Federal Circuit leaves AudienceScience’s ad-targeting patents ineligible

The story

The U.S. Court of Appeals for the Federal Circuit has affirmed that three AudienceScience internet-advertising patents are ineligible for protection under Section 101 of the Patent Act, preserving a dismissal obtained by Google and YouTube. The September 25 decision concerns an early form of personalized ad selection: combining the context of a requested web page with a user’s browsing history, assembling candidate advertisements and choosing among them with a score tied to past revenue performance.

AudienceScience’s receivership estate asserted U.S. Patent Nos. 7,747,676, 7,882,175 and 8,082,298. The related patents share a specification and claim priority to December 2004. Their common description maps pages and browsing histories to keywords, maps those keywords to advertisements and then weights the candidates according to how much revenue an ad had generated previously. One ad is selected and displayed with the requested page. AudienceScience argued that combining current-page context, individual history and performance-weighted selection improved earlier approaches.

The Federal Circuit disagreed. Applying the Supreme Court’s two-step Alice framework, the three-judge panel said the claims were directed to the abstract idea of targeting advertisements to internet users to maximize revenue. At the first step, the court found that collecting information, matching keywords and selecting an ad did not solve a problem unique to computer networks or improve the operation of the computer itself. The panel compared the claims with earlier targeted-advertising cases and concluded that adding a performance score supplied particularity, but not enough to change the claim’s overall character.

At Alice step two, the court considered whether the claims contained an inventive concept capable of transforming that abstract idea into a patent-eligible application. It found none. The patents themselves described page-context targeting and user-history targeting as conventional approaches, while the representative claim did not prescribe a precise technical method for calculating its revenue-based weight. In the court’s view, combining familiar targeting inputs and asking a generic computer to choose the better-performing ad still described the desired result rather than a new computational mechanism.

That distinction matters beyond advertising. Software patents regularly express valuable commercial ideas through data collection, ranking and selection. Section 101 asks a different question from whether an idea was useful, first in the market or economically important. A patent claim must fit within eligible subject matter, and an abstract process implemented with generic computing components may fail even before a court reaches novelty, obviousness or infringement. Here, the case ended on the pleadings; the court did not decide whether Google copied an invention or whether earlier publications anticipated the claims.

AudienceScience also challenged the district court’s decision to dismiss the complaint without permission to amend it. The Federal Circuit held that the estate forfeited that argument because it had not requested leave to amend in the district court or identified there how a revised complaint would cure the problem. That procedural holding reinforces a practical lesson for patent owners: if eligibility may turn on factual allegations about a claimed technical improvement, those allegations—and a clear request to amend—need to be developed before the case reaches appeal.

The ruling is deliberately narrow. Judges Jimmie Reyna, Richard Taranto and Leonard Stark labeled the disposition nonprecedential, so it does not establish a binding new rule for every ad-tech, recommendation or personalization patent. Nor does it say that all use of browsing history or performance data is unpatentable. A claim directed to a specific network architecture, data-processing technique or improvement in computer functionality could present a different eligibility analysis. The weakness identified here was the gap between a concrete business objective and the absence of a concrete technological means for achieving it.

INNOVOX analysis: the opinion is a drafting signal for inventors working in AI-driven ranking and recommendation systems. Modern models can combine far richer behavioral, contextual and performance signals than the 2004-era patents described, but complexity alone does not establish patent eligibility. Claims that stop at receiving data, assigning a score and selecting an output remain vulnerable. Stronger applications will need to connect the invention to a demonstrable technical constraint or improvement—such as lower latency, reduced memory use, privacy-preserving computation or a defined model-training mechanism—and describe how the system produces that result.

What to watch next is whether the receivership estate asks the panel or the full Federal Circuit to rehear the case, or seeks Supreme Court review. Neither step is guaranteed, and the nonprecedential opinion may reduce its broader reach. The larger policy question remains active: courts, applicants and technology companies continue to debate whether the Alice framework gives software inventors a predictable boundary. For now, the AudienceScience result leaves Google’s dismissal intact and adds another concrete example of how outcome-focused software claims can fail at the eligibility threshold.

INNOVOX analysis

The decision separates a commercially useful result from a patent-eligible technical invention. For software applicants, the lesson is not that personalization cannot be patented, but that a claim needs to explain a concrete computing mechanism or improvement rather than describe the information considered and the business outcome sought.

What to watch

Watch whether AudienceScience seeks rehearing or Supreme Court review, and whether future ad-tech claims emphasize particular data structures, network operations or measurable computer-performance improvements. Also watch Congress and the Supreme Court for any broader change to the Section 101 framework.