Combining a page’s keywords, a user’s browsing history, and a revenue-based score to choose an online ad may sound like an internet-era technical invention. On September 25, 2026, the Federal Circuit held that it is not patent-eligible. In The Receivership Estate of AudienceScience Inc. v. Google LLC, No. 2024-1825 (nonprecedential), the court affirmed a judgment on the pleadings that three internet-advertising patents asserted against Google LLC and YouTube LLC claim ineligible subject matter under 35 U.S.C. § 101. This overview is general information for Iowa software founders and product teams, not legal advice, and it does not create an attorney-client relationship.
The patents and the claim
The Receivership Estate of AudienceScience Inc. and its general receiver, Revitalization Partners, LLC, asserted U.S. Patent Nos. 7,747,676, 7,882,175, and 8,082,298. The patents share a specification that claims priority to December 2004. They describe choosing an ad for a publisher’s web page based on both page context and user history. Each page is linked to keywords, the pages a user has requested are tracked, and candidate ads matching those keywords are weighted by a “performance score” that reflects how much revenue each ad has generated before.
On appeal, the court treated claim 1 of the ’676 patent as representative. That claim covers adding the requested page to the user’s request history, mapping those pages to keywords, mapping keywords to a set of ads, weighting each ad by its revenue-based performance score, selecting an ad according to those weightings, and returning the page with that ad.
The case was filed in the Western District of Texas in 2021 and later transferred to the Northern District of California, where Judge Edward J. Davila granted Google’s Rule 12(c) motion. Judge Reyna wrote the Federal Circuit opinion for a panel that also included Judges Taranto and Stark.
Alice step one: targeted advertising, now on the internet
The court held that the claims are directed to the abstract idea of “targeting advertisements to internet users to maximize revenue generation.” It rejected three arguments:
- “It only exists online.” Relying on DDR Holdings, AudienceScience argued that the claims solve an internet-specific problem. The court disagreed. Choosing the most effective ad is a problem that also exists in print and television, where publishers target audiences based on past performance. Applying a known business practice to the internet doesn’t make it eligible.
- “The steps are granular.” The court compared the claim to the ad-matching claims held abstract in Chewy v. IBM. Gathering keywords, finding matching ads, and displaying one, without saying how any of that is done technically, is still targeted advertising. The revenue-based weighting added some particularity but not enough to change what the claim is directed to.
- “Other software claims survived.” The court distinguished SRI International, Finjan, and McRO. The claims in those cases recited specific solutions to technological problems. Here, the court said, the claims combine two conventional ad-targeting approaches with a performance score aimed at the non-technological goal of choosing the highest-earning ad.
Alice step two: no inventive concept on this record
AudienceScience pointed to two possible inventive concepts. The first was using page context and user history together. The patents’ own specification described each approach as conventional, and the claimed benefit, more personalized results, was just the abstract idea again. The second was the weighted performance score. The court noted that the claim recites the weighting without explaining how it is carried out. It also noted that the complaint contained no allegations that the weighting was inventive, which distinguished Weisner v. Google, where specific allegations and specification support helped claims survive the pleading stage.
The chance to amend was forfeited
The district court dismissed without leave to amend, finding amendment futile. On appeal, the Federal Circuit didn’t reach the futility question. It held that AudienceScience forfeited the issue: it never asked the district court for leave to amend, never offered a proposed amended complaint, never moved for reconsideration, and still didn’t say on appeal what new facts it would plead.
Takeaways for Iowa software founders
- Being online isn’t a technical improvement by itself. If the problem you solve (picking a better ad, offer, or recommendation) also exists offline, expect § 101 scrutiny, however computerized the implementation is.
- Claim the mechanism, not the business goal. Steps like “map to keywords” and “weight by revenue” describe what the system does. If there is a real technical contribution, such as how data is structured, processed, or delivered more efficiently, the claims and specification should describe it concretely.
- The specification can be used against you. Here, the patents’ own description of the prior approaches as conventional weakened the inventive-concept argument.
- For patent owners enforcing software patents: plead factual allegations that support an inventive concept, and ask for leave to amend if a § 101 motion is granted. Waiting until appeal may be too late.
- Nonprecedential still signals direction. This decision doesn’t bind future panels, but it applies a familiar line of ad-targeting cases.
Sources: the Federal Circuit’s September 25, 2026 opinion (PDF) and IPWatchdog’s coverage.
For related reading, see our patents overview, the software and computer-related patents page, and our earlier note on a claim that survived Alice step one.
Talk with Iowa patent counsel
Jason Sytsma works with Iowa software and technology companies on eligibility-aware claim strategy for computer-implemented inventions. Learn more on the About Jason page and the firm bio.
Contact the office to request a consultation. Any engagement is subject to conflicts review, an engagement agreement, and applicable professional-responsibility rules. This article is attorney advertising and is intended for general educational purposes only; results vary, and past experience does not guarantee a particular outcome.
