Understanding our 99.5% accuracy
Fingerprint Pro combines dozens of browser attributes to create a unique and stable
visitorId. Attributes include screen resolution, operating system, IP address, loaded fonts, and other information that your browser can access. Individual attributes may not be accurate on their own, but when many are added together the result is unique for most website visitors. For example, lots of browsers run on Macs, but not as many have an IP address in Nebraska, and fewer still have a Wingdings font – each of these attributes add up to help make a fingerprint. With the help of a machine learning system and a few other identifiers, Fingerprint Pro can recognize a returning browser through its
visitorId 99.5% of the time.
How Fingerprint Pro tests accuracy
As part of its normal operation, Fingerprint Pro saves a cookie to each visitor’s browser.
We can identify cookied browsers with 100% accuracy and can use that as a reference to check the accuracy of all other identification methods. Below are the formulas for the accuracy of visitorIDs without a cookie, and the total accuracy of the Fingerprint Pro system.
Across all our customers, the
visitorId method gives an average accuracy of 99.5%.
Why Fingerprint Pro is better than cookies alone
Cookies are accurate but are not always available. Users can delete cookies or browse in incognito mode. Fingerprint Pro is a better way to identify browsers because it can generate a unique
visitorId without cookies. Another benefit is that Pro can store the history of every attribute that we use to identify browsers. If a user switches to incognito mode, all those attributes remain the same and identification is easy. If a user upgrades their browser some attributes change, but enough remain the same that we can still identify the browser.
Factors that influence accuracy
Sometimes a visitor to a website will have all the same browser attributes as another different visitor. If we aren’t able to find a difference between the two visitors, we may give that visitor’s browser the same
visitorId. This situation is known as a false positive. Pro customers typically see a false positive rate around 0.5%.
Accuracy of the open source version
Fingerprint Pro, the market leader
In a recent browser fingerprinting study from KTH Royal Institute of Technology, only 33.6% of users were correctly identified. Other studies from the Electronic Frontier Foundation and Inria saw fingerprint accuracy between 80-90%, but those studies predate current web privacy policies and technologies. Fingerprint Pro’s 99.5% accuracy is higher than any other service on the market. Through browser fingerprinting and other techniques, Fingerprint Pro provides best-in-class identification accuracy while complying with GDPR and CCPA rules.
Updated 9 months ago