the sensible response to the decade: hold position data more loosely, and build the reporting spine out of things nobody can enjoin. That means first-party performance, cited-source share where you can measure it, and enquiries at the end.

There is even early evidence about what actually earns retrieval in the new surfaces, and it is less flattering to quick fixes than the vendor pitch suggests. A controlled study, Structured Linked Data as a Memory Layer for Agent-Orchestrated Retrieval, tested plain pages against pages with schema markup against purpose-built entity pages, and found that markup alone "provides only modest improvements" while the richer entity format reached "+29.6% accuracy improvement for standard RAG". Worth knowing that its authors work for a structured data vendor, which is a reason to read the method rather than the headline.

None of which requires abandoning rank tracking tomorrow. It requires knowing what it is: a bought signal, legally contested, describing a shrinking share of how people arrive. Keep it, price it honestly, and stop letting it carry the whole story. If you want the reporting layer rebuilt around what you own, that is what our search visibility work is for, and it is the same discipline folkfox applies to content programmes that need to prove their worth without a position average to hide behind.

For the neutral framing of what section 1201 was built to do, the United States Copyright Office section 1201 study remains the clearest official account: it prohibits circumventing technological measures used to protect access to works, and it governs trafficking in the tools for doing so. The whole dispute is an argument about who the measure belongs to.

Trade coverage of the amended filing has been quick and broadly accurate, with Search Engine Roundtable and Search Engine Journal both flagging the licensing pivot on 12 August. The filings themselves are the better read, and they are public.

Frequently asked questions#