Tracify's patented tracking technology (US 12,405,969 B2)

Tracify's core technology is patented

On September 2, 2025, the US Patent and Trademark Office (USPTO) granted patent US 12,405,969 B2, titled "System for Providing Tracking Data." The assignee is Tracify GmbH (Munich); the named inventors are Markus Rohm and Marius Rudolf; priority dates back to a European application filed December 3, 2021. The patent comprises 19 claims.

A granted patent is more than a label. The USPTO examines an application for novelty and inventive step – an examiner has to confirm that the method is new and non-obvious. For you as a user, that means: the technology underlying Tracify's measurement is an examined, publicly documented method – not a promise you simply have to believe.

The problem: why customer journeys fragment

A customer journey is rarely a continuous line. In practice it fragments into pieces – and that is exactly the starting point the patent describes. The reasons:

• Cookie lifetimes and blockers. Cookies have short lifespans and are restricted by browsers and blockers. If the identifier is gone, the trace breaks.
• Device switches. Someone discovers a product on their smartphone and buys later on the desktop. To a cookie-based setup, those are two unrelated people.
• Network and browser switches. Switching from mobile data to Wi-Fi, a VPN, or a different browser can make the same person look like several unconnected users.
• Breaks and consent. If a user interrupts their session or declines tracking, further gaps appear.

The result: you capture many individual fragments, but not the one coherent journey. And correctly reuniting these fragments is – as the patent puts it in essence – the actual challenge.

What Tracify patented – explained simply

The patented method solves exactly this reuniting. The process can be described in four steps:

1. Capture signals, not persons. An interaction produces a dataset of many technical characteristics – in the patent "entity identifiers," or a "footprint" of hardware, script, and action characteristics (such as device and browser properties or characteristic usage patterns). These are deliberately anonymous signals, not real names or account data.
2. Form an anonymous identifier. From these signals an anonymized tracking identifier (a UUID or a hash) is generated that specifically does not reveal the person's true identity.
3. Form identification groups. The system forms "identification groups" – reduced, anonymous representations per person. They are designed so that no personal information can be reconstructed from the data they contain. Data protection is thus built into the method, not bolted on afterwards.
4. Assign fragments to the right person. Using a similarity measure – at its core a distance between datasets in a parameter space – the system assigns each new fragment to the best-matching identification group. If nothing fits closely enough, a new group is created; if it turns out that two groups are the same person, they are merged. A machine learning module learns the patterns and becomes more precise as the amount of data grows.

In short: the patent describes an anonymous, AI-powered matching that assigns fragmented data traces to the same person without identifying that person.

Three properties that make the difference

Three properties follow from the method that set it apart from cookie-based standard tracking:

• Across devices and sessions. Because the assignment doesn't hang on a single cookie but on the similarity of many signals, a journey can be reunited into one person across devices, browsers, and sessions. Switching from smartphone to desktop no longer breaks the journey.
• Without login and without an account. The method works explicitly independently of login or account information. It needs no logged-in account to connect fragments.
• Privacy by design – designed to work without a consent opt-in. The data is anonymized early (in the patent, among other things via a gate module with a reverse proxy), and deliberately only signals that don't require consent are processed. That is the methodological reason why the method is designed to work without a consent opt-in.

How it all works together in Tracify's product

Attribution is ultimately a question of data quality, and that arises in three steps: capture, match, attribute. The patent sits at the heart of it – the matching.

• Capture: Hybrid tracking captures as many touchpoints of a journey as possible.
• Match: the patented AI matching reunites the fragments into the right person across devices and sessions. This is exactly where the protected method takes effect.
• Attribute: on this more complete basis, the behavior-based AI attribution evaluates the true, incremental contribution of each touchpoint – not just a learned correlation weight.

The common thread: a touchpoint you don't capture can't be assigned to any journey; what isn't assigned can't be attributed correctly. Matching is the bottleneck – and that's exactly where the patent takes effect. For how attribution works in general, see the overview of marketing attribution.

Why a patent matters for you as a user

• Verifiable instead of claimed. A granted patent is publicly accessible and was examined by an authority. Instead of "we track better," there is a documented method you can read up on.
• Examined differentiation. Novelty and inventive step are prerequisites for a grant. The method is therefore demonstrably not what standard tools do anyway.
• Data protection as a design principle. Anonymization and reduced representation are part of the method, not an afterthought filter.

What the patent does not mean

To keep the classification honest: a patent protects a method – it does not guarantee specific metrics. It proves that the method is new and non-obvious, not that a specific tracking rate is reached in every setup. Figures such as near-complete capture over 30 days remain empirical values from real-world operation, not a statement of the patent. And a patent makes no conclusive data protection assessment for an individual case. This distinction matters to us because it marks the difference between examined technology and marketing.

Conclusion

With US 12,405,969 B2, the method that forms the core of Tracify's measurement is protected: the anonymous, cross-device and cross-session reuniting of fragmented customer journeys – without a login and designed to work without a consent opt-in. For you, this means one thing above all: the data quality on which ROAS, CAC, and attribution build rests on an examined, documented method – not on a promise.