ASOpatents.com compiles a list of patents that are likely used to shape the algorithms of the Apple App Store, Google Play Store, and other major platforms. While it's uncertain whether these patents are actually implemented in the algorithms, the site provides insights into potential clues about search results, recommended apps, and other data points.

US11816694B2: The Patent Behind SKAdNetwork Attribution

Patent numberUS11816694B2
Official titleTechniques for managing advertisement attributions while preserving user privacy
AssigneeApple Inc.
InventorsErik C. Neuenschwander, Jason A. Novak, Joseph A. Tyson, Dana J. DuBois, Jakob D. Swank, Darren M. Lew, Dmitriy Y. Filyushin, Katherine B. Skinner, Ross F. LeBeau, Payam Mirrashidi, Guy L. Tribble, Jeffrey L. Robbin, Sam H. Gharabally
Priority dateFebruary 28, 2018
FiledSeptember 28, 2018
GrantedNovember 14, 2023
StatusActive. Anticipated expiry June 5, 2039
FamilyUS20190266631A1 (published application), plus a later continuation
Sourcepatents.google.com/patent/US11816694B2

What This Patent Covers and Why It Matters for ASO

Feature it describes: install attribution for app install ads without exposing a device identifier. This is the architecture behind SKAdNetwork.

Scope note. This is a measurement patent, not a ranking patent. Nothing in it decides what appears in search results. It is on this site because it governs how the acquisition side of the App Store is measured, and because the signals that feed Apple’s ranking patents, conversion and downloads per query, are exactly the signals this system deliberately withholds from advertisers.

The inventor list is the first thing worth noticing. It includes Guy L. Tribble, Jeffrey L. Robbin, Payam Mirrashidi and Sam H. Gharabally, names associated with the most senior levels of Apple’s software and store organizations. Patents at this seniority are not experiments.

The mechanism in plain terms:

  1. An ad network signs campaign parameters, including campaign identifiers, application identifiers and a UUID, with its private key.
  2. When a user taps the ad in a source app, the app store caches those signed parameters.
  3. When qualifying activity occurs in the target app, such as installation or first launch, the cached parameters are routed to an independent metrics manager.
  4. The metrics manager verifies the signature using the network’s public key and only then notifies the ad network.

What the ad network receives is campaign identifiers and the target application identifier. What it does not receive is the advertising identifier of the device or the IP address. The patent names both explicitly as the exposure this design exists to prevent.

Two observations for practitioners:

  • The signature is the anti fraud mechanism. Attribution cannot be claimed without the network’s private key, which is why spoofed installs in this model require key compromise rather than clever request forging.
  • Conversion values are not in this document. The patent describes qualifying activities such as install and first launch, but no postback URLs, no conversion value encoding and no multi touch modeling. Those arrived later. If you are looking for the origin of conversion value design, it is not here.

The priority date, February 2018, places the design work before the public arrival of the framework, which is the usual pattern and a useful reminder that a patent’s priority date is the better guide to when Apple decided something than any announcement.

Patent Summary

An app store application receives digitally signed campaign parameters from a source application displaying an advertisement. When the user interacts with the advertisement, the app store caches those parameters.

Later, when the advertised target application exhibits qualifying activity, the cached parameters are routed to a metrics manager that operates independently of both the app store and the advertising network. The metrics manager verifies the digital signature against the advertising network’s public key, establishing that the attribution claim originates from the network that actually served the advertisement, and then notifies that network of the attributed event.

Because the notification carries only campaign and application identifiers, the network learns that one of its campaigns produced an install without learning which device or user produced it. The separation of the verifying party from both the store and the network is what makes the claim distinctive rather than a description of ordinary attribution.

A detailed breakdown of the signing flow, the figures and the full claim language will follow in a later update to this article.

Bir yanıt yazın

E-posta adresiniz yayınlanmayacak. Gerekli alanlar * ile işaretlenmişlerdir