You can count on TypingDNA’s technology to recognize individual typing behavior with great confidence, depending on what typing pattern you choose to use, as well as the size and the number of typing samples enrolled for a user profile. Below, we explain the ways these variables influence how quickly TypingDNA can recognize you by the way you type.
Size of the typing sample
The size of the typing sample required for authentication is directly related to the kind of typing pattern and device you use.
The solution works if a person types a verification text that is identical to the enrollment text. In our tests, we’ve achieved great results with an average of 28 characters – that is as long as this text – pretty cool, right?
If you use typing biometrics authentication for online credit card payments, for example, the average credit card number and name usually contains 30-plus characters, which would be good enough for a first enrollment. However, for higher confidence, we recommend including more than two samples on each user profile before proceeding with the authentication.
After doing so, the card’s owner would only have to write their card number and name again for accurate authentication to occur. The same goes for login security, password reset and a couple of other scenarios. Check out these demos for more.
Using “same text” patterns the solution is able to recognize you by the way you type on mobile devices, too. Authentication on mobile works with at least one previous sample. However, we always recommend a minimum of two previous samples (i.e., typing patterns). For increased confidence we recommend 3 or more. We’ve seen impressive results on an average of 16 characters per sample, but we don’t recommend going below 8 characters.
You might need your users to write a different text each time they authenticate. In this case, the technology might require a more comprehensive typing profile recorded for each person.
Adding multiple typing samples to a profile can ensure such complexity. You can do this during your users’ first enrollment, by simply asking them to type a couple of distinctive tweet-long texts (good results require an average of 140 characters per text).
For example, in an online learning authentication scenario, a student can type two different texts when they enroll for the first time. The same student would then type a different text to be authenticated at any time during their course or examination.
No. of enrollments required
Overall, regardless of the kind of typing pattern you choose (same text or any text), the technology works with a minimum of one typing sample per user. But the more samples you collect, the more accurate the authentication will be.
Try our online demos to see how quickly TypingDNA can recognize you by the way you type.
Alternatively, if you’re more tech-savvy, sign up for a free account and check out our API documentation to start using typing biometrics in your own projects.